Lockfile Statistics Analysis - 2026-02-25 #18381
Replies: 3 comments
-
|
🤖 Beep boop! The smoke test agent was here! Running diagnostic checks... All systems nominal! The automation hamsters are spinning their wheels at peak efficiency. 🐹⚡ This message was left by your friendly neighborhood smoke test bot. Don't mind me, just making sure everything works!
|
Beta Was this translation helpful? Give feedback.
-
|
💥 WHOOSH! 🦸 The Smoke Test Agent has arrived! ⚡ KAPOW! Claude engine validation complete — all systems are GO! ✨
🔥 ZAP! BOOM! BANG! — Claude was here and everything survived! 🦾
|
Beta Was this translation helpful? Give feedback.
-
|
This discussion was automatically closed because it expired on 2026-02-26T16:53:56.686Z.
|
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Overview
This report analyzes all 158
.lock.ymlfiles in.github/workflows/as of 2026-02-25. The repository has reached a stable count of 158 workflows, with total lockfile storage at ~9.8 MB (avg 63.5 KB per file). The dominant pattern is scheduled + manually-dispatchable workflows using thedynamicengine — suggesting a mature, primarily automated workflow suite built around periodic reporting and analysis tasks.Executive Summary
codex-github-remote-mcp-test(24.5 KB)smoke-claude(142.8 KB)File Size Distribution
The overwhelming majority (91.8%) fall in the 50–100 KB band, indicating a remarkably uniform structural template across workflows.
Top 5 Largest & Smallest Files
Largest:
smoke-claude.lock.ymlsmoke-copilot-arm.lock.ymlsmoke-copilot.lock.ymlpoem-bot.lock.ymlSmallest:
codex-github-remote-mcp-test.lock.ymlchroma-issue-indexer.lock.ymlexample-custom-error-patterns.lock.ymlexample-permissions-warning.lock.ymlfirewall.lock.ymlTrigger Analysis
Most Popular Triggers
workflow_dispatchschedulepull_requestissue_commentissuespull_request_review_commentdiscussion_commentdiscussionworkflow_runpushMost Common Trigger Combinations
schedule + workflow_dispatchworkflow_dispatchonlypull_request + schedule + workflow_dispatchpull_request + workflow_dispatchissuesonlyissue_comment + issues + pull_requestworkflow_runonlySchedule Pattern Breakdown
Schedule Frequency Categories
1-5)Most Repeated Cron Expressions
0 14 * * 1-50 13 * * 1-50 11 * * 1-50 9 * * 1-50 */6 * * *0 16 * * 1-5Working-hours (weekday) scheduling is prevalent, clustering around 9 AM–4 PM UTC — suggesting many workflows are tied to developer activity windows.
Safe Outputs Analysis
Safe Output Types Distribution
create_discussionnoopmissing_datamissing_toolcreate_issuecreate_pull_requestadd_commentcreate_pull_request_review_commentadd_labelsupdate_issueclose_issuepush_to_pull_request_branchclose_discussionsubmit_pull_request_reviewremove_labelsdispatch_workflowlink_sub_issueclose_pull_requesthide_commentupdate_pull_requestcreate_code_scanning_alertcreate_agent_sessioncreate_project_status_updateupdate_projectassign_to_userupdate_releaseadd_reviewerresolve_pull_request_review_threadunassign_from_userDiscussion Categories Used
auditsannouncementsreportsartifactsdevresearchagent-researchdaily-newssecurityauditsis by far the most-used discussion category, used by ~28% of all discussion-creating workflows — establishing it as the primary sink for workflow reports.Structural Characteristics
Job Complexity & Timeout
unbloat-docs,daily-copilot-token-report)Timeout Distribution
15 and 20 minutes dominate, covering the vast majority of job-level timeout configurations. The rare 90-minute and 180-minute timeouts indicate a small number of intensive, long-running workflows.
Concurrency Settings
concurrency:blockcancel-in-progress: trueEvery single workflow uses concurrency groups — a 100% adoption rate. This ensures workflows from the same repo/ref don't overlap unintentionally. The 14 with
cancel-in-progress: trueare typically smoke tests where the latest run supersedes older ones.Permission Patterns
Most Granted Permissions
contents:readissues:writediscussions:writepull-requests:writecontents:writeissues:readpull-requests:readactions:readdiscussions:readsecurity-events:readactions:writesecurity-events:writePermission Distribution
contents:readis near-universal (appearing 633 times across jobs), whileissues:writeis the most common write permission, reflecting the heavy use of issue creation and updates as a primary output channel.MCP Server & Tool Patterns
Most Used MCP Servers
githubsafeoutputsplaywrightserenatavilyagenticworkflowssafeinputsThe
githubandsafeoutputsMCP servers are essentially universal — every workflow integrates GitHub API access and controlled safe outputs.playwrightappears in ~28% of workflows, indicating a substantial number of browser-automation or screenshot-capable agents.serena(likely a semantic/code intelligence server) appears in 15% — used for deeper code analysis tasks.Engine Distribution
dynamicclaudecodexgeminiThe
dynamicengine (67%) dominates, allowing runtime engine selection. Explicitclaudeengine designation covers 24% of workflows, withcodexat 8% and a singlegeminiworkflow.Interesting Findings
Near-perfect safe output coverage: 95.6% of workflows declare
create_discussion,noop,missing_data, andmissing_tooloutputs — these four have become a near-universal baseline, likely enforced by the workflow template system.Scheduling cluster around business hours: Of the 23 weekday-only cron schedules, they cluster tightly around 9 AM–4 PM UTC, suggesting workflows are deliberately scheduled to align with developer working hours for immediate visibility.
100% concurrency adoption: Every single lock file uses a concurrency group. This is a rare case of universal compliance with a best practice across a large workflow fleet.
playwrightis surprisingly common: With 45 workflow jobs using Playwright (~28%), browser automation is not a niche capability but a core tool in the agentic toolkit — far more prevalent than external search (tavilyat only 5).Stable fleet at 158 files: The file count has been exactly 158 for at least 3 consecutive days (Feb 23–25), though total bytes grew by ~300 KB between Feb 23–24 (suggesting prompt/content updates to existing workflows), then stabilized. The fleet appears mature and in maintenance mode rather than active expansion.
Historical Trends
The significant single-day growth on Feb 24 (~296 KB across unchanged file count) indicates a batch update to workflow prompts or embedded content, not new workflow additions.
Recommendations
Formalize the
auditscategory as the canonical sink for automated reports — it already dominates at 28.5% of discussion outputs. Consider creating sub-categories (e.g.,audits/security,audits/performance) as volume grows.Review the 9 read-only workflows: With 94.3% of workflows having write permissions, the 9 read-only workflows may be candidates for enhancement to surface their findings through issues or discussions rather than silently completing.
Investigate the 14
cancel-in-progressworkflows to ensure the other 144 are intentionally not cancelling — for long-running scheduled workflows, allowing concurrent runs may cause duplicate reports or resource waste.Monitor the
dynamicengine cohort: At 67%, the dynamic engine selection is dominant, but the criteria for runtime engine selection aren't visible in lock files. Periodic audits of which engine gets selected dynamically could reveal optimization opportunities.Timeout calibration opportunity: With 176 job instances at 15 min and 173 at 20 min, there may be room to right-size timeouts. Workflows using
playwrightorserenamay warrant higher limits while lightweight report generators could reduce to 10 min.Methodology
/tmp/gh-aw/cache-memory/scripts/full_analysis.py).github/workflows/*.lock.yml/tmp/gh-aw/cache-memory/history/; today's data saved as2026-02-25.jsonReferences:
Beta Was this translation helpful? Give feedback.
All reactions