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
Updates the task-specific kanon-workers prompt with two focused conventions backed by recurring review findings from Kanon implementation PRs.
First, the worker now tests every affected command, flag, scope, and guard outcome directly instead of relying on another path as indirect coverage:
Kanon PR Add axon CLI for managing Tasks #9 review found that the new update command had only an early-return test and no pull-to-apply happy-path coverage.
Kanon PR Add cron source to TaskSpawner #40 review found that testing apply -n did not cover the distinct update -n pull-plus-no-write contract.
Second, the worker now treats external repository inputs as one trust boundary across subprocess arguments, filesystem traversal, error output, and persistent caches:
Kanon PR axon install/uninstall feature #29 reviews repeatedly found external-source boundary gaps: symlink traversal could read files outside the repository, expanded URL credentials could leak through git errors, and root clone metadata could persist credentials in the source cache.
The rules remain in the implementation worker prompt because they address mistakes made while writing Kanon changes. Shared skills remain only in self-development/base-agent.yaml.
The audit did not promote Kanon PR #68's single README finding, Kanon PR #22's feature-specific schema findings, or Kanon PR #45's line-ending edge case into general rules. Other findings were already covered by Kanon's current repository instructions, including co-owned configuration preservation and resource-naming errors.
Which issue(s) this PR is related to:
N/A
Special notes for your reviewer:
There were no Kanon PRs updated or merged during the last-seven-days window of August 17-24, 2026. The newest PR activity in the requested recent-20 sets was July 25, 2026, and the newest merge was July 2, 2026. The audit therefore inspected the diffs, reviews, inline comments, and conversations for the review-bearing and generated-by-kelos PRs in those sets, while retaining the existing recurring evidence from PRs #9 and #16.
Only self-development/kanon/kanon-workers.yaml is changed. make verify passes.
kelos-botBot
changed the title
Make Kanon worker follow the PR template and label gates
Give the Kanon worker standing PR-template guidance for the label gate
Jun 1, 2026
Good question — I dug into why self-development works without this and the answer points to a real gap rather than a redundancy:
kelos-workers complies with the PR template not because of its prompt, but because the kelos repo has a CLAUDE.md ("Always follow .github/PULL_REQUEST_TEMPLATE.md…"). That standing guidance is what clears the label gate. The kanon repo has no CLAUDE.md (verified via the GitHub API and a code search), so kanon-workers got no equivalent guidance from anywhere — and after kanon PR #8 added check-pr-labels, its PRs would trip the gate.
I've reworked the change to match how self-development does it: the convention now lives in the kanon-workers-agentagentsMD (the agent's standing instructions, the kanon analog of kelos's CLAUDE.md), and the numbered procedure steps are reverted to be identical to kelos-workers. Net diff is now just a 10-line ## Pull Requests section.
If you'd prefer this live in a CLAUDE.md added to the kelos-dev/kanon repo itself (which would also cover the other kanon agents and human contributors), I'm happy to open that as a follow-up against kanon instead — let me know.
kelos-botBot
changed the title
Give the Kanon worker standing PR-template guidance for the label gate
Give the Kanon worker PR-template and testing conventions
Jun 1, 2026
kelos-botBot
changed the title
Give the Kanon worker PR-template and testing conventions
Refine Kanon worker test and documentation guidance
Jul 14, 2026
kelos-botBot
changed the title
Refine Kanon worker test coverage guidance
Refine Kanon worker implementation guidance
Aug 23, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What type of PR is this?
/kind cleanup
What this PR does / why we need it:
Updates the task-specific
kanon-workersprompt with two focused conventions backed by recurring review findings from Kanon implementation PRs.First, the worker now tests every affected command, flag, scope, and guard outcome directly instead of relying on another path as indirect coverage:
updatecommand had only an early-return test and no pull-to-apply happy-path coverage.apply -ndid not cover the distinctupdate -npull-plus-no-write contract.Second, the worker now treats external repository inputs as one trust boundary across subprocess arguments, filesystem traversal, error output, and persistent caches:
Motivating reviews:
The rules remain in the implementation worker prompt because they address mistakes made while writing Kanon changes. Shared skills remain only in
self-development/base-agent.yaml.The audit did not promote Kanon PR #68's single README finding, Kanon PR #22's feature-specific schema findings, or Kanon PR #45's line-ending edge case into general rules. Other findings were already covered by Kanon's current repository instructions, including co-owned configuration preservation and resource-naming errors.
Which issue(s) this PR is related to:
N/A
Special notes for your reviewer:
There were no Kanon PRs updated or merged during the last-seven-days window of August 17-24, 2026. The newest PR activity in the requested recent-20 sets was July 25, 2026, and the newest merge was July 2, 2026. The audit therefore inspected the diffs, reviews, inline comments, and conversations for the review-bearing and
generated-by-kelosPRs in those sets, while retaining the existing recurring evidence from PRs #9 and #16.Only
self-development/kanon/kanon-workers.yamlis changed.make verifypasses.Does this PR introduce a user-facing change?