Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Review
v1.0.0Pre-landing PR review. Analyzes diff against the base branch for SQL safety, LLM trust boundary violations, conditional side effects, and other structural is...
⭐ 0· 75·3 current·3 all-time
by@loocor
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The name/description (pre-landing PR review) align with the instructions (git/gh-based diff analysis, design checks, auto-fixes). However the skill metadata declares no required binaries or credentials even though the runtime instructions rely heavily on git, gh, grep, jq and shell tools. This omission is inconsistent and should be corrected.
Instruction Scope
Instructions read arbitrary repository files (TODOS.md, DESIGN.md, multiple checklists), run networked git fetches (origin/<base>), and explicitly perform AUTO-FIX operations that modify files in-place. While reading the repo is expected for a review tool, the explicit auto-apply behavior (apply fixes directly) means the agent will make changes to the user's working tree without an explicit metadata-level declaration or a required user-confirmation step in the metadata. The SKILL.md also assumes presence of files (.claude/skills/review/checklist.md) and will stop if they are missing, which is brittle.
Install Mechanism
Instruction-only skill with no install spec — lowest risk of arbitrary code download. There is no installer to evaluate.
Credentials
The skill declares no required env vars or credentials, yet it uses the GitHub CLI (gh) and performs git fetch against remotes; those tools typically rely on local auth state (stored tokens/config). The skill should at minimum declare the dependency on gh/git and document that it will use existing CLI authentication. Lack of declared binaries/credentials is an inconsistency that could confuse users about what the skill will access.
Persistence & Privilege
always:false and autonomous invocation allowed (default). The main privacy concern is that the instructions include automatic in-place repo modifications (AUTO-FIX) and applying fixes after asking only for ASK items — this grants write privileges to the agent when invoked and should be made explicit to users and gated by confirmation.
What to consider before installing
This skill is coherent with a PR review purpose but has important mismatches and risky behaviors to consider before installing:
- It uses many command-line tools (git, gh, grep, jq, etc.) but the skill metadata declares no required binaries. Expect failures or silent assumptions if those CLIs are absent — the publisher should list them.
- The instructions perform network operations (git fetch origin/<base>) and use the GitHub CLI; those will use whatever credentials are configured in the user's environment (gh auth, git remotes). The skill doesn't declare or request authorization, so confirm you are comfortable with the tool using your existing CLI auth/session.
- The SKILL.md includes AUTO-FIX steps that modify files directly. Ask the publisher to clarify whether the agent will always write changes automatically, and request a confirm-before-write option. Prefer an explicit prompt that shows edits and requires approval before applying them.
- The skill expects repo-specific checklist files (.claude/skills/review/checklist.md, design-checklist.md). If these don't exist the skill may stop or behave unexpectedly; check the repository for these files or request fallback behavior.
Before installing or enabling this skill, ask the publisher to
1) declare required binaries and their minimum versions, 2) document that it uses existing GH/Git auth and what it will access, and 3) add an explicit confirmation step for any in-place changes (or make auto-fixes opt-in). If you cannot verify those, treat the skill as potentially dangerous for automated write operations.Like a lobster shell, security has layers — review code before you run it.
latestvk9771zphqqq73k2x09c96mx2g983ancq
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
