The Agent Skills Directory

[SAFE]: No malicious instructions, obfuscation, or safety bypass attempts were found in the skill metadata or body.\n- [COMMAND_EXECUTION]: The skill interacts with the local environment exclusively through the qodo CLI, using structured commands like qodo read pr-review-session and qodo pr-review-session mark-implemented to perform its intended functions.\n- [EXTERNAL_DOWNLOADS]: The skill mentions the official get.qodo.ai domain for manual CLI installation but includes a clear guardrail instructing the agent to never pipe an installer directly into a shell or invent checksums.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of pull request review findings, which is a potential injection surface.\n
Ingestion points: Data enters the agent context via qodo read pr-review-session findings in SKILL.md.\n
Boundary markers: The skill enforces human-in-the-loop review; findings are not pre-selected, and the agent must explicitly ask the user which findings to resolve unless an explicit autofix argument is provided by the user.\n
Capability inventory: The skill can modify local source code to fix findings and use the qodo CLI to record resolutions in the vendor's database.\n
Sanitization: Instructions require the agent to independently evaluate findings against the actual code and PR intent before presenting recommendations, acting as a technical filter rather than a blind executor.

qodo-review-resolver