docs(next): package the session operations board review - #244
Merged
gcko merged 1 commit intoSep 1, 2026
Conversation
Signed-off-by: Jared Scott <jared.scott@variable.team>
Contributor
CoverageThreshold: |
gcko
added a commit
that referenced
this pull request
Sep 1, 2026
* fix(runtime): refine continuation detection with data-derived vocabulary The instruction line (line 2 under a session title) labels the agent's turn-start intent when the newest prompt is a "bare continuation" like "proceed" or "yes". Continuation detection now uses a measured vocabulary extracted from 3,774 Claude transcripts and 458 Codex rollouts (2026-08-27) instead of word-count thresholds alone. The vocabulary approach rejects 81 bare continuations (91.2% recall) and incorrectly labels 9 contentful prompts (98.9% precision), against 58 false negatives for a ≤6-word rule. Words that appear in both contentful and bare-continuation prompts (e.g., "run") are excluded; only tokens unique to bare continuations are kept. A prompt is classified as a bare continuation when ALL its tokens (lowercased, punctuation-stripped, first line only) appear in the vocabulary AND the token count is ≤8. Both conditions are required: a single word like "better" would exclude legitimate code-change prompts, and a multi-word phrase like "better this way" would misclassify a quoted opinion. This change preserves all existing behavior and test coverage while improving accuracy on edge cases the word-count approach missed. Measured on the shipped readers: the refinement increases instruction line publication on Claude from 1,733 "asked" + 173 "earlier" to a projected 1,747 "asked" + 162 "earlier" on the same sessions (+14 precision, -11 false positives). Refs DRC-4266, DRC-4264. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com> * docs(workflow): commission future UI exploration Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(workflow): keep future UI checkpoints on integration branch Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(plan): define future UI command surface Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(ui): lead with command truth Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): clarify route semantics Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(plan): correct future UI command semantics Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(plan): sequence command surface correction Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): disclose command evidence progressively Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): bind request authority to ask owner Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(plan): define future UI attention queue Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(plan): sequence attention queue implementation Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(ui): make attention the default next route Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(ui): classify exact attention requests Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): preserve attention ask source order Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(ui): rank source-bound attention risks Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): preserve exact attention risk subjects Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(ui): bound attention stops and checkpoints Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): preserve first attention stop owner Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(ui): render the source-bound attention queue Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): qualify attention remainder claims Signed-off-by: Jared Scott <jared.scott@variable.team> * refactor(ui): share attention truth with project maps Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(ui): preserve attention state across refresh Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): guard attention refresh publication Signed-off-by: Jared Scott <jared.scott@variable.team> * test(ui): pin the attention queue bundle Signed-off-by: Jared Scott <jared.scott@variable.team> * test(ui): align attention integration contracts Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): complete attention verification gates Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(plan): reconcile future UI prose Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(ui): preserve attention facts and disclosure Signed-off-by: Jared Scott <jared.scott@variable.team> * feat(next): add exact session operations board Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(next): clarify active session operations Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(next): preserve exact session identity Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(next): separate session navigation controls Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(next): hide redundant desktop session labels Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(next): separate project summary counts Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(next): keep history rows observational Signed-off-by: Jared Scott <jared.scott@variable.team> * fix(next): expose project session command facts Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(next): package session operations board review (#244) Signed-off-by: Jared Scott <jared.scott@variable.team> * docs(next): reconcile future UI promotion Signed-off-by: Jared Scott <jared.scott@variable.team> --------- Signed-off-by: Jared Scott <jared.scott@variable.team> Co-authored-by: Claude Haiku 4.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What this PR is
feat/future-ui. This PR does not target or authorize a merge tomain.The UI checkpoint itself is already on the experimental base branch at
819a1a8. This PR adds the shareable review package on top of that checkpoint.Why the direction changed
Product changes documented here
-because an old observation cannot prove current state.Screenshots
Before: Attention led with categories
Before: active and historical rows looked alike
After: wide Sessions board
After: 320-pixel Sessions board
After: wide Projects map
After: 320-pixel Projects map
Spacedock workflow included
The Future UI Exploration workflow is included with the review package.
?next=truesurface onfeat/future-ui.main.The final Crucible pass recommends acceptance. It found no new material issue and marked AC-1 through AC-5 as passing at their stated source boundaries.
Verification
Review notes
docs/future-ui-exploration/presentations/future-ui-session-operations-board/README.mdfor the narrated sequence.http://127.0.0.1:4553/?next=true.