The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to download and execute the 'serena' tool directly from an unverified third-party GitHub repository (github.com/oraios/serena) using uvx. This pattern executes remote code at runtime without verifying the source against a trusted vendors list.
[COMMAND_EXECUTION]: The skill triggers several local shell commands and scripts, including bash scripts/serena_gate.sh, bash scripts/lint.sh, gh pr diff, and git diff. This assumes that the local scripts and environment are secure and trustworthy.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from code diffs and pull request feedback. Malicious instructions embedded in code comments could influence the agent's behavior or conclusions during the review process.
Ingestion points: SKILL.md (via git diff, gh pr diff, and external review feedback).
Boundary markers: Absent; there are no delimiters or instructions to ignore embedded commands within the processed code.
Capability inventory: Extensive shell execution capabilities including bash, uvx, gh, and git across multiple steps.
Sanitization: Absent; the skill does not specify any filtering or escaping of the ingested code content before processing.

vibe-review-code