The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or data exfiltration attempts were detected. The skill defines a standard engineering process for code analysis.
[SAFE]: The skill contains specific guidance to protect sensitive information, instructing the agent to never include secrets, credentials, or API keys in the review ledger.
[PROMPT_INJECTION]: The skill is designed to process external, untrusted content (source code, diffs, and pull request descriptions), which represents a surface for indirect prompt injection. Malicious instructions could be embedded in the code under review to attempt to manipulate the agent's findings.
Ingestion points: Code diffs, source files, project documentation, and PR descriptions in SKILL.md.
Boundary markers: Absent; there are no specific markers defined to isolate the content being reviewed from the agent's instructions.
Capability inventory: Tool-based operations including doctrine review raise (ledger logging), backlog new (task creation), and memory storage (/record-memory).
Sanitization: Not explicitly defined; the skill relies on the agent's underlying safety filters and the user's manual review of the generated ledger.

code-review