dbk-pr-feedback
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the GitHub CLI (
gh) to interact with the GitHub API. This is a standard and expected behavior for the described workflow. - Evidence:
gh api repos/{owner}/{repo}/pulls/{pr_number}/commentsandgh api repos/{owner}/{repo}/issues/{pr_number}/commentsinSKILL.md. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external PR comments which could contain malicious instructions designed to influence the agent's behavior during code modification.
- Ingestion points: Fetches review comments and PR comments via
gh apicalls defined inSKILL.md. - Boundary markers: None. The skill lacks specific instructions to treat fetched comments as untrusted or to ignore embedded formatting/instructions that might attempt to override the agent's logic.
- Capability inventory: The agent has the capability to modify project files based on the evaluation of these comments (Step 4: "只修改文件" in
SKILL.md). - Sanitization: No sanitization or validation of the fetched comment content is mentioned before it is processed for code changes.
Audit Metadata