[Issue Arborist] Issue Arborist Report — 2026-07-29

[github-actions[bot]]

🌳 Issue Arborist Report

Issues analyzed: 60

Parent issues created

None — all clusters had existing parent issues.

Links created

Parent	Sub-issue	Reasoning
#1535 (AI-credits rate-limit failures)	#1711	[aw] Rust PR Reviewer hit AI credits rate limit
#1535	#1692	[aw] Issue Arborist hit AI credits rate limit
#1535	#1678	[aw] Issue Arborist hit AI credits rate limit
#1535	#1665	[aw] Issue Arborist hit AI credits rate limit
#1535	#1641	[aw] Issue Arborist hit AI credits rate limit
#1549 (non-rate-limit AW execution failures)	#1709	[aw] Bash Step Hygiene Auditor failed
#1549	#1686	[aw] Bash Step Hygiene Auditor failed
#1549	#1680	[aw] Copilot Model List Updater reported incomplete result
#1549	#1661	[aw] Bash Step Hygiene Auditor failed
#1549	#1627	[aw] Bash Step Hygiene Auditor failed
#1283 (upstream dep release notes)	#1305	copilot-cli upstream release action items
#1283	#1304	awf upstream release action items
#1283	#1253	mcpg upstream release action items
#1286 (compiler internals improvements)	#1378	Unify AzureCLI@2 token-mint steps
#1286	#987	Split agentic-pipeline shape out of standalone_ir.rs
#1286	#384	YAML generation uses manual string building in compile_gate_step_external
#1286	#1081	feat(ir): add EnvValue::RuntimeExpression variant
#1649 (agent runtime & execution pipeline)	#1486	Adopt Copilot sidecar (SDK-driver) execution model
#1649	#432	Inject runtime parameter values via ado-script
#1649	#860	Built-in PR context precomputation
#1649	#1375	Reject unwireable cross-job provider-env macros
#1649	#1720	Support allowlisted Copilot CLI session flags
#1321 (MCPG infrastructure & safe-output tooling)	#254	Surface MCPG backend failures in pipeline logs
#1321	#945	Custom MCP container env passthrough missing
#1321	#1621	Port gh-aw's set-issue-type safe output
#1321	#1103	add-pr-comment: render inline threads as suggestions
#1321	#381	fix(safe-outputs): add file staging to upload-workitem-attachment

Total links: 27

Suggested (not linked — for maintainer review)

• [agent-issue]: Parity gap with gh-aw — safe-outputs.threat-detection.prompt is not supported #1701 (threat-detection.prompt not supported) + Proxy Stage 1 Azure DevOps reads without exposing SC_READ_TOKEN #1652 (Proxy Stage 1 ADO reads) + debug: permissions.read token is not exported to Azure CLI in Agent sandbox #1717 (permissions.read token not exported to Azure CLI) + 🔴 Red Team Audit — High: Shell injection via runtimes.lean.toolchain #855 (shell injection via runtimes.lean.toolchain) form a loose security/sandboxing cluster, but at only 4 issues and with heterogeneous sub-themes, the threshold wasn't met. Could warrant a parent if more security hardening issues accumulate.
• chore(workflows): recompile safe-output fixtures with ado-aw v0.38.0 #1156, chore(workflows): recompile safe-output fixtures with ado-aw v0.37.0 #1041, chore(workflows): recompile safe-output fixtures with ado-aw v0.36.0 #1027 (recompile safe-output fixtures for v0.38, v0.37, v0.36) are a natural series but only 3 issues — not enough for a parent cluster.
• feat: align FrontMatter fields with gh-aw schema — run-name, labels, metadata, tracker-id, source, strict #451 (align FrontMatter with gh-aw schema) and feat: decouple prompt body from compiled pipeline output (runtime-import approach) #237 (decouple prompt body, runtime-import approach) could both fit under [Parent] Compiler internals and pipeline generation improvements (IR types, YAML gen, prompt decoupling) #1286 or [Parent] Track agent runtime and execution pipeline improvements #1649; left unlinked due to ambiguity.
• docs: fix AGENTS.md accuracy — add trigger-e2e to ado-script tree and export-fact-catalog to CLI index #1583 (docs AGENTS.md accuracy) and docs(site): refresh IR typed-helper docs to builder-struct API #1192 (refresh IR typed-helper docs) are documentation fixes — no existing docs parent exists to link them to.

Observations

• The backlog has a healthy set of tracking parents that are actively absorbing new failure reports.
• The rate-limit parent ([Parent] Track and resolve recurring AI-credits rate-limit failures in scheduled agentic workflows #1535) continues to accumulate instances — may warrant a root-cause investigation or capacity bump.
• Bash Step Hygiene Auditor is the most frequently failing non-rate-limit workflow (4 failures visible in this batch alone).
• Several orphan issues remain after linking: security hardening, runtime additions (TLA+, config overrides), and a few gh-aw parity features — these may benefit from new parents if the clusters grow.

Generated by Issue Arborist · 61.9 AIC · ⌖ 6.27 AIC · ⊞ 6.1K · ◷

• expires on Aug 5, 2026, 8:43 PM UTC

[github-actions[bot]]

This discussion has been marked as outdated by Issue Arborist.

A newer discussion is available at Discussion #1743.