Skillv1.0.0

ClawScan security

Create Pr · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignMar 9, 2026, 8:13 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: Instructions, required capabilities, and requested access align with a Gitee PR-creation helper — nothing disproportionate or unrelated is requested.
Guidance: This skill appears coherent for creating Gitee pull requests through a configured MCP server. Before installing/using it: ensure your platform's Gitee MCP credentials are correct and scoped appropriately (the skill will create PRs and can append 'closes #N' to close issues), review any generated PR title/body before the skill submits it, and avoid granting broader repo permissions than necessary. If you do not want the agent to create PRs autonomously, keep the skill user-invocable only and require explicit confirmation from you before calling create_pull.

Purpose & Capability: okThe skill's name and description match its instructions: it builds a PR title/body from a repo diff and calls Gitee MCP operations (compare_branches_tags, create_pull, etc.). Requiring a configured Gitee MCP server is appropriate for this purpose.
Instruction Scope: okSKILL.md limits actions to fetching diffs, analyzing changes, generating a title/body, and invoking create_pull. It does not instruct reading unrelated system files, environment variables, or posting to external endpoints outside the declared MCP tools. It explicitly states to ask the user if information is missing.
Install Mechanism: okInstruction-only skill with no install spec and no code files — nothing is written to disk or downloaded during install.
Credentials: okNo environment variables or extra credentials are requested by the skill itself. It depends on a pre-configured Gitee MCP server (platform-managed credentials), which is proportionate for creating PRs.
Persistence & Privilege: okalways is false and disable-model-invocation is not set to force disable; the skill is user-invocable and may be invoked autonomously by the agent (normal platform behavior). The skill does not request persistent system-wide privileges or modify other skills' configs.