Feat: Add a link back to the skill when opening a link issue (breadcrumb)

[esnible]

Summary

I'd like the issue opened by the link checker to include a 'breadcrumb' back to the link checker itself. This will help the developer find the skill that generated the report. The developer can then decide if the skill itself could be improved, in addition to fixing the reported problem.

I noticed there were only 2 forks of this repo, suggesting that Rossoctl developers are unaware of the potential of contributing automation to increase our work velocity.

[clawgenti]

Docs-only change that adds a breadcrumb link back to the scanner skill and a note about auto-close behavior — clear and useful context for issue recipients. All checks pass. Ready for human review.

---

Reviewed by clawgenti using github:pr-review

[rubambiza]

Suggested change to reconsider the assertion that the automation does not create PRs with fixes.

[rubambiza]

@esnible I agree that the automation auto-closes the issues. However, at least for internal links, the automation (i.e., link-health-fixer) does create commits

automation/scripts/link-health-fixer.sh

Line 477 in 094bc0d

# Apply each fix

and push PRs

automation/scripts/link-health-fixer.sh

Line 508 in 094bc0d

# Create PR

with fixes.

[esnible]

Converted to draft while I add references to https://github.com/rossoctl/automation/blob/main/scripts/link-health-fixer.sh

[rubambiza]

@esnible If you are referring to particular lines in link-health-fixer.sh, let me know because I am working toward a refactor that makes all of the scripts more portable beyond the rossoctl org. The spec for it can be found at https://github.com/rossoctl/automation/blob/main/docs/specs/2026-07-29-org-portability-design.md

[rubambiza]

Nice breadcrumb idea — we just adopted the same "link back to the skill" pattern for the pr-review program's review footer. One portability caveat worth surfacing here:

The (cron: link-health-scanner) part of the breadcrumb is only accurate when the skill runs under our OpenClaw setup (clawgenti on the automation host). Once a skill is importable via /plugin and run in someone else's environment, there is no link-health-scanner cron and no standing-orders/... to point at — so a hardcoded cron/standing-order breadcrumb becomes wrong for every external importer.

For pr-review (whose skill now lives in rossoctl/agent-skills, importable) we resolved this by making the footer skill-link only — it names the one thing true in any environment:

*Reviewed by clawgenti using the [github-pr-review skill](https://github.com/rossoctl/agent-skills/blob/main/skills/github-pr-review/SKILL.md)*

The OpenClaw-specific provenance (cron job, schedule, scope) stays documented in the automation standing order (standing-orders/pr-review.md), where it's accurate and version-controlled, rather than baked into the portable skill's output.

link-health's skill still lives in automation today, so the cron breadcrumb is fine as-is here. But if/when it migrates to agent-skills (like pr-review did), the same skill-link-only (or env-aware) treatment would keep the breadcrumb correct for importers.