v5 fix: unblock the e2e tier on the v5 base by mikolalysenko · Pull Request #288 · SocketDev/socket-patch

Mikola Lysenko (mikolalysenko) · 2026-09-28T11:35:33Z

Waste-review follow-up W0, the fast lane from #286. This PR is kept small so it can land first.

Finding	Outcome
F77: a red base test keeps the e2e tier, which sits behind `needs: test`, from running	Done. The cargo assertion the review cites (`e2e_redirect_cargo_build.rs:829`) was already fixed by #280 (`686e5fb`). `test (windows-latest)` was also red on the base in `covgap_commands_scan_mod`: on `8ae7dc3` (run 36352437716), on #280's own head (36383945958) and on #281 (36414725782). The cause was `scan_hosted_paths_run_once_per_project_directory` (`62f07c7`), which compared the `== apps/a ==` header with a literal `/` path. `scan` prints the directory `glob` matched, which Windows spells `apps\a`; the test now builds the expected header from path components (`bd632f5`). Proof: on `bd632f5`, `test` passed on ubuntu, macos and windows, and the e2e tier ran for the first time on a v5 head (run 36420943134).
F77 fallout: e2e-tier tests that had not run since `get` began defaulting to hosted (`5e5f5ed`)	Fixed. `e2e_vlt` `vlt_pinned_matrix_agent_get_and_remove` (`b9286f9`, the same change as `cc5f1b6` on #285) and the three `e2e_safety_pnpm` apply tests (`c699754`) now pass `--mode agent`. These are one-line test fixes, the same kind `5e5f5ed` already applied to the default-tier tests.
F78: CI concurrency cancels long runs	Done, as scoped. CI and the go, pdm, poetry, bun and vlt compatibility workflows now cancel a run only when it is a `pull_request` run superseded by a newer push to the same PR. Push, dispatch and scheduled runs always finish; before, `ci.yml` cancelled every run not on `main`, and go, pdm and poetry could cancel `main` push runs mid rust-cache save. CI groups PR runs by PR number, which is equivalent to `refs/pull/N/merge`. Deferred: gating heavy CI on drafts; the cited cancelled run (36371317869) was a real same-PR supersession.

Surfaced but not fixed here (not caused by this PR):

yarn-classic 1.0.2 / 1.6.0 / 1.7.0 / 1.9.4 mode_migration_npm is a real v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280 gap. yarn < 1.10 writes no integrity line, so the ledger-free upstream restore refuses vendor over a hosted pin, and rollback cannot land on the pristine old-format lock. It reproduces locally on 1.0.2 and 1.9.4, and 1.22.22 passes. The proposed WS1 fix is in the PR comments.
vlt patch compatibility install-proof legs come from v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280's hosted vlt rollback (hosted_wiring_contested).
macOS hosted legs in Bun, Poetry and vlt fail on runner reachability to patch.socket.dev (Errno 8, transport failure). WS5: one VendoredBackend for vendored apply/revert/repair; cut repair's ledger rebuild #283 passed the same legs on the same base code.

Savings: 15+/14− lines of workflow YAML plus 7+/6− lines of test fixes. The e2e tier now runs on v5 PRs instead of being skipped, and same-PR supersession is unchanged.

CI equivalence: no job, step or trigger changes; only the concurrency blocks. A group still holds at most one pending run, so a third dispatch on the same ref displaces a queued one.

Tests: locally, covgap_commands_scan_mod passes 45/45 and e2e_redirect_cargo_build 3/3, and e2e_vlt / e2e_safety_pnpm compile. The real-vlt and live-API legs cannot reach npm or patches-api through this sandbox's proxy, so CI is their proof. Clippy on the CLI tests is clean.

🤖 Generated with Claude Code

https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp

Note

Low Risk
Changes are CI concurrency policy and test harness fixes; no production CLI behavior or auth/data paths are modified.

Overview
Unblocks the e2e tier on v5 by fixing CI cancellation and two failing tests.

GitHub Actions concurrency is tightened across ci.yml and the bun, go, pdm, poetry, and vlt compatibility workflows: cancel-in-progress is true only for pull_request events, so pushes to main, workflow_dispatch, and scheduled runs are no longer cancelled mid-job (protecting rust-cache save-if on main). ci.yml also groups PR runs by github.event.pull_request.number instead of ref alone, matching other workflows.

Tests: scan_hosted_paths_run_once_per_project_directory builds expected == … == headers with Path::join so Windows matches glob output (apps\a vs apps/a). fresh_checkout in the NuGet dotnet e2e suite copies .socket/ only when present, reflecting v5 hosted state where the lock alone holds pins.

^{Reviewed by Cursor Bugbot for commit 7e1758b. Configure here.}

A CI or compatibility run is now cancelled only when a newer push to the same pull request replaces it. Push, dispatch and scheduled runs always finish, so a manually dispatched run on the v5 base branch (its only CI verdict, since push CI runs on main alone) is no longer killed by a later dispatch or by the non-main cancel rule, and main keeps finishing its rust-cache saves. CI now groups PR runs by PR number, like the compatibility workflows already do. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp

Mikola Lysenko (mikolalysenko) · 2026-09-28T11:40:18Z

[agent] CI note: the vlt patch compatibility install-proof legs are red on this PR (for example install-proof (ubuntu-latest, 0.0.0-16), job 108909893148). This PR does not cause them. It changes only the workflows' concurrency blocks; vlt ran here only because vlt-compatibility.yml is in its path filter.

Failing tests: vlt_pinned_matrix_hosted_{crlf_lock,idempotence,resave_crlf_rollback,resave_install_rollback,rollback_byte_exact} in e2e_redirect_vlt_build.rs. Hosted rollback returns partial_failure with hosted_wiring_contested on vlt-lock.json. vlt_pinned_matrix_agent_get_and_remove in e2e_vlt.rs:156 also fails (Absent, expected Patched).
These are base-inherited from v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280. Its last vlt run, 36383945937 on fc9a52a, failed 61 install-proof legs, and v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280 landed with that result. Every PR that has since merged the base fails the same way: WS3: one lockfile model per ecosystem #281 (36414725689), WS5: one VendoredBackend for vendored apply/revert/repair; cut repair's ledger rebuild #283 (36414561626 CI / 36414561657 vlt) and v5: remove setup (WS7) + patch UI streamlining (WS8) #279 (36414963421).
Existing fixes: cc5f1b6 on Add the NuGet vendoring design (docs only) #285 (--mode agent for the vlt get test) fixes only agent_get_and_remove. No fix exists yet for the hosted vlt rollback regression, so porting cc5f1b6 alone would not get this job green, and I haven't ported it. The failures are deterministic, so I am not re-running them. The hosted vlt rollback needs its own WS1 follow-up.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T11:41:13Z

[agent] CI note: Poetry patch compatibility failed native (macos-latest, 1.0.10) (job 108909964306). This PR changes only the concurrency blocks, so it should not cause this.

The direct and populated hosted legs errored with NewConnectionError ... [Errno 8] nodename nor servname provided: pip could not resolve the hosted artifact host while installing urllib3-1.26.18. The crlf hosted leg then failed its install checks on the same leg.
The same workflow is green on every PR that has merged the current base: v5: remove setup (WS7) + patch UI streamlining (WS8) #279 (36414963477), WS3: one lockfile model per ecosystem #281 (36414725712), v5 WS4/WS6: one hosted engine for disk + memory; unified Ledgers view #282 (36415924625), WS5: one VendoredBackend for vendored apply/revert/repair; cut repair's ledger rebuild #283 (36414561750). v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280's head (36383945905) was green too.
No fix is needed or exists to port. I re-ran the failed jobs once. If the job fails again, I will treat it as real and root-cause it.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T11:44:54Z

[agent] CI note: Bun patch compatibility failed native (macos-latest, 1.1.39) (job 108910543180). The hosted legs fail with urlopen error [Errno 8] nodename nor servname provided, the same macOS DNS error as the Poetry leg above. The vendored legs pass.

This PR changes only the concurrency blocks. v5: remove setup (WS7) + patch UI streamlining (WS8) #279, which includes the current base, passed Bun (36414963363).
WS3: one lockfile model per ecosystem #281, WS5: one VendoredBackend for vendored apply/revert/repair; cut repair's ledger rebuild #283 and Keep NuGet patches installed with <clear /> and NuGet.config #284 all failed Bun in the same ~11:40 UTC window, which points to a runner DNS problem rather than code.
No fix exists to port. I will re-run the failed jobs once when this run completes. A second failure gets treated as real.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T12:16:53Z

[agent] note from the #286 triage owner: there is a second red test on the v5 base, besides F77.

test (windows-latest) fails one target on release/v5-prerelease @ 686e5fb: -p socket-patch-cli --test covgap_commands_scan_mod. The failure reproduces on two docs-only PRs, #285 (job) and #289 (job). Ubuntu and macOS pass. Unblocking the v5 base is in scope for this PR, so please take this one too, or say in the PR body that it is out of scope.

Generated by Claude Code

scan_hosted_paths_run_once_per_project_directory compared the per-directory `== apps/a ==` header against a literal forward-slash path, but scan prints the directory glob matched, which Windows spells `apps\a`. The Windows test leg failed on this since 62f07c7, and because every e2e job waits on `test`, the whole e2e tier was skipped on v5 PRs. Build the expected header from path components. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp

Mikola Lysenko (mikolalysenko) · 2026-09-28T12:18:16Z

[coordinator] #288 is first in the v5 landing order (fast lane), but it can't land yet because it is still a draft. It merges cleanly and already contains release/v5-prerelease @ 686e5fb. Please mark it ready for review when it's done. Before it lands, CI on its head also needs to finish with only base-inherited failures, and it needs no unresolved review threads.

Other PRs, checked this run:

WS5: one VendoredBackend for vendored apply/revert/repair; cut repair's ledger rebuild #283 (next in core): CI run 36417124615 on 9d46742 is still in progress.
docs: v5 waste review and repacking design (reference) #289 (docs-only): CI run 36417036132 is still in progress (test-release).

Nothing landed this run.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T12:18:33Z

[agent] Picked up in bd632f5, and the PR body now covers it under F77. The Windows failure in covgap_commands_scan_mod comes from scan_hosted_paths_run_once_per_project_directory, which expected a == apps/a == header. scan prints the directory glob matched, which is apps\a on Windows. It has been red since 62f07c7 (8ae7dc3's run 36352437716 fails the same target). The test now builds the expected header from path components.

For the coordinator: I will mark this ready and post [agent] ready to land once CI on bd632f5 shows the Windows leg green and the e2e tier running.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T12:25:06Z

[agent] CI note on bd632f5: the macOS hosted legs fail in two workflows. The vendored and agent legs pass in both.

Bun patch compatibility native (macos-latest, 1.2.0) (job 108923824532). The one leg that printed a cause shows urlopen error [Errno 8] nodename nor servname provided.
vlt patch compatibility native (macos-latest, 1.2.0) (job 108924240471). Every hosted leg reports transport failure; retrying three times.

Hosted mode downloads from the real patch.socket.dev, so these runs depend on that server being reachable from the runner.

This is not this PR's code. WS5: one VendoredBackend for vendored apply/revert/repair; cut repair's ledger rebuild #283 ran the same base code minutes earlier and passed every macOS Bun leg, hosted included (run 36417124687). The Ubuntu legs of these runs pass too. This PR changes only the concurrency blocks and one test file.
Cause: some macOS runners intermittently cannot resolve or reach patch.socket.dev. The same failure hit WS3: one lockfile model per ecosystem #281 and Keep NuGet patches installed with <clear /> and NuGet.config #284 at about 11:40 UTC. No fix exists to port.
When each run completes, I will re-run its failed jobs once. A repeat will be reported as a runner or network problem that needs the owner, not something this PR can fix. The vlt install-proof legs are still the separate base-inherited failure described above.

Generated by Claude Code

get defaults to hosted mode since 5e5f5ed, so the real-vlt get_and_remove leg ran a hosted get and found the installed copy unpatched (Absent, expected Patched). Pass --mode agent, as the other agent-mode fixtures already do. Same change as cc5f1b6 on #285; it blocked the e2e tier's e2e_vlt jobs now that they run. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp

get defaults to hosted mode since 5e5f5ed, so the three pnpm safety tests ran a hosted redirect and found proj_a's installed copy unpatched and no pnpm-layout note. They test the in-place apply path, so pass --mode agent. These e2e-tier tests had not run since that change because a red base test skipped the tier. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp

Mikola Lysenko (mikolalysenko) · 2026-09-28T13:05:47Z

[agent] The e2e tier now runs. On bd632f5 (CI run 36420943134), test passed on ubuntu, macos and windows, so the needs: test jobs ran for the first time on a v5 head. Here is the triage of every failure they surfaced:

e2e failure	Cause	Outcome
`e2e_vlt` (every vlt version, all OSes): `vlt_pinned_matrix_agent_get_and_remove` (Absent, expected Patched). This is the only failing test in those jobs	`get` defaults to hosted since `5e5f5ed`; this e2e-tier test still expected agent mode	Fixed in `b9286f9` (same change as `cc5f1b6` on #285)
`e2e_safety_pnpm` (ubuntu, macos): 3 tests	Same cause: `get` ran a hosted redirect (the log shows `redirect_pnpm_trust_lockfile`)	Fixed in `c699754` with `--mode agent`
`e2e_nuget_dotnet_build` 6/7/8/10: hosted leg `NotFound` at :319	Test drift from #280: v5 hosted mode writes no `.socket/`, and `fresh_checkout` copied that tree unconditionally	Fixed in `10a7724`: copy it only if present
`yarn-classic 1.0.2 / 1.6.0 / 1.7.0 / 1.9.4`: `mode_migration_npm` classic	Real #280 gap, not fixed here. yarn < 1.10 writes no `integrity`, so the ledger-free upstream restore refuses `vendor` over a hosted pin (`redirect_revert_failed`), and `rollback` cannot land on the pristine lock. Reproduced locally on 1.0.2 and 1.9.4; 1.22.22 passes	Proposed WS1 fix: the yarn-classic restorer drops `integrity` when the entry's lock format has none (like #280's Pipenv `index`), and the mock mirrors the lock
`e2e_redirect_vlt_build`, `mode_migration_vlt` (4 tests), `e2e_redirect_uv_build` `hosted_uv_transitive_override_manifestless_vex` (every uv version)	Real #280 gap, not fixed here. Hosted rollback reports `hosted_wiring_contested` (`patched_ref_unattributable`) on `vlt-lock.json` / `pyproject.toml`. This is the same class as the vlt compat install-proof failures	WS1: the ledger-free attribution of vlt pins and uv transitive overrides

So after this PR the e2e tier runs, and the red that remains is real #280 regressions it had been hiding. A skip could not have surfaced them. The macOS patch.socket.dev reachability notes above still apply.

Generated by Claude Code

v5 hosted mode writes no `.socket/` directory (the lock pins are the whole hosted state), so the hosted leg's fresh_checkout panicked with NotFound copying a tree that no longer exists. Copy it only when the run left one; the vendored leg still carries its ledger through. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp

Mikola Lysenko (mikolalysenko) · 2026-09-28T13:18:43Z

#283 landed on release/v5-prerelease as 06437d2; please merge origin/release/v5-prerelease again, resolve conflicts, get green, and keep it ready.

Generated by Claude Code

…nblock-e2e-tier

Mikola Lysenko (mikolalysenko) · 2026-09-28T13:21:51Z

[agent] I merged release/v5-prerelease @ 06437d2 (#283) as 7e1758b. There were no conflicts; clippy on the workspace and covgap_commands_scan_mod pass. Once CI on 7e1758b finishes with only the #280 failures triaged above, I'll mark this ready and post [agent] ready to land. That should be about 50 min, the Windows leg's run time.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T14:08:32Z

[agent] ready to land

CI on head 7e1758b, which includes release/v5-prerelease @ 06437d2, is CI run 36428080359: 224 jobs passed and 22 failed.

test passes on ubuntu, macos and windows, and the e2e tier ran in full.
Everything this PR fixed passes: e2e_vlt on every vlt version and OS, e2e_safety_pnpm on all three OSes, and e2e_nuget_dotnet_build on SDKs 6 to 10.
The 22 failures are all the v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280 regressions triaged in the comment above, none caused by this PR: 14 e2e_redirect_uv_build jobs, 3 e2e_redirect_vlt_build jobs and 1 mode_migration_vlt job (all hosted_wiring_contested), and 4 yarn-classic < 1.10 jobs (mode_migration_npm).
Also from v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280, not from this PR: the vlt compat install-proof legs, and the macOS hosted legs that cannot reach patch.socket.dev.

There are no review threads.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T14:19:02Z

[coordinator] #288 is first in the landing order. It is not a draft, it is MERGEABLE, it contains release/v5-prerelease @ 06437d2, and it has no review threads. It still can't land under the coordinator's rule, because 23 failing checks on 7e1758b don't count as base-inherited:

22 e2e-tier jobs have no run on the base. These are 14 e2e_redirect_uv_build, 3 e2e_redirect_vlt_build, 1 mode_migration_vlt and 4 yarn-classic 1.0.2/1.6.0/1.7.0/1.9.4. The base's latest CI run (36352437716 on 8ae7dc3) never ran the e2e tier because it was skipped behind needs: test, so these job names have no base result to match. The triage above calls them v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280 regressions, and that is plausible. The rule, though, needs the same job name failing on the base, and nothing can show that until the base runs this tier.
native (macos-latest, 1.1.45) (run 36428080424) fails here and passed on the base. This looks like the macOS patch.socket.dev reachability problem noted above. A green re-run would clear it.
The 23 vlt install-proof failures are base-inherited, and they are fine.

To unblock, either land the fixes for the #280 hosted-rollback and yarn-classic regressions first, or the owner explicitly approves landing #288 with those 22 e2e jobs red.

Other PRs, checked this run:

WS3: one lockfile model per ecosystem #281 (next in core) contains the base. Its only non-inherited failure is Bun native (macos-latest, 1.2.0) (run 36429829217, attempt 1), the same macOS hosted-host failure. It needs one green re-run.
docs: v5 waste review and repacking design (reference) #289 (docs-only): its [agent] ready to land (11:40) is older than its last push (13:19). It needs a fresh ready comment.
Add the NuGet vendoring design (docs only) #285: test (windows-latest) is still running.

Nothing landed this run.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T14:19:31Z

[agent] Reply to the coordinator:

Bun native (macos-latest, 1.1.45): I re-ran the failed jobs of run 36428080424 once. The cause is the macOS patch.socket.dev reachability problem.
The 22 e2e-tier jobs cannot get a base result as things stand. The base (06437d2) still has the Windows covgap_commands_scan_mod failure that this PR fixes. A CI dispatch on release/v5-prerelease would fail test (windows-latest) again, and every one of these jobs would be skipped behind needs: test. So the rule's "same job name red on the base" can only be checked after v5 fix: unblock the e2e tier on the v5 base #288 lands. Holding v5 fix: unblock the e2e tier on the v5 base #288 until then is circular.
For the same reason, the two unblock routes need the owner. One is to land the WS1 fixes for the v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects #280 hosted-rollback and yarn-classic regressions first; those PRs would hit the same skipped-tier gate unless they carry this PR's Windows fix. The other is for the owner to explicitly approve landing v5 fix: unblock the e2e tier on the v5 base #288 with those 22 jobs red. I've asked the owner. The PR stays ready in the meantime.

Generated by Claude Code

Mikola Lysenko (mikolalysenko) · 2026-09-28T16:35:40Z

Owner approval: land #288 with the 22 known #280 e2e-tier failures red. More generally, I approve landing every open v5 PR into release/v5-prerelease once it contains the current base and its CI shows only known or base-inherited failures. Agents should merge the base in and re-validate after each landing.

…etup-and-ui Take the base's configuration.md deferred-defaults paragraph, which already accounts for `setup` being removed in v5. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PQaKzoW5dSw9u5pgAAvVRj

…ed-engine Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KHZ8uzdXfkG2zH8ZYDG8ju

Brings in the squashed staged-rollout plan (#290), the e2e-tier fix (#288: Windows path separators in a scan test, CI cancel rules) and the docs-only design PRs. The plan doc's later edits (two-phase policy selection for the in-memory engine) merge with this branch's implementation notes unchanged. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

Brings in the merged staged-rollout plan (#290), the Windows scan test fix (#288) and the docs-only design PRs. The plan's late two-phase memory selection is implemented in the next commit. Claude-Session: https://claude.ai/code/session_01BKsyzefGhAnPkYmXCwq3H3 Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* Lead the CLI help with the v5 workflow List scan, vex, vendor and list first, then the agent-mode commands (get, apply, setup, rollback, remove, repair), and add a short "Typical workflow" footer to the root help. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Prefer the newest merged patch when choosing one per package A merged patch folds several advisories into one blob and is the package's cumulative fix, so the newest one the user can download now wins outright. Packages with no merged patch keep the old order: highest severity, then newest. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Make scan default to hosted mode and never prompt A bare scan now runs hosted mode: it rewrites lockfiles so only the patched dependencies resolve to Socket-hosted packages. A path-scoped, --prune or global scan with no mode only reports, since it has no lockfile to rewire. scan asks nothing any more: no confirm prompt in any mode and no patch menu. It takes the top-ranked patch per package. New --package filter (repeatable or comma-separated, env SOCKET_SCAN_PACKAGES) scopes a scan to packages by name or purl. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Detect hosted patch updates from the lockfile's own pins scan's updates[] now also sees the hosted pins the lockfiles wire, not only the redirect ledger's records, so a hosted project that never committed its ledger still reports a superseding patch. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Share the crawler-options and ecosystem-scope helpers across commands GlobalArgs gains crawler_options(), is_global(), ecosystem_selected() and purl_ecosystem_selected(). They replace eight hand-copied CrawlerOptions literals, seven --ecosystems predicates and seven global checks across scan, vendor, vex, apply, setup, rollback, get and repair. vendor's predicate also matched case-insensitively and accepted for golang. clap validates the names first, so neither form ever reached it. It now uses the same exact match as everything else. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Let hosted and vendored scans take project directories In hosted and vendored mode, and so in a bare scan, positional PATHs now name project directories: each directory, or directory glob such as apps/*, is scanned on its own as if it were --cwd, under a "== <dir> ==" header. The exit code is the worst of the runs. A PATH that is not a directory is a usage error. --json takes a single directory so stdout stays one document. Agent-mode PATHs keep their installed-path glob meaning. The changelog's v5 section now leads with the scan → vex → vendor workflow and describes the new scan defaults. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Rank [UPDATE] detection by the same rule scan installs by candidate_supersedes still put severity above merge state, so scan's updates[] and [UPDATE] marker could name a different patch from the one scan selects. It now calls a new ranking::batch_supersedes, which uses the same merged-first key as the selection and ignores the tier and uuid tiebreaks and missing dates. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Default get to hosted mode, like scan socket-patch get <id>, and the bare-UUID shortcut, now redirect the package to its Socket-hosted patched copy. Agent mode (manifest + in-place apply) is --mode agent. It stays the default with --save-only, which records a manifest entry, and with --global/--global-prefix, since those have no project lockfile to rewrite. The agent-mode test fixtures now pass --mode agent. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Rewrite the README, CLI contract and docs for v5 The README now leads with the scan → vex → vendor workflow: hosted mode first, vendored for offline installs, agent mode as the older install-hook flow, and the command reference in that order. The contract and README now match the code on these points: - a bare scan and get run hosted mode - scan never prompts - --package - PATH semantics in each mode - the merged-first patch ranking - the exit-2 cases - the env var table The Bundler plugin README now says setup writes a path: source and that failures warn by default. The docs/testing notes drop references to files and flags that no longer exist. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Clean up stale and narrative comments across the tree Comments and doc comments now describe the current code: - references to renamed or deleted functions, files, tests and flags are fixed; - notes about pre-v5 scan behavior (prompts, report-only non-TTY scans, severity-first ranking, PATH rejection) are gone; - TODOs for finished work are removed; - history narration ("used to", bug diaries, PR and audit tags, dated verification notes) is rewritten as the invariant it protects; - doc comments attached to the wrong item are moved. Two CI path filters that named the deleted vendor/lock_inventory.rs now match vendor/lock_inventory/**. No code behavior changes. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Fix follow-ups from the stale-doc sweep - An exported-but-empty SOCKET_SCAN_PACKAGES= filtered every package out of a scan. It is now scrubbed like the other local env vars, and blank --package specs are ignored. - Manifest-less VEX liveness for a rebuilt cargo vendor entry (no recorded wiring) now also probes the root Cargo.toml, where v5 writes the [patch.crates-io] table, not only the pre-v5 .cargo/config files. - setup names the dependency it writes, socket-patch[hook]. The repair_vendor hint and the hosted --prune warning name the current flags. - Drop scan's unreachable patch-menu spacing and the vendored arm's leftover blank line from the removed confirm prompt. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Add the v5 plan: decisions and workstreams Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * v5 WS1+WS2: ledger-free hosted mode, upstream-restore rollback, vendor ejects hosted projects (#280) * Stop writing the hosted redirect ledger from scan Hosted scan keeps its edits and records in memory only; the lockfiles are the record of a redirect. Replays the parked WIP (which was snapshotted on an older tree) as just its hosted.rs delta. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * Restore hosted pins to their upstream registry entries instead of replaying the ledger Imports the stopped local WS1 agent's work-in-progress (backup/local-v5- ledger-free-hosted): core patch::redirect::upstream re-resolves npm-family, cargo and golang registry entries for every hosted pin vex::discover finds in the lockfiles, and rollback/remove/vendor route their hosted legs through it instead of the redirect ledger. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Update rollback prompt unit test for the upstream-restore wording Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Derive list, scan updates and takeover state from lockfile hosted pins v5 keeps no hosted ledger, so every reader that consulted .socket/vendor/redirect-state.json now reads the hosted pins lockfile discovery finds: - list shows each hosted pin with the lockfiles wiring it (details.lockfiles); a pre-v5 ledger only supplies the details of pins it still describes. - scan's updates[] fold, redirectState block and the agent-flow hosted_wiring_retained probe read the pins. - The hosted-over-vendored takeover classifier reads the pins; the vendored-over-hosted ledger reconcile (vendor_supersedes_redirect) is gone: once the lock routes a package to .socket/vendor/ no hosted state is left to go stale. - vex treats a malformed pre-v5 redirect ledger as an advisory. scan/mod.rs unit tests still need rewriting (WIP). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Drop the hosted ledger from get, repair, the in-memory engine and the scan fold - get's vendored lock-text gates read the lockfiles' hosted pins instead of the redirect ledger (DownloadParams carries --patch-server-url for it). - repair's hosted-only no-op fires for hosted lockfile pins (or a pre-v5 ledger). - The in-memory hosted engine neither reads nor emits .socket/vendor/redirect-state.json. - Disk hosted scan no longer folds edits into a throwaway ledger; its records feed only the stale-install probes and in-run VEX. - The cargo vendor backend's hosted_redirect_live refusal names rollback / git checkout instead of a ledger. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * vendor ejects a hosted project; vendoring over any hosted pin restores upstream first WS2: standalone `vendor` with no manifest now takes its patch set from the lockfiles' hosted pins (purl + the uuid in each hosted URL), fetches each record from the API, vendors it into .socket/vendor/ through the same step `scan --mode vendored` and `get --mode vendored` use, and rewires the lock from hosted to vendored. Without hosted pins it keeps the no-manifest no-op. The vendor takeover now restores the upstream registry entry before vendoring for every ecosystem, not only cargo/npm/golang, so the vendor ledger always records the upstream entry as its original and `vendor --revert` returns to upstream rather than to hosted. A pin whose upstream entry cannot be restored is refused with the checkout remedy. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite scan/mod.rs takeover unit tests for lockfile-derived hosted state v5 hosted mode keeps no ledger, so the scan/mod.rs unit tests now make a purl "hosted" by writing a lockfile that pins the hosted URL instead of planting .socket/vendor/redirect-state.json: - overlap / classify_overlap_takeover tests use hosted package-lock, yarn (classic + berry), bun and cargo sparse-index pins; the non-default-host test configures --patch-server-url and pins that an unconfigured host is no pin. - New behavior pinned: a pre-v5 ledger on disk is never hosted state; a lock routed to vendored (npm or cargo) yields no pin and no overlap; an edits-only state names no package; a half-migrated project whose locks name both sides stays silent; the redirectState block has no ledger/ledgerKey fields. - hosted_wiring_retained_purls / redirect_state_json tests read the lockfile-derived state, keeping the probe's own liveness gate covered. - Removed tests of retired behavior: note_vendor_supersedes_redirect reconcile (wet/npmrc/dry-run/no-op/persist-failure), the edits-only fallback (degraded ledger, vlt tilde keys) and following the vendored remediation. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Restore hosted gem and composer pins to their upstream entries Add the RubyGems and Composer restorers to the v5 ledger-free hosted unwind (`redirect::upstream`). Gem (Gemfile.lock / gems.locked + Gemfile / gems.rb): a converged Socket GEM section is removed and its spec moved back, in name order, into the upstream section (the single remaining one, else the manifest's global source or rubygems.org, else refused as ambiguous); a bundler <= 2.1 merged section loses only the Socket remote; the DEPENDENCIES `!` pin is dropped; CHECKSUMS is re-pinned from the rubygems.org compact index. The Gemfile source block becomes `gem "n", "v"[, opts]` again (the original constraint is not derivable), or is removed with its DEPENDENCIES entry only when provably a transitive append. The pre-2.6 mixed state is undone when a pin is supplied, keeping the untouched lock's own constraint. Composer: dist {type,url,reference,shasum} and the dropped source block are rebuilt from packagist p2 metadata (composer/2.0 minified, expanded), cross-checked against the lock's dist.reference, in the lock's indent, slash style and line endings. Non-packagist entries are refused. UpstreamClient gains cached rubygems and packagist lookups (SOCKET_RUBYGEMS_URL, SOCKET_PACKAGIST_URL). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Fix clippy findings in the eject and takeover paths Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Restore hosted PyPI pins to their upstream registry entries Add the Python family to the v5 hosted -> upstream restore: Pipfile.lock, requirements.txt, Hatch direct references (pyproject.toml / hatch.toml), poetry.lock, pdm.lock, and uv.lock / PEP 723 script locks / PEP 751 pylock files with their paired pyproject / script metadata. Each restorer rewrites only entries whose reference is a hosted URL for an in-scope patch uuid and re-derives what the hosted rewrite overwrote from PyPI's JSON API (UpstreamClient::pypi_files, base overridable with SOCKET_PYPI_JSON_API, cached like the other lookups). Where a field is not derivable the pin is refused instead of guessed: requirements hash-checking mode that no other line settles, a Pipfile.lock index that is not PyPI, PDM locks without cross_platform (or uv locks) when the release ships platform- or interpreter-specific wheels, uv locks with no sibling registry package to show the artifact shape, several or non-PyPI registries, exclude-newer / no-binary / no-build filtering, multi-clause uv specifiers with no spelling evidence, uv 0.2 [[distribution]] locks, offline runs and registry failures. The golden harness round-trips the native Poetry (1.0-2.4), PDM (every supported lock_version) and Pipenv fixtures plus synthetic requirements/Hatch/uv/pylock projects through the real hosted rewriter; the shared requirements golden restores modulo the grant's name casing and the uv golden (no registry sibling) is refused. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Restore vlt, Maven and NuGet hosted pins to their upstream entries vlt-lock.json: slot [2] from npm dist.integrity, slot [3] per the lock's own convention (same-era default-registry siblings, else DepID era and options.registries), every hosted instance of name@version together. Maven (no network): base versions back, added dependencyManagement entries, socket-patch repositories and emptied wrappers removed, trusted-checksum lines dropped and .mvn files deleted only when nothing but hosted content is left; module poms and stray suffix uses refuse. NuGet: socket-patch source and mapping removed, a mapping that only fans * out to every source dropped; packages.lock.json contentHash re-derived from nuget.org's catalog packageHash (SOCKET_NUGET_URL), refusing when the restored config does not resolve the id from nuget.org alone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Drop dead fixture fields clippy flags in covgap_commands_rollback Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Retire a pre-v5 hosted ledger when rollback finds nothing else; update the cargo guard test rollback in a project whose only state is a stale pre-v5 redirect-state.json (no manifest, no vendor ledger, no hosted pin in the lockfiles) removes that file and exits 0 instead of failing on the missing manifest. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Feed the hosted scan's in-run VEX this run's records The in-run `scan --mode hosted --vex` attestation read its hosted records from the ledger the run had just written. With no ledger, the run's fetched records reach the VEX builder in memory (VexBuildParams hosted_records), merged over any pre-v5 ledger's. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Refresh doc comments that still described the hosted ledger Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * vendor --revert restores hosted pins a pre-v5 vendor ledger re-creates A package vendored over hosted wiring before v5 recorded the hosted fragment as its pre-vendor original, so reverting it wired the lock back to the patch server. After a wet revert, any hosted pin on a reverted purl is restored to its upstream registry entry (warning vendor_revert_restored_upstream; a refused restore is a failed event, hosted_restore_failed, exit 1). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Refuse around contested hosted wiring; refuse offline eject before any request Review follow-ups: - core HostedInventory keeps raw hosted wiring apart from attributable pins: a hosted identity discovery recognizes but cannot attribute (locks that disagree, a malformed reference, a lockless registry pin) is contested wiring. rollback (unscoped), remove, list and vendor refuse around it with hosted_wiring_contested, naming the files and the git checkout remedy, instead of reporting a bare project. - vendor's eject refuses offline (flag or env, wet or dry) with offline_eject_unavailable before it builds any request. - Drop the stray upstream/client.rs.orig merge backup. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Make eject one transaction: plan, restore upstream, vendor, or roll back Review follow-ups on the hosted -> vendored eject: - Every patch record is fetched first; one the API cannot serve refuses the whole eject (eject_refused) before anything is touched. - The upstream restore is planned first (a dry resolve of every pin); a pin that cannot be restored refuses the whole eject with its remedy. - A dry run stops at the verified plan (eject_planned events) and writes nothing, not even .socket/. - The wet run takes the apply lock once, snapshots every file the eject can touch (root files, pin and restore files, cargo/maven config, the vendor ledger, vendored uuid dirs), restores the pins upstream BEFORE the vendor engine inventories sources (so a fresh checkout with nothing installed resolves the pristine registry package), vendors, and on any failure puts the snapshot back (eject_rolled_back): a failed eject leaves the project hosted, byte for byte. Adds failure-injection and dry-run coverage. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite hosted rollback/remove coverage-gap tests to the v5 upstream restore Hosted state is the lockfile pin: rollback/remove restore the default upstream registry entry (mock npm registry via SOCKET_NPM_REGISTRY), a refused pin (offline, registry 404, missing integrity) fails closed with the git-checkout remedy, and a pre-v5 ledger is never replayed but retired once no pin remains. Ledger-persist failure tests become restored-lockfile write failure tests. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite hosted rollback/remove/vex-step tests to the ledger-free v5 contract - in_process_rollback_hosted (+ vlt): hosted pins come from the lockfiles and are restored from a mock npm registry; per-pin refusal, scoped restores, legacy-ledger retirement (never replayed), preserve-state, and the vlt heal following restored pins. A vlt project re-locked onto the registry has no hosted state left to roll back. - coverage_fix_rollback_ecosystem_scoped_replay -> ..._scoped_hosted: --ecosystems never restores another ecosystem's pin nor retires the pre-v5 ledger while a pin remains. - e2e_golang_hosted_state: rollback and vendor takeover restore go.sum from a mock checksum database; offline refuses. - redirect_npm_allow_remote: no ledger records the .npmrc edit; restore deletes only a pristine scaffold .npmrc and reports a kept allow-remote=all line (npm_allow_remote_left). - hosted_symlinked_files, repair_invariants: no ledger is written; a lockfile-pin-only project takes the redirect_only_project skip. - vex_pipenv_pip_steps: a reverted checkout with no ledger is the plain manifest_not_found error; apply --vex fetches the record online when no ledger supplies it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite hosted scan/get integration tests for the ledger-free v5 contract Hosted scan/get write no .socket/vendor/redirect-state.json: assertions on the ledger become "no ledger" plus lockfile facts; pre-v5 (incl. corrupt) ledgers are pinned as ignored and left byte-identical; ledger-write failure tests become "a read-only .socket/vendor or a squatting dir no longer blocks the run"; re-runs are pinned idempotent on the lock bytes. Rollback round trips now name the mock host with --patch-server-url and mock the upstream registry (SOCKET_NPM_REGISTRY / SOCKET_PYPI_JSON_API); berry CRLF/BOM, bun digestless, pnpm, poetry, pdm and pipenv restores are checked against the pristine locks. bun.lockb rollback pins the refusal. Manifest-less VEX legs attest from lock + API, and use a synthesized pre-v5 ledger as the extra local record source for the offline and redirect_unwired legs. scan redirectState/hosted_wiring_retained/updates are pinned to lockfile pins; vendor_supersedes_redirect is gone. vlt_hosted_common gains assert_no_ledger, legacy_record_from_view and write_legacy_ledger (additive). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Tidy hosted remove/rollback messages and legacy-ledger residue - remove's hosted_revert_failed message is the restore's own refusal (it already names the pin and the remedy) instead of wrapping it twice. - Retiring a pre-v5 ledger prunes the emptied .socket/vendor/. - The in-run hosted VEX summary says patches are attested from their patch records, not from a ledger. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite list and vex ledger tests for lockfile-derived hosted state list: hosted entries are lockfile pins (details.mode hosted, details.lockfiles, no details.ledger); a pre-v5 ledger only details a matching pin and never lists a record by itself. vex: a malformed pre-v5 redirect ledger is the redirect_ledger_corrupt warning, the run proceeds and the file is left byte-identical. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite manifest-less VEX suites for the ledger-free hosted mode v5 hosted mode writes no .socket/vendor/redirect-state.json, so the shared harnesses (vex_pdm_hatch_common, vex_pipenv_pip_common, vex_e2e_common/bun and vlt, npm_e2e_common/manifestless) now assert the hosted wiring run left NO ledger, while vendored keeps its .socket/vendor/state.json cells. Hosted cells that relied on the ledger now pin the new contract: offline with no local record is record_unavailable, online attests from the API, and a reverted hosted lock leaves nothing to discover. vendor over hosted pins is the eject flow, so the manifest-less embedded cells drive apply for hosted checkouts and a new cell pins the eject path. One focused cell shows a committed pre-v5 ledger still lets a hosted pin attest offline (new additive helpers: write_legacy_redirect_ledger, assert_no_hosted_ledger, LEGACY_REDIRECT_LEDGER). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Teach the real-uv VEX matrix the ledger-free hosted mode A hosted uv flow writes no ledger: assert that, run the embedded steps online (no local record), expect nothing discovered once the wiring is reverted, and take a hosted production leg's record from the public patch API instead of the removed ledger. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Run the real-uv hosted revert against the v5 rollback contract A hosted uv flow commits no .socket/, so copy_tree tolerates a missing source; rollback needs --patch-server-url for a pin on the mock origin, and restores each pin to its upstream registry entry or refuses it with the version-control hint (asserting nothing was written) instead of replaying ledger bytes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Delete the hosted ledger replay engine; keep a read-only legacy loader v5 derives hosted state from lockfile pins and restores upstream entries by re-resolving them from the registry, so nothing reverts recorded ledger fragments any more. Remove the replay machinery and the writers behind it: - patch/redirect/replay.rs and takeover.rs (fragment replay, per-purl npm/cargo/golang revert, redirect_revert_supported) - state.rs persist_redirect_state, drop_superseded_purl, quarantine and the unclassified-edit guards; load_redirect_state stays for migration reads and save_redirect_state stays doc(hidden) for laying down pre-v5 fixtures - npmrc unwind planners, bun.lockb snapshot restore, pipenv/vlt/bun text inverses and the EOL fragment respelling that only replay used - tests that only exercised replay; rewrite round-trips in the poetry/uv suites keep their forward and idempotency assertions (the upstream restore of those formats is covered by upstream_restore_golden) hosted_url_names/hosted_url_version move to a small hosted_url module shared by the bun rewriter and VEX discovery. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Document the ledger-free hosted contract and hosted eject Update CLI_CONTRACT.md, README.md, CHANGELOG.md [Unreleased] and the testing/ecosystem docs to v5's WS1/WS2 behaviour: hosted mode writes no redirect ledger; rollback/remove restore hosted pins to their upstream registry entries (per-format coverage, refusals incl. --offline and bun.lockb, the git checkout remedy); list/vex/scan/repair derive hosted state from the lockfiles (details.lockfiles, the new redirectState shape, redirect_ledger_corrupt as a warning); vendor ejects a hosted project and vendor --revert returns to upstream; vendor_supersedes_redirect and hosted_revert_unsupported are gone; the pre-v5 ledger is read for migration only and retired by rollback; new registry env overrides. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite vendor takeover covgap tests to lockfile hosted pins; add WS2 eject tests covgap_commands_vendor: a pre-v5 redirect ledger (even malformed) is now ignored; the takeover guard is driven by hosted pins in package-lock.json on a --patch-server-url origin (dry-run advisory against a mock registry, wet takeover + revert back to the upstream registry entry, offline refusal redirect_revert_failed, unconfigured origin is not hosted). The redirect-ledger write-failure test has no subject anymore and is removed. vendor_eject (new): standalone vendor in a hosted npm project ejects the pins into .socket/vendor/ (no ledger, no manifest), vendor --revert returns to upstream, a failed view fetch is patch_fetch_failed/exit 1, and no pins keeps the no-manifest no-op. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite in_process_vendor hosted takeovers to the v5 upstream restore The berry CRLF takeover and the pnpm/package-lock hosted->vendored conversions no longer see a redirect ledger: vendor runs online against a mock npm registry (SOCKET_NPM_REGISTRY) with the patch-server origin configured, restores the upstream entry, and vendor --revert / rollback land on the upstream registry entry. Adds an offline-refusal test (redirect_revert_failed with the checkout remedy). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * mode_migration_npm: hosted takeovers restore the upstream registry entry Hosted mode writes no ledger; vendor over a hosted yarn pin now runs online with --patch-server-url and restores the upstream entry first (classic legs read a registry document mirrored from the pristine lock via SOCKET_NPM_REGISTRY, so the unwind is hermetic). The reverse classic leg replaces the ledger-originals check with a rollback back to the pristine registry lock. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * mode_migration_cargo: hosted takeovers restore the crates.io entry Vendor over a hosted cargo pin runs online with --patch-server-url and a sparse-index mirror (SOCKET_CRATES_INDEX) of the pristine checksum; the ledger-deletion fail-closed test becomes an offline-refusal test (redirect_revert_failed + checkout remedy) keeping the half-reverted hosted_redirect_live backstop, which now names rollback / git checkout. The hosted leg of the manifest-less VEX matrix fetches its record from the API (no hosted ledger). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * mode_migration_bun: no hosted ledger; unwinds restore the upstream 4-tuple Every run carries SOCKET_PATCH_SERVER_URL (the mock patch origin) and SOCKET_NPM_REGISTRY (a mirror of the pristine lock's integrities), so vendor takeovers, rollback and remove restore the registry 4-tuple; the ledger record/edit assertions become no-ledger assertions. VEX over a stale manifest may name the superseded uuid only in vex_record_superseded. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * in_process_vendor_bun_takeover: hosted pins without a ledger bun: every run names http://patch.test the patch server and points SOCKET_NPM_REGISTRY at one shared registry mirror of the pristine integrities; hand-written hosted fixtures are the lock's URL 3-tuples only, and the ledger record/edit assertions become no-ledger assertions (the unwinds restore the registry 4-tuple byte-exactly). vlt: scan/vendor/get/rollback over a hosted pin run online against the mock origin and a registry mirror; the fixture lock records options.registries so the upstream restore re-derives slot [3] exactly. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e_bun_lockb: binary hosted pins are refused by rollback and vendor takeover v5 restores a hosted pin's upstream entry instead of replaying a ledger, which a binary bun.lockb cannot get: vendor over the hosted pin (dry and wet) and rollback of it now refuse with the git-checkout remedy and write nothing; the tests apply that remedy and continue the round trip. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e_redirect_bun_build: no hosted ledger; rollback restores the upstream 4-tuple The hosted run writes only bun.lock (asserted: no ledger). Rollback and its dry run carry the mock origin as the patch server and a registry mirror of the pristine integrity, so the upstream restore lands on the pre-redirect lock byte for byte; the repeat-run heal is checked in the lock alone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e_redirect_yarn_classic_build: hosted run writes only yarn.lock Assert no redirect ledger is written, and carry .socket/ into the fresh checkout only when it exists (v5 hosted mode writes nothing there). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e_redirect_gem_stale_install: no ledger fallback in hosted mode The re-fire-from-ledger test becomes a re-scan with a failing record fetch: record_fetch_failed with the v5 VEX-omission detail, exit 0, wiring byte-identical, no ledger. Manifest-less VEX drops the ledger-kept cells: a stale unconverged pair and a reverted pair attest nothing. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e_redirect_pnpm_build: no hosted ledger; rollback restores the resolution The hosted run and its idempotent re-run write no ledger (the trust setting and lock splice are checked on disk). The v5/v6 synthetic legs replace the ledger-original checks with a rollback that re-resolves the upstream integrity from a registry mirror and lands on the pristine lock byte for byte; the pinned matrix rollback runs online the same way, and its vex checks fetch the record from the API (no ledger to read offline). .socket/ is copied into fresh checkouts only when present. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e_redirect_cargo_shapes: remove restores crates.io entries without a ledger The post-install vex fetches records from the API (no hosted ledger), and remove runs with a sparse-index mirror of the pristine checksums and the mock origin named the patch server, restoring every shape byte for byte. One inherent v5 difference is pinned: a .cargo config that lacked a final newline gets it back, since without a ledger fragment the rewriter's separator is indistinguishable from a terminated file's. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e_redirect_cargo_build: the three-file rewrite is the whole hosted state Assert no ledger after scan/get --mode hosted; the post-install vex and the manifest-less matrix fetch the record from the API (the ledger-kept cells go), and reverted locks attest nothing. .socket/ travels into the fresh checkout only when present. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * golang hosted e2e: go.mod/go.sum is the whole hosted state get --mode hosted in a module and in a go.work root now asserts no redirect ledger is written; the manifest-less VEX tail already covers the ledger-less shape. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Format the rewritten vendor/hosted test files Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Pin that a transactional eject emits no per-purl takeover warning Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Document the transactional eject, offline refusal and contested wiring Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Drop a stale comment on the hosted unwind error Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Prove eject from a fresh hosted checkout; verify hash-pinned requirements The eject restores every hosted pin's upstream registry entry before the vendor engine takes its source inventory, so a fresh checkout (no node_modules, empty CARGO_HOME, no virtualenv) fetches the pristine source from the registry and verifies it against the restored checksum. New subprocess tests pin that for npm, cargo and pypi. The pypi case exposed that requirements.txt `==` pins never carried their `--hash=sha256:` digests into the lock inventory, so a hash-pinned pin with no installed copy was unverifiable. Carry them as the entry's integrity (any-of, like Pipfile.lock) while the file resolves from the public index; an index option keeps every pin unverifiable. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Give the stale-Pipfile rollback test a derivable hash mode A requirements.txt whose every line is a hosted pin is refused by the upstream restore (hash-checking mode is not derivable). Add an unhashed, unpatched sibling so the test still pins a successful restore. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * node addon smoke: a wet hosted session writes no ledger The in-memory hosted engine emits only lockfile/config edits in v5, so assert no .socket/ output instead of requiring the redirect ledger. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * covgap scan_hosted: unreadable-workspace case asserts no ledger Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Rewrite the CI-only e2e suites and backtests to the ledger-free hosted contract v5 hosted mode writes no .socket/vendor/redirect-state.json: every #[ignore]d real-toolchain capstone and every compatibility backtest that read the redirect ledger now asserts it is absent and reads the facts from the lockfile (or the public API record) instead. - npm / yarn berry / composer / gem / maven / nuget / rush / poetry / pdm / hatch / pypi-real / yarn-classic matrices: no-ledger assertions; vex offline with no local record is record_unavailable; a reverted hosted lock names the patch nowhere (manifest_not_found, exit 2); a manifest-less `vendor --vex` is no longer run over hosted pins (it ejects them now). - Mock-origin hosted pins pass --patch-server-url to vex / rollback / remove / vendor / scan --mode vendored; the vlt harness gains rollback_upstream / Fixture::rollback with SOCKET_NPM_REGISTRY pointed at the harness registry. - Rollback legs expect the upstream registry entry back: npm (real registry, JSON-equal lock, .npmrc removed), composer 2 (byte-identical from packagist; composer 1's inline repository refuses with the git checkout remedy), maven (--offline, byte-identical pom), poetry (real PyPI), vlt (byte-exact per era); the production npm leg gains a real-registry rollback on a copy. - vlt: URL-less / update-dropped pins now find no state ("Manifest not found"); the TS-written lock restores to its input and the pre-v5 ledger is retired; mode migrations restore upstream before vendoring. - Backtests (bun, pdm, pipenv, poetry, vlt, uv docs): hosted records come from the public /patch/view/<uuid>; noLedger checks are unconditional for hosted; rollback checks expect the upstream entry (bun custom registry slot comes back as "", bun.lockb refuses with the remedy); the vlt downgrade leg asserts the pin with no ledger. Core fix: the vlt upstream restore added a slot [3] tarball URL to a single-node lock whose options record `registry` (vlt rc.33 .. 1.0.4 with config.registry), which vlt itself never writes (save.ts omits the resolved URL when it starts with the configured registry); rollback was not byte-exact on those eras. The no-siblings fallback now honours that rule (unit-tested). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Hold the vlt configured-registry slot [3] rule pending real-vlt evidence The rule in 44240a2 drops slot [3] when a lock records options.registry, matching vlt rc.33..1.0.4, but it breaks the vlt-lock-v1-both-registry-keys golden, and newer vlt releases have not been checked yet. Restore the previous behaviour until real captures decide it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Backtests: send a User-Agent on public patch-view fetches The hosted backtest legs now read the patch record from the public proxy's /patch/view/<uuid>, and CI got HTTP 403 for Python's default urllib User-Agent (the poetry 1.4-2.x native legs). Send an explicit agent, as the vlt backtest's api_get already does. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Re-land the vlt configured-registry slot [3] rule on real-vlt evidence vlt's lockfile save (identical in 1.0.0-rc.33, 1.0.4, 1.0.10 and 1.2.0) writes a default-registry node's resolved URL (slot [3]) only when no `registry` is configured or the URL does not start with it, and records that `registry` in `options.registry`. Real installs of left-pad@1.3.0 with `config.registry` + `registries.npm` set to the default registry produce a byte-identical 3-tuple lock on all four releases. - upstream::vlt: the no-siblings fallback of `records_url` no longer adds slot [3] under a recorded `options.registry` the node resolves under (unit test covers both sides). - fixtures: `lock-v1-both-registry-keys` was hand-written with slot [3]; its input (and expected-edits original) now match the real lock byte for byte, and the real capture is added as `capture-1.2.0-config-registry` (vex-discover golden extended). - docs/testing/vlt-compatibility.md records the rule and the evidence. Without the rule the real-vlt legs that restore a single-node lock of a `config.registry` project (the harness configures it for rc.33 .. 1.0.4) are not byte-exact: hosted idempotence / crlf_lock and the migration scoped_unwind / rollback_from_mixed / agent_rollback_after_takeovers legs. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * hosted-e2e uv leg: serve the record of the uuid the lock pins pypi_uv_lock_hosted_install_proof handed all three PYPI_UUIDS to uv_vex::production_manifestless. With no local record (v5 hosted keeps no ledger) production_record takes the FIRST of those the public proxy answers for, so the VEX stand-in served de58c8b8 while the resolver had granted (and uv.lock pinned) e828efa5: every manifest-less cell then asked the stand-in for e828efa5 and got `vex_record_not_found` / `record_unavailable`. Pass exactly the wired uuid instead; if production ever grants a uuid its public view will not serve, production_record now fails naming it rather than masking it behind another patch's record. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * PDM static_urls restore writes files in URL order; bun backtest User-Agent on artifact fetches PDM orders a static_urls entry's files by URL (the sdist under 0c/39 precedes the wheel under b0/53 for urllib3 1.26.18); the upstream restore wrote them in filename order, so hosted rollback wasn't byte-exact. The golden now uses real bucketed URLs so the order differs from filename order. The bun backtest's lockb digest and asset downloads now send the same User-Agent as the patch-view fetch (patch.socket.dev 403s urllib's default). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * covgap rollback: macOS-only blob-pin check derives hashes from the fixture bytes The fixture no longer carries before_hash/after_hash; the macos-gated manifest-write-failure test still read them and broke the macOS build. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Round-trip hosted rollback of unpopulated Poetry 1.0/1.1 locks Poetry 1.0/1.1 record `[metadata.files] <name> = []` against today's PyPI JSON API, while locks written earlier list every release file. The hosted rewrite replaced either with the one-entry patched array, so rollback could not tell them apart and always re-derived the full release list: the backtest's `direct` and `crlf` shapes (literal `urllib3 = []`) failed rollbackRestoresLockBytes on Poetry 1.0.10 and 1.1.15. The rewriter now keeps that bit in the patched entry's layout: one file per line (Poetry's own rendering) when the original listed files, inline when it was `[]`; a re-run keeps the layout it finds. The restore reads it back and writes the full release list or `[]`. Adds a golden round-trip over every native fixture generation (1.0.10 to 2.4.3), LF and CRLF, alongside the existing populated-shape one. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * yarn berry e2e: manifest-less VEX matrix expects no hosted ledger The berry matrix still asserted the pre-v5 .socket/vendor redirect ledger after a hosted flow. v5 hosted mode writes none: the offline cell now expects record_unavailable, a reverted hosted checkout discovers nothing (exit 2 manifest_not_found), and manifest-less apply --vex is a calm no-op. The yarn4 pnpm-linker and workspaces fresh checkouts copy .socket/ only when it exists, as the node-modules berry test already does. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * e2e vex poetry: accept the one-file-per-line metadata.files pin A populated lock-1.0/1.1 [metadata.files] entry now keeps Poetry's multi-line layout after the hosted rewrite (so rollback can tell it from an originally empty one); the pin-spelling matrix strips that form too. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Keep Pipenv's index on hosted entries so rollback restores it exactly The hosted Pipfile.lock rewrite dropped each entry's `index`, and the ledger-free upstream restore then guessed it from sibling registry entries (none -> the PyPI source name; all siblings index-less -> none). Pipenv's own choice cannot be re-derived from the lock or the Pipfile: for the same Pipfile it depends on the release and the locking environment (measured with real `pipenv lock`): shape 2018.11.26 2020-2022 2023.12.1-2026.8.0 direct pypi pypi pypi extras table pypi pypi (none) marker-excluded pypi (none) (none) transitive (none) (none) (none) So the backtest's rollbackRestoresLockBytes failed on 2022.12.19 extras (transitive pysocks sibling has no index -> index dropped) and on 2022.12.19 / 2026.8.0 marker-excluded (no siblings -> "pypi" added), and rollbackAfterRelockRetires failed on 2026.8.0 marker-excluded: the relock hybrid (our `file` kept, `version` and registry `hashes` restored, no `index`) was restored with an `index` Pipenv never wrote. The hosted rewrite now keeps `index` exactly as Pipenv wrote it (present or absent) and only drops `version`; the restore carries the entry's `index` back unchanged instead of choosing one, refusing when it (or the Pipfile's explicit index) does not name a PyPI source in `_meta.sources`. A `file` entry carrying `index` installs the referenced wheel itself (direct_url.json present) on Pipenv 2018.11.26, 2020.11.15, 2021.11.23, 2022.12.19, 2023.12.1, 2024.4.1, 2025.1.3 and 2026.8.0 (`install --deploy`, `sync`, `verify`), and a marker-excluded one stays uninstalled; Pipenv 7-11 ignore `index` on a `path` entry (convert_deps_to_pip skips it for file/path deps). Tests: real Pipenv 2018.11.26 / 2022.12.19 / 2026.8.0 locks for the extras and marker-excluded shapes (tests/fixtures/pipenv-shapes) round trip byte for byte in LF and CRLF, and the 2026.8.0 marker-excluded relock hybrid restores the pristine bytes. The backtest's allCategoriesRewritten now expects hosted entries to keep the pristine `index` (vendored still drops it). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Restore hosted bun.lockb pins natively for the vendor takeover With the hosted ledger gone, `vendor` over a live hosted pin in a binary bun.lockb (Bun 0.8.1-1.1.x's default lock, Bun 1.2's legacy lock) was refused `redirect_revert_failed`: format_of() mapped bun.lockb to Unsupported, so the backtest matrix's hosted-then-vendored cell exited partial_failure. The upstream restore gains a bun.lockb restorer (patch/redirect/upstream/bun_lockb.rs). It rebuilds each hosted remote-tarball record as Bun's npm registry record for name@version, from the registry's dist.tarball / dist.integrity (SOCKET_NPM_REGISTRY aware), via the new BunLockb::set_registry_package. That function re-interns the URL (re-using the original pool offset), drops the hosted URL string from the pool tail and re-derives the metadata hash. The staged restore view now carries binary files. To make the rebuild byte-exact, set_package keeps the registry record's inactive bytes (padding, semver) when it writes a remote tarball record. Early writers leave uninitialized padding there (Bun 0.8.1), so this matters. A re-pin to a later grant's URL now drops the superseded URL from the pool. A record without the retained bytes is rebuilt the way Bun writes one; prerelease versions are refused in that case. The rebuild is exact for every fixture writer except a format-1 lock (kept promoted) and workspace locks (behaviors kept normalized). So only the vendor takeover and eject opt in (RestoreOptions::bun_lockb): their vendor ledger records the rebuilt record, and `vendor --revert` returns the pre-hosted bytes. `rollback` / `remove` keep refusing a hosted bun.lockb pin with the `git checkout -- bun.lockb` remedy, as the harness's rollbackLockbRefused expects. An offline vendor still refuses. Tests: core restorer tests over every real fixture, codec tests for the rebuild (retained and zeroed records, re-pin), a hermetic CLI test (vendor_eject_bun_lockb.rs: takeover, eject, rollback and offline refusals on Bun 0.8.1 / 1.1.38 / 1.2.0 locks), and the real-Bun e2e_bun_lockb takeover now vendors online and reverts byte-exact. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * bun backtest: inject the custom-registry slot with byte I/O On Windows, Path.write_text turned the injected bun.lock's LF into CRLF, so the LF pre-injection bytes could never match the (EOL-preserving) upstream restore: custom-registry hosted rollbackOriginalFiles failed on Windows only. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Demote hosted format-1 bun.lockb locks exactly on the vendor takeover The bun backtest's binary job failed only for the format-1 writers (Bun 0.1.1 / 0.1.6, readers 0.5.9 and 1.4.2) after 65aa074: the hosted rewrite promotes a format-1 lock to format 2 (format 1 has no URL column), and the new native upstream restorer rebuilt the registry record inside that promoted lock. The vendor ledger then recorded the promoted bytes as its pre-vendor original, so `vendor --revert` returned a format-2 lock and e2e_bun_lockb's "the revert restores the pre-hosted bytes exactly" assertion failed. A promoted lock is byte-for-byte indistinguishable from one Bun 0.1.7+ wrote (same pool order, same metadata hash), and hosted mode keeps no ledger, so the restorer could not tell. The codec now marks a lock whenever an edit had to normalize it: seven magic bytes and a flag byte in the last eight bytes of the root package's resolution (the root resolution's value union, which no Bun reader reads; early writers leave uninitialized bytes there, and 1.4.2 / 0.5.9 read such locks). For a promoted lock those bytes are new, so the mark overwrites nothing. promote_legacy_format sets NORMALIZED_FORMAT_1; normalize_workspace_behaviors sets NORMALIZED_WORKSPACE when it changes a dependency behavior or workspace literal. The vendor ledger's layout_original path normalizes the original the same way, so its exact revert is unchanged. The bun.lockb upstream restorer then: - demotes a NORMALIZED_FORMAT_1 lock back to format 1 once every hosted record is rebuilt (BunLockb::demote_legacy_format, which drops the URL column and the URLs the promotion appended to the pool, and succeeds only if promoting the result again reproduces the lock byte for byte). Otherwise the pins are refused with the `git checkout -- bun.lockb` remedy. - refuses a NORMALIZED_WORKSPACE lock outright with that remedy, since clearing the workspace behavior bit cannot be undone. It no longer takes the lock over non-exactly. Tests: the upstream restorer's byte-exact test now covers 0.1.1 / 0.1.6. Workspace-normalized extension locks refuse, and so does a lock whose mark carries an unknown flag. The codec rebuild test checks the exact demotion. vendor_eject_bun_lockb adds the 0.1.1 / 0.1.6 takeover with an exact revert and a workspace-lock refusal. Locally with real Bun, the 1.4.2 reader x 0.1.1 / 0.1.6 writer cells now pass the takeover and the exact revert. They then stop at the 0.5.9 legacy reader, which segfaults on any networked install in this sandbox. The 1.1.45 / 1.2.0 / 1.4.2 cells pass everything except the `extensions` shape, which needs api.github.com (403 here). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * ci: give the Windows test leg a 50-minute budget The Windows leg runs the same suite ~1.6x slower than macOS. On the base branch it already took 34m40s of the flat 35-minute budget, and with this PR's added tests it was cancelled at the limit mid-run (no failures). Linux and macOS keep 35 minutes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ * Check out the Poetry and Pipenv lock fixtures byte-exact on Windows The Windows test leg's CRLF checkout (core.autocrlf) turned the LF-committed tests/fixtures/poetry, pipenv and pipenv-shapes locks into CRLF, so the byte-exact upstream-restore round trips (and their derived CRLF variants, which became \r\r\n) and the Poetry VEX pin-spelling test failed on Windows only. Mark them -text like the other captured-lock fixtures. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018tncU7u7FK48tubFNA12UQ --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * WS5: one VendoredBackend for vendored apply/revert/repair; cut repair's ledger rebuild (#283) * Consolidate vendored apply/revert/repair into one VendoredBackend vendor, scan/get --mode vendored, vendor --revert, rollback's vendored leg, remove and repair now go through one VendoredBackend { apply, revert, repair } over the shared engine (vendor_records_reusing, dispatch_revert_one_opts). The boxed_* scan shims collapse into one boxed_vendor_step; the engine future stays boxed inside apply for the Windows 1 MiB main-thread stack. repair no longer re-synthesizes vendor ledger entries from lockfiles. A lockfile reference with no ledger entry fails with vendor_ledger_missing (artifact-level event: uuid + details.{ecosystem,path}); the remedy is restoring state.json from version control. Missing or corrupt artifacts are re-vendored through the same engine as vendor, so the patch service's prebuilt artifact is downloaded first under --vendor-source auto, with the local build as the fallback. The fingerprint post-verify, set-aside of corrupt bytes and carried-inventory refresh are kept. Removed with the rebuild: repair_vendor.rs, gem Gemfile wiring reconstruction, and registry_fetch::fetch_npm_unverified. The packing code (npm_pack, pypi_wheel, berry_zip, registry_fetch, prestage) stays: depscan does not call it (verified against depscan master 784013d6), but it is the CLI's own --vendor-source build/auto fallback. Tests for the reconstruction path are replaced by vendor_ledger_missing pins per flavor; CLI_CONTRACT, README, CHANGELOG and the v5 plan are updated. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa * Point the vlt coverage map at the renamed repair test vlt-coverage.json still named vlt_repair_reconstructs_the_ledger_from_the_lock, and vendor_vlt_lock_out_of_sync lost the only assertion the coverage check could see when the lock-only reference test switched to vendor_ledger_missing. vendor_vlt_out_of_sync now asserts the refusal detail. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa * Drop the ledger-less mirror leg from the bun binary workspace test native_binary_alias_and_transitive still expected repair to rebuild a workspace mirror after deleting state.json. Repair now reports that as vendor_ledger_missing (pinned in native_binary_hosted_vendored_takeover_roundtrip), so the leg is gone; the missing/corrupt mirror legs keep running and assert the ledger stays byte-identical. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa * Pass --mode agent to the gated agent-mode get fixtures get defaults to hosted since 5e5f5ed, which moved the agent-mode fixtures to --mode agent but missed the #[ignore]d real-toolchain suites (e2e_vlt, e2e_npm, e2e_pypi, e2e_gem, e2e_safety_pnpm). Their plain `get <uuid>` now redirects instead of applying in place, so e.g. vlt_pinned_matrix_agent_get_and_remove saw the copy Absent. This PR touches the vlt-compatibility path filter, which surfaced it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa * Keep an unpersistable inventory refresh loud, drop reconstruction persist tests A carried-inventory refresh whose ledger write fails now reports vendor_inventory_refreshed next to vendor_state_write_failed and keeps the member-verified rebuild on disk, instead of falling through to vendor_artifact_rebuild_failed and removing it. The persist-failure tests for the removed backfill / anchored / soft reconstruction paths go with them; they only run as non-root, which is why the root sandbox skipped them. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa * Restore the macOS-only PatchedFixture hash fields The dead-field clippy fix removed before_hash/after_hash, but the macOS immutable-flag rollback tests read them, so test (macos-latest) no longer compiled. Keep the fields and allow dead_code off macOS only. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa * Keep repair on the recorded artifact identity Review of #283 found two regressions in the shared-engine repair: - A corrupt artifact was moved aside before staging, so its afterHash-verified members were no longer harvested: an offline repair that the previous implementation completed failed with "no local source". The members are now harvested first and passed as the seed. - The post-verify reloaded the ledger the re-vendor had just written, so a service archive with different bytes (same members, new gzip mtime) was committed as `rebuilt` with a rewired lock and new fingerprint. Repair now verifies against the original entry; when the result is not the recorded artifact, that candidate's wiring files and ledger entry are put back from a pre-run snapshot, and a service copy falls back to a build-only rebuild. Legitimate backend migrations of a verified rebuild (the cargo version retag) are kept. Both reproductions are pinned in repair_vendor_e2e. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa * Make repair's wiring undo FIFO-safe, atomic and symlink-preserving Bugbot on #283: the identity-undo snapshot read lockfiles with bare tokio::fs::read (a FIFO at a wiring path blocks open(2)) and skipped symlinked lockfiles, and the put-back wrote them in place (no stage+fsync+rename, mode bits dropped). The snapshot now reads through read_regular_to_bytes and records a symlinked lockfile's link text; the undo re-links a link that the engine's rename replaced, then writes the target through atomic_write_bytes_preserving_mode. Pinned by repair_identity_undo_follows_a_symlinked_lockfile. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DMhWChtaNX5FJYNDq3NBJa --------- Co-authored-by: Claude <noreply@anthropic.com> * v5 fix: unblock the e2e tier on the v5 base (#288) * Cancel only superseded PR runs in CI A CI or compatibility run is now cancelled only when a newer push to the same pull request replaces it. Push, dispatch and scheduled runs always finish, so a manually dispatched run on the v5 base branch (its only CI verdict, since push CI runs on main alone) is no longer killed by a later dispatch or by the non-main cancel rule, and main keeps finishing its rust-cache saves. CI now groups PR runs by PR number, like the compatibility workflows already do. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp * Expect native path separators in scan headers scan_hosted_paths_run_once_per_project_directory compared the per-directory `== apps/a ==` header against a literal forward-slash path, but scan prints the directory glob matched, which Windows spells `apps\a`. The Windows test leg failed on this since 62f07c7, and because every e2e job waits on `test`, the whole e2e tier was skipped on v5 PRs. Build the expected header from path components. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp * Run the vlt agent get test in agent mode get defaults to hosted mode since 5e5f5ed, so the real-vlt get_and_remove leg ran a hosted get and found the installed copy unpatched (Absent, expected Patched). Pass --mode agent, as the other agent-mode fixtures already do. Same change as cc5f1b6 on #285; it blocked the e2e tier's e2e_vlt jobs now that they run. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp * Run the pnpm safety e2e gets in agent mode get defaults to hosted mode since 5e5f5ed, so the three pnpm safety tests ran a hosted redirect and found proj_a's installed copy unpatched and no pnpm-layout note. They test the in-place apply path, so pass --mode agent. These e2e-tier tests had not run since that change because a red base test skipped the tier. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp * Let the NuGet hosted e2e run without .socket/ v5 hosted mode writes no `.socket/` directory (the lock pins are the whole hosted state), so the hosted leg's fresh_checkout panicked with NotFound copying a tree that no longer exists. Copy it only when the run left one; the vendored leg still carries its ledger through. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AfGnDQMy2FvpExneXmR1qp --------- Co-authored-by: Claude <noreply@anthropic.com> * Add the NuGet vendoring design (docs only) (#285) * Drop unused fields from the rollback covgap fixture `cargo clippy --all-targets -- -D warnings` failed on the dead before_hash/after_hash fields of PatchedFixture. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01R11RZRvYFL3fzmEnkFAU4A * Add an opt-in NuGet fallback vendoring layout Setting SOCKET_PATCH_NUGET_LAYOUT=fallback vendors a patched NuGet package under a Socket-only version V' (the upstream version plus a 4th part derived from the patch uuid). The package is committed already extracted, as a NuGet fallback package folder under .socket/vendor/nuget/<uuid>/. It is wired through a generated socket-patch.targets file…

Mikola Lysenko (mikolalysenko) mentioned this pull request Sep 28, 2026

docs: v5 waste review + repacking-to-depscan design #286

Closed

Mikola Lysenko (mikolalysenko) mentioned this pull request Sep 28, 2026

docs: v5 waste review and repacking design (reference) #289

Merged

Mikola Lysenko (mikolalysenko) mentioned this pull request Sep 28, 2026

v5 CI: build e2e binaries once and tier the PM matrix #291

Merged

Claude (claude) added 2 commits September 28, 2026 13:02

Merge remote-tracking branch 'origin/release/v5-prerelease' into v5/u…

7e1758b

…nblock-e2e-tier

Mikola Lysenko (mikolalysenko) mentioned this pull request Sep 28, 2026

v5 prerelease: scan → vex → vendor workflow, hosted by default #277

Merged

Mikola Lysenko (mikolalysenko) marked this pull request as ready for review September 28, 2026 14:08

Mikola Lysenko (mikolalysenko) merged commit c7aa5a5 into release/v5-prerelease Sep 28, 2026
539 of 586 checks passed

Mikola Lysenko (mikolalysenko) deleted the v5/unblock-e2e-tier branch September 28, 2026 16:35

Mikola Lysenko (mikolalysenko) mentioned this pull request Sep 28, 2026

WS3: one lockfile model per ecosystem #281

Merged

This was referenced Sep 28, 2026

v5: socket.yml patch rollout config and filtering #293

Merged

v5: cap new patches per scan, most critical first #294

Merged

v5 WS4/WS6: one hosted engine for disk + memory; unified Ledgers view #282

Merged

Conversation

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026 • edited by cursor Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026 •

edited by cursor Bot

Loading

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026 •

edited

Loading

Mikola Lysenko (mikolalysenko) commented Sep 28, 2026 •

edited

Loading