You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Top combination: schedule+workflow_dispatch (208 workflows), then workflow_dispatch alone (38), then pull_request+schedule+workflow_dispatch (29)
Most common cron: 0 0 */2 * * (every 2 days) — 42 workflows
Safe outputs analysis
create_discussion_workflows: 92, all 92 resolved to a category (discussion_category_detected = 92, 0 unresolved) — self-check passed
Discussion categories: audits (79), announcements (5), artifacts (2), dev (2), research (2), general (1), daily-news (1)
Most common non-baseline safe outputs: create_issue (146, +1 vs prior day), create_discussion (92), add_comment (77), create_pull_request (64)
safe_outputs_config_missing: 0 (config extracted from every lockfile)
Permission patterns (agent job)
Scope
read
write
none
contents
298
0
0
issues
260
0
38
pull-requests
255
0
43
actions
113
0
185
discussions
52
0
246
permissions_unknown: 0 — all 298 agent jobs parsed. Union-of-all-jobs shows 298/298 workflows grant some write scope (mostly via the safe-output post-processing jobs), vs. the read-mostly agent job itself — expected, since safe-output jobs perform the actual writes.
Engine distribution
copilot (120), codex (76), claude (56), pi (29), crush/aider/opencode/goose (3 each), cursor/deepseek-harness/gemini/kiro/pydantic-ai (1 each). engine_unknown: 0.
Tool & MCP patterns
safeoutputs MCP present in all 298 workflows; github MCP in 120; agenticworkflows in 44; serena in 25
mcp_fallback_used_count: 0 (manifest-based detection worked for all lockfiles)
Interesting findings
Near-zero week-over-week churn: comparing to 2026-09-02, only total_bytes (+0.37%), create_issue count (+1), and averages shifted marginally — the fleet of 298 lockfiles is stable, no new/removed workflows in the last day.
Audits dominate discussion output: 79 of 92 create_discussion workflows (86%) post to the audits category — this very report included.
copilot is the leading engine (120/298, 40%) ahead of codex (76) and claude (56), despite claude models otherwise appearing across model strings (claude-sonnet-5, claude-opus-4.8, etc.) — engine choice and model string diverge for copilot-branded agents.
Permission hygiene looks solid: 0 agent jobs grant write on contents, issues, discussions, etc. at the primary agent-job level — all direct write actions are deferred to dedicated safe-output jobs (100% union write coverage vs. 0% agent-job write).
Cron clustering: 42 workflows (14%) share the exact same 0 0 */2 * * schedule, suggesting a common default/template cadence.
Recommendations
No corrective action needed — self-checks passed (yaml parse OK, safe-output config found for all lockfiles, discussion categories fully resolved).
Given 42 workflows share an identical cron slot, consider jittering schedules to smooth GitHub Actions runner load at that instant.
Continue daily snapshots; the low day-over-day variance makes this a good baseline for detecting anomalies (e.g. a sudden permission-scope change or new engine appearing).
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Analysis of 298 compiled
.github/workflows/*.lock.ymlfiles (0 malformed/skipped). Methodology: single-script compact JSON analysis (cached analyzer, schema v4).Key metrics
Trigger analysis
workflow_dispatch: 290,schedule: 241,pull_request: 43 (dominant triggers)schedule+workflow_dispatch(208 workflows), thenworkflow_dispatchalone (38), thenpull_request+schedule+workflow_dispatch(29)0 0 */2 * *(every 2 days) — 42 workflowsSafe outputs analysis
create_discussion_workflows: 92, all 92 resolved to a category (discussion_category_detected= 92, 0 unresolved) — self-check passedcreate_issue(146, +1 vs prior day),create_discussion(92),add_comment(77),create_pull_request(64)safe_outputs_config_missing: 0 (config extracted from every lockfile)Permission patterns (agent job)
permissions_unknown: 0 — all 298 agent jobs parsed. Union-of-all-jobs shows 298/298 workflows grant somewritescope (mostly via the safe-output post-processing jobs), vs. the read-mostly agent job itself — expected, since safe-output jobs perform the actual writes.Engine distribution
copilot (120), codex (76), claude (56), pi (29), crush/aider/opencode/goose (3 each), cursor/deepseek-harness/gemini/kiro/pydantic-ai (1 each).
engine_unknown: 0.Tool & MCP patterns
safeoutputsMCP present in all 298 workflows;githubMCP in 120;agenticworkflowsin 44;serenain 25github:get_commit/get_file_contents/etc. (118 workflows each)mcp_fallback_used_count: 0 (manifest-based detection worked for all lockfiles)Interesting findings
total_bytes(+0.37%),create_issuecount (+1), and averages shifted marginally — the fleet of 298 lockfiles is stable, no new/removed workflows in the last day.auditscategory — this very report included.claudemodels otherwise appearing across model strings (claude-sonnet-5,claude-opus-4.8, etc.) — engine choice and model string diverge for copilot-branded agents.writeoncontents,issues,discussions, etc. at the primary agent-job level — all direct write actions are deferred to dedicated safe-output jobs (100% union write coverage vs. 0% agent-job write).0 0 */2 * *schedule, suggesting a common default/template cadence.Recommendations
Methodology note: single-script compact JSON analysis.
Warning
Firewall blocked 1 domain
The following domain was blocked by the firewall during workflow execution:
api.anthropic.comTo allow these domains, add them to the
network.allowedlist in your workflow frontmatter:See Network Configuration for more information.
All reactions