The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data including git diffs, PR content, and project documentation (spec files and context docs) to perform its review. If this content contains malicious instructions, it could attempt to influence the agent's review logic or triage process.
Ingestion points: Reads {diff_output} via git commands and loads text from {spec_file} and other documentation files.
Boundary markers: The workflow does not explicitly mention the use of delimiters or 'ignore' instructions for the ingested code content.
Capability inventory: The skill can write to story files and the backlog, and execute local subprocesses (Python, Git, GitHub CLI).
Sanitization: No specific sanitization or filtering of the ingested diff or spec content is described.
[COMMAND_EXECUTION]: The skill executes local Python scripts and standard developer tools to manage its workflow.
Evidence: Calls python3 {project-root}/_bmad/scripts/resolve_customization.py during initialization and at the end of the workflow to handle configuration merging and final actions. It also uses git and gh for context gathering.
[DYNAMIC_EXECUTION]: The skill supports dynamic behavior by executing commands defined in project-level configuration files.
Evidence: The on_complete hook in step-04-present.md resolves a command string from the workflow.on_complete configuration key and executes it via a Python wrapper script at the end of the review process.

gds-code-review