[copilot-agent-analysis] Daily Copilot Agent Analysis - 2026-04-30

[github-actions[bot]]

🤖 Copilot Agent PR Analysis - 2026-04-30

Analysis Period: Last 24 hours
Total PRs (agent_prs_total): 64 | Merged (agent_prs_merged): 40 (62.5%) | Completed Success Rate: 74.1% | Avg Merge Duration: 60 min

Performance Metrics

Date	PRs	Merged	Closed	Open	Success Rate	Avg Merge Duration
2026-04-30	64	40	14	10	74.1%	60 min
2026-04-29	38	31	7	0	81.6%	124 min
2026-04-28	58	46	12	0	79.3%	81 min

Trend: 📈 Avg merge duration improved significantly (-52% vs yesterday). 📉 Success rate declined slightly (74.1% vs 81.6% yesterday — but 10 PRs still open and may yet merge).

Agent Task Texts

PR #	Status	Task Text (first 100 chars)
#29281	Closed	- [x] Analyze root cause of "session not found" errors on long runs\n- [ ] Add HTTP timeout disablin
#29273	Open	The Anthropic API rejects any input_schema that uses oneOf, allOf, or anyOf at the top level
#29270	Open	## Summary\n\nWhen NO safe outputs are configured explicitly by the user (including when there is no
#29269	Open	## Summary\n\nAdds a new job to the agentic maintenance action that triggers when an issue or pull req
#29264	Open	The [!CAUTION] security alert for threat detection was appended as part of footer generation, bury
#29260	Closed	Users driving agentic workflows from inside a coding agent currently have no native surface — agents
#29259	Open	When Copilot CLI exits mid-stream due to server errors, the on-disk session credential written by th
#29257	Open	Reusable workflow_call workflows couldn't parameterize threat detection at runtime — `threat-detec
#29254	Open	smoke-gemini has been failing with API_KEY_INVALID on every run since Apr 25. The root cause was
#29253	Closed	Bumps four tool versions in pkg/constants/version_constants.go and regenerates all 205 workflow lo
#29242	Closed	workflow_call reusable workflows couldn't parameterize tool policies — bash allowlists, GitHub MCP
#29239	Merged	The lint-go CI job was failing on main due to an unused name string field in the `listFieldCas
#29236	Open	## Summary\n\nAdds support for a new github_ref: true/false constraint on import-schema input typ
#29235	Merged	Safe-output boolean fields (create-issue, hide-older-comments) only accepted literal booleans, b
#29230	Merged	Reusable workflow_call workflows had no way to let callers choose protected-files policy or `pat
#29229	Merged	The add_reviewer tool handler and Go config generation already supported team reviewers end-to-end
#29227	Merged	delight.md was burning 34 inference turns on a fully read-only run (0 write actions, 0.49% output
#29226	Closed	The daily skill optimizer was allowed to run against a dirty working tree, risking noisy or conflict
#29225	Merged	## Summary\n\nAdds a new .github/aw/skills.md reference file that explains how agentic workflows ca
#29223	Open	## Summary\n\nEnable SHA256 checksum validation by default when installing the gh-aw binary.
#29222	Open	AWF's --config (path) support already exists but the compiler never used it, instead generating ev
#29212	Merged	List-valued safe-output fields (labels, allowed-repos, allowed-base-branches) required literal
#29210	Merged	Any workflow with cache-memory: true but without unrestricted bash (bash: *) would fail to read
#29209	Closed	copilot-token-audit.lock.yml (and its source imports) used `curl -fsSL .../refs/heads/main/install
#29205	Merged	On first run or after cache expiration, sergo.md tried to cat non-existent cache files, causing
#29204	Merged	Agent spans were emitting the non-standard gen_ai.provider.name attribute with raw gh-aw engine ID
#29198	Merged	map[T]bool is used as a set throughout the codebase (153+ production files), while `map[T]struct{}
#29197	Merged	~37% of fmt.Errorf calls in pkg/workflow and pkg/parser discarded the original error, breaking
#29195	Merged	GetIntFromEnv accepted a *logger.Logger but only used it for success messages — warnings for inv
#29192	Closed	## Summary\n\nAdds a new shared agentic workflow shared/gh-skill.md that installs GitHub agent skil
#29188	Merged	Some AI agents support @filename syntax to expand file contents inline into tool arguments. The sa
#29182	Merged	Two step naming violations against the project's Title Case imperative-mood convention and symmetric
#29181	Closed	The js-integration-live-api CI job was intermittently failing when the GitHub Actions installation
#29170	Merged	## Summary\n\nAdds a --stdin flag to both gh aw logs and gh aw audit commands that reads workfl
#29169	Merged	The Daily Syntax Error Quality Check consumed 6.07M tokens in a single run — ~14% of the daily Copi
#29168	Merged	The Design Decision Gate was configured with max-turns: 20 but averaged only 7.1 turns — a 5× var
#29167	Closed	When a safe-output field exceeds ~16,000 tokens, the framework silently replaces the value with a `[
#29162	Merged	## Summary\n\nCreates shared/otel.md as a new OpenTelemetry shared import that wraps the existi
#29156	Merged	compiler_validators.go was 477 lines — 59% over the 300-line hard limit in AGENTS.md — mixing mult
#29155	Merged	ContainerPin was defined identically in both pkg/actionpins and pkg/workflow, forcing an expli
#29154	Merged	## Investigation\n\nInvestigated failure at https://github.com/github/gh-aw/actions/runs/25116260447
#29153	Merged	With preserve-branch-name: true, create_pull_request previously aborted (and fell back to openin
#29151	Merged	When a workflow is imported, its bots field was extracted into MergedBots by `import_field_extra
#29141	Merged	Automated CLI consistency inspection surfaced 6 help text issues across 3 commands — jargon, ambigu
#29140	Closed	CI lint-go job failed on main (run #1972) due to a manual for loop in `isValidWorkflowRunConclu
#29137	Closed	Adds detection and automatic expansion of @/absolute/path file references in agent-generated text
#29135	Merged	Sentrux was unable to download language grammar plugins due to releaseassets.githubusercontent.com
#29133	Merged	## Summary\n\nUpdates the Daily Caveman Optimizer workflow to process 5 files per run instead of
#29131	Merged	## Summary\n\nFixes two issues in the release workflow script that caused the agent job to fail (see
#29130	Merged	golangci-lint v2.8.0 introduced the modernize/slicescontains rule, which flagged a manual for-lo
#29128	Merged	The ! in "🍪 **Issue Monster selected this for Copilot!**" triggered bash history expansion (!*
#29127	Merged	## Summary\n\nUpdates the missing_tool tool description to explicitly suggest that agents call this
#29126	Merged	## Summary\n\nUpdates the bash tool configuration in daily-caveman-optimizer.md from a restricted
#29123	Merged	The safe-outputs security harness blocks shell expansion patterns at runtime (${var@op}, ${!var}
#29111	Merged	## Summary\n\nAdds a daily agentic workflow that applies the Julius Brussee principles to a file per
#29101	Merged	## Summary\n\nAdds .github/aw/reuse.md, a new instruction file that explains how to reuse and refac
#29100	Closed	The Visual Regression Checker workflow failed because starlight-links-validator rejected a broken
#29099	Closed	Daily News has been failing consistently since March 25, 2026 (runs 169–194) because AWF v0.25.29's
#29098	Merged	Semantic function-clustering analysis identified 14+ inline []any→[]string type-switch duplicati
#29097	Closed	Patch bump for @playwright/mcp to pick up updated playwright alpha dependency (1.60.0-alpha) and s
#29096	Merged	When an agent explicitly specifies the repository's default base branch (e.g. base: "main") in its
#29095	Merged	schema_suggestions_test.go used raw t.Errorf/t.Fatalf with manual strings.Contains and slice
#29094	Merged	toolsets: "default" was rejected with a bare "got string, want array" — no valid values, no example
#29089	Merged	## Summary\n\nAdds support for `on.workflow_run.conclusion: string

Notable PRs

Issues ⚠️ (Closed without merge)

• #29281: Fix safe outputs returning 'session not found' error — closed without merge
• #29260: Users driving agentic workflows (native agent surface) — closed without merge
• #29253: Tool version bumps — closed without merge
• #29242: workflow_call reusable workflow tool policies — closed without merge
• #29226: Daily skill optimizer dirty tree fix — closed without merge
• #29209: Lock file curl fix — closed without merge
• #29192: New shared agentic workflow gh-skill.md — closed without merge
• #29181: js-integration-live-api CI fix — closed without merge
• #29167: Safe-output field truncation fix — closed without merge
• #29140: CI lint-go for-loop fix — closed without merge
• #29137: @filename path expansion — closed without merge
• #29100: Visual Regression Checker fix — closed without merge
• #29099: Daily News AWF version fix — closed without merge
• #29097: playwright/mcp patch bump — closed without merge

Open PRs ⏳

• #29273: Fix: Remove top-level anyOf from add_reviewer MCP tool schema
• #29270: feat: auto-inject create-issue safe output
• #29269: feat: add disable_agentic_workflow label-triggered job
• #29264: [!CAUTION] security alert visibility fix
• #29259: Copilot CLI session credential fix
• #29257: threat-detection parameterization for workflow_call
• #29254: smoke-gemini API_KEY_INVALID fix
• #29236: github_ref constraint on import-schema
• #29223: SHA256 checksum validation for gh-aw install
• #29222: AWF --config path compiler support

Key Insights

• 📉 Success rate dipped to 74.1% (vs 81.6% yesterday) partly due to 10 still-open PRs; if all open PRs merge, effective rate would rise to ~93.8%.
• 📈 Merge turnaround improved dramatically — avg 60 min today vs 124 min yesterday, suggesting faster review cycles.

---

Generated by Copilot Agent Analysis (Run: §25163831745)

Generated by Copilot Agent PR Analysis · ● 136.4K · ◷

• expires on May 1, 2026, 11:58 AM UTC

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-05-01T11:58:13.115Z.

Closed by Workflow