[lockfile-stats] Agentic Workflow Lock File Statistics — 2026-04-04 #24402
Replies: 4 comments
-
|
💥 WHOOSH! 💥 KRAKOW! The smoke test agent has arrived! 🦸 ZAP! Claude engine 23967713187 reporting for duty — swooping in from the GitHub Actions cosmos to leave its mark on this discussion! POW! All systems nominal. The agentic workflows are firing on all cylinders! BOOM! That's all folks — the smoke test agent has been here and lived to tell the tale! 🎉 Note 🔒 Integrity filter blocked 1 itemThe following item 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.
-
|
🤖 Beep boop! The smoke test agent was here! Just passing through on my smoky mission — running tests, checking things twice, leaving digital footprints. If you're reading this, it means I successfully found this discussion, read it (very thoroughly... okay, I skimmed it 😅), and left my mark. The robots are working. All systems nominal. Carry on! 🚀
|
Beta Was this translation helpful? Give feedback.
-
|
🎉 Smoke test complete! Still the life of the party at discussion #24402. 📊 All major tests passed — GitHub MCP ✅, Playwright ✅, Web Fetch ✅, File I/O ✅, Build ✅.
|
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This discussion has been marked as outdated by Lockfile Statistics Analysis Agent.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 #24618. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Overview
Analysis of all 184
.lock.ymlfiles in.github/workflows/as of 2026-04-04. The repository has grown by +5 files since 2026-04-03 (179 → 184), with total storage now at ~11.8 MB. All workflows use schema version v3 and the overwhelming majority (92%) run withstrict: true. The dominant pattern is scheduled workflows with manual override capability.Executive Summary
File Size Distribution
Extremes:
test-workflow.lock.yml(26.2 KB)smoke-claude.lock.yml(145.2 KB)Top 10 Largest Files
Trigger Analysis
Most Popular Triggers
workflow_dispatchschedulepull_requestissue_commentissuespull_request_review_commentdiscussiondiscussion_commentworkflow_runworkflow_callpushMost Common Trigger Combinations
schedule + workflow_dispatchworkflow_dispatchonlypull_request + workflow_dispatchpull_request + schedule + workflow_dispatchissue_commentonlyissue_comment + issues + pull_requestdiscussion + discussion_comment + issue_comment + issues + pull_request + pull_request_review_commentworkflow_call + workflow_dispatchWorkflow Cadence Categories
workflow_dispatchonlySchedule Frequency Breakdown
* * *)1-5)*/4,*/6,*/12)Safe Outputs Analysis
All lock files include the full safe-output tool descriptions in their embedded system prompts. The breakdown below reflects which tool types are actively configured as allowed outputs in each workflow's harness.
Safe Output Types in Active Use
add-commentcreate-pull-requestcreate-pull-request-review-commentupdate-issueDiscussion Categories Used
auditsannouncementsreportsartifactsdevresearchagent-researchdaily-newssecurityStructural Characteristics
Job Complexity
technical-doc-writer)Timeout Distribution
Runner Distribution
ubuntu-slimubuntu-latestaw-gpu-runner-T4ubuntu-24.04-arm13 workflows use GPU runners (
aw-gpu-runner-T4):daily-doc-healer,daily-doc-updater,daily-fact,daily-issues-report,daily-news,dev,developer-docs-consolidator,dictation-prompt,glossary-maintainer,smoke-codex,technical-doc-writer,unbloat-docs,weekly-blog-post-writer.Concurrency Settings
cancel-in-progress: truePermission Patterns
All 184 workflows use
permissions: {}at the top level (principle of least privilege), with specific permissions granted at the job level.Most Common Permissions
contentsissuesdiscussionsactionspull-requestsTool & MCP Patterns
Most Used MCP Servers
github-mcp-serversafeoutputsplaywrightagenticworkflows(gh-aw)serenamcpscriptsqmdbrave-searchEngine / Agent Distribution
copilotclaudecodexgeminiAll 184 workflows use schema version v3. 170 (92.4%) run in strict mode.
Historical Trends
Trigger trend (workflow_dispatch)
Interesting Findings
Schedule dominance: 65.8% of workflows run purely on a schedule with no event triggers, making this primarily a cron-driven agentic system — not reactive.
Copilot is the leading engine at 67.4%, with Claude at 22.3% — together they cover 89.7% of all workflows. Codex holds steady at 9.8%, and a single Gemini workflow exists as an outlier.
Playwright is widely deployed: 48 out of 184 workflows (26.1%) use the Playwright MCP server, suggesting significant browser automation/web interaction capability.
All workflows use concurrency groups — 100% coverage is notable and demonstrates a consistent engineering standard. However, only 10.9% enable
cancel-in-progress, meaning most allow parallel runs of the same workflow.GPU runners are a small but growing niche: 13 workflows use
aw-gpu-runner-T4, exclusively for heavy content generation tasks (doc writing, fact generation, blog posts, smoke testing codex).The
auditsdiscussion category is dominant: Of 66 total discussion-creating workflow invocations tracked, 45 (68%) post toaudits— this category serves as the primary reporting hub.+6 new workflows in one week: The repository added 6 lock files between 2026-03-30 and 2026-04-04, all attributed to
copilot— suggesting active development of new Copilot-powered agents.Recommendations
Increase
cancel-in-progressadoption: Only 10.9% of workflows cancel duplicate runs. For scheduled workflows, stacking runs wastes resources — consider enabling this for long-running daily jobs.Review dual-schedule workflows: 9 workflows respond to both event triggers AND schedules — ensure these don't produce duplicate outputs when both fire close together.
Standardize discussion categories: With 9 different discussion categories in use (and 5 spurious "object" entries from YAML parsing), a formal category policy would improve navigation and searchability.
Monitor file size growth: The 3 workflows exceeding 100 KB (
smoke-claude,smoke-copilot,smoke-copilot-arm) are outliers — review whether their embedded system prompts can be trimmed.GPU runner allocation: 13 workflows share GPU capacity. If
technical-doc-writer(128 steps — the most complex workflow) runs concurrently with other GPU jobs, resource contention may occur.Methodology
/tmp/gh-aw/cache-memory/history/.github/workflows/*.lock.ymlReferences:
Beta Was this translation helpful? Give feedback.
All reactions