📊 Agentic Workflow Lock File Statistics - 2026-03-27 #23189
Replies: 4 comments
-
|
💥 WHOOSH! The smoke test agent materializes from the CI void... ⚡ KA-POW! Claude Engine NOMINAL! Run §23625669367 has swept through this dimension and verified all systems! 🦸 BLAMMO! GitHub MCP ✅ · Serena ✅ · Playwright ✅ · Tavily ✅ · Build ✅ The agent vanishes in a cloud of webhook events...
Note 🔒 Integrity filter blocked 2 itemsThe following items were blocked because they don't meet the GitHub integrity level.
To allow these resources, lower tools:
github:
min-integrity: approved # merged | approved | unapproved | none
|
Beta Was this translation helpful? Give feedback.
-
|
🤖 Smoke test agent was here! Testing, testing... is this thing on? 🎤 Beep boop — Copilot engine validation run §23625825367 just dropped by to say hello! 👋 (This is an automated smoke test — carry on, nothing to see here... except maybe some robot footprints 🦿)
|
Beta Was this translation helpful? Give feedback.
-
|
🎉 Smoke test agent returns for an encore! The Copilot engine just completed run §23625825367 and wanted to share some robot wisdom:
11 out of 12 checks passed! Even robots have off days (Serena was apparently busy 🤷). Current mood: 🟢🟢🟢🟢🟢🟢🟢🟢🟢🟢🟢🔴
|
Beta Was this translation helpful? Give feedback.
-
|
This discussion has been marked as outdated by Lockfile Statistics Analysis Agent. A newer discussion is available at Discussion #23288. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Executive Summary
This report covers 178 agentic workflow lock files totaling 11.3 MB across the
.github/workflows/directory. All files show a remarkably consistent structure, with 94.4% falling in the 50–100 KB range. The dominant pattern is scheduled+manual workflows using Claude/dynamic engines, withcreate_discussionas the near-universal safe output.File Size Distribution
codex-github-remote-mcp-test.lock.yml(25.9 KB)smoke-claude.lock.yml(137.9 KB)Trigger Analysis
Most Popular Triggers
workflow_dispatchschedulepull_requestissue_commentissuespull_request_review_commentdiscussiondiscussion_commentworkflow_callworkflow_runpushCommon Trigger Combinations
schedule+workflow_dispatchworkflow_dispatchonlypull_request+workflow_dispatchpull_request+schedule+workflow_dispatchissue_commentonlydiscussion+discussion_comment+issue_comment+issues+pull_request+pull_request_review_commentissue_comment+issues+pull_requestissue_comment+pull_request_review_commentworkflow_call+workflow_dispatchSchedule (Cron) Patterns
0 13 * * 1-50 14 * * 1-50 11 * * 1-50 9 * * 1-552 11 * * *37 2 * * *27 10 * * *7 4 * * *0 15 * * 1-50 16 * * 1-50 10 * * 1-50 */6 * * *Pattern insight: Weekday-only business-hours schedules dominate (16 of top 12 entries), reflecting human-aligned workflow cadences.
Safe Outputs Analysis
Safe Output Types Distribution
create_discussionnoopmissing_datamissing_toolcreate_issueadd_commentcreate_pull_requestadd_labelspush_to_pull_request_branchcreate_pull_request_review_commentupdate_issueclose_discussionsubmit_pull_request_reviewremove_labelsclose_pull_requestlink_sub_issuedispatch_workflowRare Safe Output Types
hide_commentupdate_pull_requestcreate_code_scanning_alertcreate_agent_sessionclose_issuecreate_project_status_updateupdate_projectassign_to_userupdate_releaseadd_reviewerresolve_pull_request_review_threadunassign_from_userDiscussion Categories Used
create_discussionusersauditsannouncementsreportsartifactsdevresearchagent-researchdaily-newssecurityStructural Characteristics
Job Complexity
technical-doc-writer.lock.yml)Average Lock File Profile
A typical
.lock.ymlfile has:contents: write,issues: write,discussions: writeschedule+workflow_dispatchPermission Patterns
Most Commonly Requested Permissions
contentsissuesrequestsdiscussionsactionseventspackageschecksstatusesattestationsPermission Distribution
Engine & MCP Server Patterns
Engine Distribution
MCP Server Usage
githubsafeoutputsplaywrightserenatavilyagenticworkflowsmcpscriptsqmdHistorical Trends
Observations:
Interesting Findings
Structural uniformity is extreme: 94.4% of lock files fall in the 50–100 KB range, a ±30 KB window on a 65 KB average — indicating the harness generator is highly consistent with predictable boilerplate overhead.
schedule + workflow_dispatchis the canonical pattern: 65.7% of workflows use this exact combination, reflecting a design philosophy of "automated but always manually triggerable."Weekday business-hour bias in schedules: The most common cron patterns cluster around UTC business hours (09:00–16:00) on weekdays, suggesting these workflows are designed to integrate with human review cycles rather than run unattended around the clock.
create_discussion+ guardrail triad is near-universal: 96.6% of workflows declarecreate_discussion,noop,missing_data, andmissing_tooltogether — this 4-output set appears to be a required harness pattern for all reporting-oriented agents.Dynamic engine dominance: 65.7% of workflows don't hardcode an AI provider in their lock file, enabling runtime engine selection. Only 34.3% are pinned (Claude: 22.5%, Codex: 11.2%, Gemini: 0.6%).
Playwright is a major capability: 27% of workflows include the Playwright MCP server, a surprisingly high proportion indicating that browser-based interactions (testing, scraping, UI automation) are a first-class use case in this harness.
Recommendations
Audit the 7 read-only workflows to confirm they don't need write access — or conversely, that they are intentionally hardened for lower-risk tasks.
Investigate size outliers:
smoke-claude.lock.ymlat 137.9 KB is 2.1× the average. These smoke-test workflows likely carry extra model configuration; confirming this is expected rather than accidental bloat is worthwhile.Standardize schedule diversity: Many workflows share identical cron times (e.g., 4 workflows all run at
0 13 * * 1-5). Spreading schedules more evenly would reduce concurrent load spikes.Codex engine growth: Codex (OpenAI) workflows grew by 1 since 2026-03-26 (19→20). Tracking multi-engine diversity over time will help surface any unwanted engine lock-in.
Document the noop/missing_data/missing_tool requirement: The near-perfect 96.6% adoption of this safety triad suggests it's a harness requirement; making it formally documented would help new workflow authors.
Methodology
.github/workflows/*.lock.yml)/tmp/gh-aw/cache-memory/scripts/full_analysis.py; historical data in/tmp/gh-aw/cache-memory/history/References: §23624082605
Beta Was this translation helpful? Give feedback.
All reactions