The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection during its exploration phase.
Ingestion points: In WORKFLOW.md, the 'Explore' sub-agent (Step 1) is instructed to navigate the codebase naturally and record friction, meaning it reads and processes arbitrary content from the files within the repository.
Boundary markers: The instructions provided to the sub-agent do not include specific delimiters or directives to ignore instructions that might be embedded within the codebase (e.g., in code comments or documentation).
Capability inventory: The agent has access to Read, Grep, Glob, Write, and Bash(git *) tools.
Sanitization: There is no evidence of sanitization or filtering of the content read from the files before it is processed by the AI.
[COMMAND_EXECUTION]: The skill uses the Bash tool restricted to git * commands to perform churn analysis and navigate the repository, as specified in the SKILL.md metadata and referenced in the workflow.

arch-lens