The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automatically infers and executes shell commands to run project test suites (e.g., 'npm test', 'pytest', 'go test') based on the detected languages and repository structure. This is an intended function for code verification.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and processes untrusted data from the project repository.
Ingestion points: The skill reads various markdown files and source code diffs (SKILL.md).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard instructions potentially embedded within the source files being reviewed.
Capability inventory: The agent can execute inferred test commands, perform file system modifications (applying fixes), execute git commits, and delete/archive project directories.
Sanitization: No explicit validation or filtering of input file content is performed prior to processing.
[SAFE]: The skill implements standard developer operations including git management and directory cleanup. While it possesses capabilities for data modification and deletion (archiving/removing tracks), these operations are gated by user confirmation requests or are central to the tool's defined workflow.

conductor-review