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
agent-shield.yml is a thin caller stub for the org reusable workflow. Investigation of the run history found two distinct contributors to the flagged 90.9% failure rate:
Genuine startup failures (conclusion failure, "workflow file issue") β caused by the stub referencing the reusable via a floating tag@v1. When that tag is missing/moved, the run fails at startup. The org source-of-truth stub (petry-projects/.github/standards/workflows/agent-shield.yml) pins the reusable to an immutable commit SHA: @208ec2d69b75227d375edf8745d84fbac05a76b2 # v1. The repo has drifted from this canonical form.
action_required runs (0s, on bot-pushed branches) β these are produced by GitHub's repo-level workflow-approval gate for non-collaborator actors. This is outside the scope of this workflow file (it's a repo-settings concern) and cannot be fixed by editing the stub.
Scope
Sync agent-shield.yml to the org source-of-truth by pinning the reusable-workflow uses: ref to the canonical immutable SHA (# v1), eliminating tag-resolution startup failures. This is the same "sync org-standard stub" pattern as the previous commit.
Implementation Checklist
.github/workflows/agent-shield.yml β change uses: from @v1 to @208ec2d69b75227d375edf8745d84fbac05a76b2 # v1 to match the source-of-truth (no other lines change; triggers and job name preserved)
Tests to Write
Per AGENTS.md Phase-2 exception: this is a behavior-preserving configuration change (the v1 tag currently resolves to exactly this SHA), so no automated test is added. Verification is via YAML-validity + lint + diff review. There is no existing workflow-lint/test harness in the repo.
Test Command
bash .dev-lead/scripts/dev-lead-lint.sh (+ npm run lint and a YAML parse check)
Out of Scope / Follow-up
The residual action_required runs stem from a repo-level workflow-approval setting for bot/outside-collaborator branches, not from this file. Recommend a separate follow-up to adjust repo Actions approval settings (e.g. via apply-repo-settings).
Root-caused the 90.9% failure rate: genuine failure runs were "workflow file issue" startup failures from referencing the org reusable via a floating tag@v1; the org source-of-truth pins an immutable SHA.
Synced .github/workflows/agent-shield.ymluses: ref to the canonical pinned SHA @208ec2d69b75227d375edf8745d84fbac05a76b2 # v1 (same reusable, same v1).
Verified the file is now byte-identical to petry-projects/.github/standards/workflows/agent-shield.yml.
Triggers (push/pull_request on main), uses: target, and job name agent-shield (the required status check) all preserved.
Test Results
Per AGENTS.md Phase-2 exception, this is a behavior-preserving configuration change (the v1 tag currently resolves to exactly this SHA), so no automated test was added; there is no workflow-lint/test harness in the repo. Verification output:
npm run lint/prettier were not run because node_modules is absent in this environment and a full npm install is disproportionate for a one-line workflow-ref change; the changed file is byte-identical to the org-validated canonical stub, so it conforms to org formatting/lint by construction.
Files Changed
.github/workflows/agent-shield.yml: pinned the reusable-workflow uses: ref from floating @v1 to the immutable SHA @208ec2d6β¦# v1, matching the org source-of-truth. Eliminates tag-resolution startup failures and SHA-pins this security workflow (best practice).
Notes
Residual action_required runs are out of scope for this file. A portion of the flagged failures are action_required (0s) runs on bot-pushed branches (including non-dependabot agent branches). These come from GitHub's repo-level workflow-approval gate for outside-collaborator/bot actors, not from the workflow content β they cannot be fixed by editing this stub.
Recommended follow-up: open a separate issue to adjust the repo's Actions approval settings (e.g. via apply-repo-settings) so trusted bot/agent branches don't sit in action_required. That is the lever that will move the remaining failure-rate contribution.
β Auto-resolved by Fleet Monitor. This workflow's failure rate has been below the 10% threshold (or the workflow was removed) since 2026-06-25 (20 days) β the monitor stopped flagging it. Closing. It will reopen automatically if the workflow crosses the threshold again.
Activity
don-petry commentedon Jun 18, 2026
Dev-Lead Implementation Plan
Issue: #288 β [Fleet Monitor] petry-projects/TalkTerm β .github/workflows/agent-shield.yml
Root-Cause Analysis
agent-shield.ymlis a thin caller stub for the org reusable workflow. Investigation of the run history found two distinct contributors to the flagged 90.9% failure rate:failure, "workflow file issue") β caused by the stub referencing the reusable via a floating tag@v1. When that tag is missing/moved, the run fails at startup. The org source-of-truth stub (petry-projects/.github/standards/workflows/agent-shield.yml) pins the reusable to an immutable commit SHA:@208ec2d69b75227d375edf8745d84fbac05a76b2 # v1. The repo has drifted from this canonical form.action_requiredruns (0s, on bot-pushed branches) β these are produced by GitHub's repo-level workflow-approval gate for non-collaborator actors. This is outside the scope of this workflow file (it's a repo-settings concern) and cannot be fixed by editing the stub.Scope
Sync
agent-shield.ymlto the org source-of-truth by pinning the reusable-workflowuses:ref to the canonical immutable SHA (# v1), eliminating tag-resolution startup failures. This is the same "sync org-standard stub" pattern as the previous commit.Implementation Checklist
.github/workflows/agent-shield.ymlβ changeuses:from@v1to@208ec2d69b75227d375edf8745d84fbac05a76b2 # v1to match the source-of-truth (no other lines change; triggers and job name preserved)Tests to Write
v1tag currently resolves to exactly this SHA), so no automated test is added. Verification is via YAML-validity + lint + diff review. There is no existing workflow-lint/test harness in the repo.Test Command
bash .dev-lead/scripts/dev-lead-lint.sh(+npm run lintand a YAML parse check)Out of Scope / Follow-up
The residual
action_requiredruns stem from a repo-level workflow-approval setting for bot/outside-collaborator branches, not from this file. Recommend a separate follow-up to adjust repo Actions approval settings (e.g. viaapply-repo-settings).don-petry commentedon Jun 18, 2026
Dev-Lead: Implementation Complete
Plan Execution
failureruns were "workflow file issue" startup failures from referencing the org reusable via a floating tag@v1; the org source-of-truth pins an immutable SHA..github/workflows/agent-shield.ymluses:ref to the canonical pinned SHA@208ec2d69b75227d375edf8745d84fbac05a76b2 # v1(same reusable, samev1).petry-projects/.github/standards/workflows/agent-shield.yml.push/pull_requestonmain),uses:target, and job nameagent-shield(the required status check) all preserved.Test Results
Per AGENTS.md Phase-2 exception, this is a behavior-preserving configuration change (the
v1tag currently resolves to exactly this SHA), so no automated test was added; there is no workflow-lint/test harness in the repo. Verification output:npm run lint/prettierwere not run becausenode_modulesis absent in this environment and a fullnpm installis disproportionate for a one-line workflow-ref change; the changed file is byte-identical to the org-validated canonical stub, so it conforms to org formatting/lint by construction.Files Changed
.github/workflows/agent-shield.yml: pinned the reusable-workflowuses:ref from floating@v1to the immutable SHA@208ec2d6β¦# v1, matching the org source-of-truth. Eliminates tag-resolution startup failures and SHA-pins this security workflow (best practice).Notes
action_requiredruns are out of scope for this file. A portion of the flagged failures areaction_required(0s) runs on bot-pushed branches (including non-dependabot agent branches). These come from GitHub's repo-level workflow-approval gate for outside-collaborator/bot actors, not from the workflow content β they cannot be fixed by editing this stub.apply-repo-settings) so trusted bot/agent branches don't sit inaction_required. That is the lever that will move the remaining failure-rate contribution.feat: implement issue #288 β [Fleet Monitor] petry-projects/TalkTerm β¦
feat: implement issue #288 β [Fleet Monitor] petry-projects/TalkTerm β¦
feat: implement issue #288 β [Fleet Monitor] petry-projects/TalkTerm β¦
don-petry commentedon Jul 15, 2026
β Auto-resolved by Fleet Monitor. This workflow's failure rate has been below the 10% threshold (or the workflow was removed) since 2026-06-25 (20 days) β the monitor stopped flagging it. Closing. It will reopen automatically if the workflow crosses the threshold again.