docs(#3531): add AGENTS.md guidance for Konflux Tekton task substitutions - #3532
docs(#3531): add AGENTS.md guidance for Konflux Tekton task substitutions#3532fullsend-ai-coder[bot] wants to merge 1 commit into
Conversation
…ions Add a new section documenting the distinction between routine Tekton task bundle digest/version bumps and task name substitutions in Konflux/MintMaker PRs. Digest-only bumps are low-risk, but task name changes (e.g., clair-scan to roxctl-scan) require cross-file analysis because Go source, shell scripts, and documentation reference task names as string literals for SLSA provenance filtering. Stale references silently return empty results rather than errors. The guidance directs reviewers to grep for old task names in benchmark/, cmd/, pkg/, hack/, docs/, and pkg/schema/examples/, and to prioritize production code paths over developer utility scripts in hack/. Pre-commit hook (check-commit-message) could not be verified in-sandbox due to network restrictions blocking the remote hook repository fetch. Closes #3531
|
🤖 Finished Review · ✅ Success · Started 6:45 PM UTC · Completed 6:58 PM UTC Commit: Runtime: claude · Model: opus → claude-opus-4-6 · Effort: high · Cost: $3.00 |
Codecov Report✅ All modified and coverable lines are covered by tests.
Flags with carried forward coverage won't be shown. Click here to find out more. 🚀 New features to boost your workflow:
|
|
Risk Assessment: low (1/5) DetailsDocumentation-only PR by a known bot adding 31 lines of guidance to AGENTS.md, with issue scope perfectly matching PR size, no security-sensitive files, and no regression or coupling signals. |
ReviewFindingsMedium
Low
|
| — example JSON and AsciiDoc references to task names. | ||
|
|
||
| When stale references are found, distinguish **production code paths** | ||
| (Go source in `benchmark/`, `cmd/`, `pkg/` — higher priority, should |
There was a problem hiding this comment.
[low] technical accuracy
The guidance states that docs/ and pkg/schema/examples/ contain 'example JSON and AsciiDoc references to task names' in the context of task substitutions like clair-scan to roxctl-scan. However, the actual references in those locations are to get-clair-scan (a different Tekton task that retrieves scan results) and test:get-clair-scan (a policy rule name), not to the clair-scan scan task itself. This imprecision could lead reviewers to conflate the two tasks when evaluating task substitution PRs, potentially causing over-flagging or under-flagging of stale references.
What:
Add a new section to AGENTS.md documenting review guidance for Konflux/MintMaker PRs that substitute Tekton task names (as opposed to routine digest or version bumps).
Why:
PR #3513 substituted clair-scan with roxctl-scan, which left stale string-literal references in benchmark/offliner/scans.go and hack/view-clair-reports.sh. The review agent caught this correctly, but the detection relied on general reasoning rather than explicit repo-specific guidance. Codifying the distinction between digest bumps (low-risk, no cross-file analysis needed) and task substitutions (higher-risk, require grepping for old task name references) makes the behavior more reliable across model versions and helps human reviewers recognize the risk pattern.
The guidance covers:
Tickets:
Related: #3508 (hack/ directory conventions), #3513 (the Konflux PR that motivated this), #3530 (follow-up fix for benchmark offliner)
Closes #3531
Post-script verification
agent/3531-konflux-tekton-task-guidance)af3fe00e4120446b6a808687cc3a5af96ce3d779..HEAD)