chore(deps): update mattpocock/skills digest to c55ee46

[renovate]

This PR contains the following updates:

Package	Update	Change
mattpocock/skills	digest	84fdeff → c55ee46

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • Between 12:00 AM and 03:59 AM, only on Monday (* 0-3 * * 1)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[toolhive-release-app]

🛡️ Skill Security Scan Results

✅ ask-matt

• Status: Passed
• Findings: 0

✅ codebase-design

• Status: Passed
• Findings: 0

✅ diagnosing-bugs

• Status: Passed
• Findings: 1

✅ domain-modeling

• Status: Passed
• Findings: 0

✅ git-guardrails-claude-code

• Status: Passed
• Findings: 2

✅ grill-me

• Status: Passed
• Findings: 1
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: mattpocock/skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

✅ grill-with-docs

• Status: Passed
• Findings: 0

✅ grilling

• Status: Passed
• Findings: 1

✅ handoff

• Status: Passed
• Findings: 1

✅ implement

• Status: Passed
• Findings: 0

✅ migrate-to-shoehorn

• Status: Passed
• Findings: 1

✅ prototype

• Status: Passed
• Findings: 0

✅ research

• Status: Passed
• Findings: 0

✅ resolving-merge-conflicts

• Status: Passed
• Findings: 1

✅ scaffold-exercises

• Status: Passed
• Findings: 2
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: mattpocock/skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

✅ setup-matt-pocock-skills

• Status: Passed
• Findings: 1

✅ setup-pre-commit

• Status: Passed
• Findings: 3
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: mattpocock/skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

✅ tdd

• Status: Passed
• Findings: 0

✅ teach

• Status: Passed
• Findings: 1

✅ to-questionnaire

• Status: Passed
• Findings: 2
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: mattpocock/skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

✅ to-spec

• Status: Passed
• Findings: 3
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: mattpocock/skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

✅ to-tickets

• Status: Passed
• Findings: 2
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: mattpocock/skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

✅ wait-what

• Status: Passed
• Findings: 1
• Allowed (not blocking): 1
  • MANIFEST_MISSING_LICENSE (Allowed: mattpocock/skills is licensed MIT at the repository root; upstream does not embed an SPDX license identifier in per-skill SKILL.md frontmatter.)

✅ wayfinder

• Status: Passed
• Findings: 2

✅ wizard

• Status: Passed
• Findings: 0

✅ writing-for-agents

• Status: Passed
• Findings: 0

---

Summary: Completed 26 of 26 skill scan(s), all passed security checks. ✅

[danbarr]

Pushed follow-up commits allowlisting the `skill-security-scan` findings on `skills/wizard/spec.yaml` and `skills/triage/spec.yaml`:

• wizard: 3 findings (`.env` mention, `bash -n` syntax-check flag, `chmod +x`) — all describe the wizard's own documented, human-run setup procedure, not a credential leak or privilege escalation.
• triage: 1 finding (`ATR_2026_00051`, "all files") — matches OUT-OF-SCOPE.md's instruction to read all files in the project-local `.out-of-scope/` directory, a scoped and bounded read.

Note: I did not touch `skills/improve-codebase-architecture/spec.yaml` in this PR. Its scan failure looked like the same kind of keyword noise at first glance, but the actual blocking findings are two real `LLM_PROMPT_INJECTION` (HIGH) issues: the generated HTML report configures Mermaid with `securityLevel: 'loose'` (disabling its XSS protections, and the report content is derived from codebase analysis that could include attacker-influenced names/comments), and the skill instructs the agent to read and act on externally-controlled files (`CONTEXT.md`, ADR files, git history) before analysis. Those are worth a human security call rather than a blind allowlist — flagging separately rather than merging as part of this bump.

[danbarr]

Pushed a follow-up commit allowlisting two of the four blocking skill-security-scan findings on skills/improve-codebase-architecture/spec.yaml:

• Two LLM_PROMPT_INJECTION findings (reading CONTEXT.md/ADR files, and using git log --oneline for hot-spot detection) — same class of accepted risk as skills/hf-mcp and skills/huggingface-tool-builder: the skill's documented purpose requires processing project-local content the user already trusts.

Needs a maintainer decision, not touched here:

• LLM_COMMAND_INJECTION and LLM_DATA_EXFILTRATION — both anchored in HTML-REPORT.md's Mermaid securityLevel: 'loose' configuration. This is a real, unaddressed risk: the generated HTML report embeds codebase-derived content (file/module names) into diagram labels, and loose mode disables Mermaid's HTML/script sanitization, so a crafted file or module name could execute script when the report is opened. Checked upstream (mattpocock/skills) — no existing issue covers this. I haven't filed one yet; let me know if you want that opened, or if you'd rather accept the risk and allowlist it (the practical exploit path requires the attacker to already have write access to the analyzed codebase).

[renovate]

Edited/Blocked Notification

Renovate will not automatically rebase this PR, because it does not recognize the last commit author and assumes somebody else may have edited the PR.

You can manually request rebase by checking the rebase/retry box above.

⚠️ Warning: custom changes will be lost.

[danbarr]

Security-scan triage (trusted upstream publisher does not mitigate these execution/instruction risks):

• No new exclusions added. PG_EXFIL_HTML_TAG remains the existing false-positive exclusion for the local static report scaffold.
• Blocking — LLM_COMMAND_INJECTION (improve-codebase-architecture): Mermaid securityLevel: "loose" disables sanitization for a report populated with repository-controlled content and opened by the workflow; this is an unmitigated XSS path.
• Blocking — LLM_COMMAND_INJECTION (matt-pocock-code-review): the user-supplied fixed point is interpolated into Git shell commands without validation or safe argument handling.
• Blocking — LLM_PROMPT_INJECTION (matt-pocock-code-review): untrusted issue-tracker content is passed to a sub-agent as specification without a data-only boundary.
• Blocking — LLM_PROMPT_INJECTION (triage): untrusted issue/PR discussions and reproduction steps are followed without prompt-injection handling or dangerous-command confirmation.
• Blocking — LLM_UNAUTHORIZED_TOOL_USE (triage): external PR checkout and test/command execution lack sandboxing or immediate maintainer authorization.

No files changed or commit pushed; the five findings remain correctly blocking pending upstream mitigations or an explicit maintainer risk acceptance.

[danbarr]

Follow-up: I trialed deferring only the three originally flagged updates (improve-codebase-architecture, matt-pocock-code-review, and triage) while retaining the rest of this batch. The remaining update set still has unallowlisted security findings in ask-matt (including critical ATR_2026_00063), so I did not push the partial revert. No unrelated exclusions were added.

[danbarr]

Correction to the prior follow-up: the local scanner output was not the authoritative PR result. The three flagged updates (improve-codebase-architecture, matt-pocock-code-review, and triage) have now been deferred in 92e30f6; all other Matt Pocock updates remain. Final status should be determined by the GitHub Trusted Skill Scan.