The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use common shell utilities (find, grep, ls, git submodule status) to explore the project structure and extract architectural patterns from source code. These operations are essential for the skill's purpose and are scoped to the project's directory.
[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and interpreting data from the project being analyzed, such as manifest files (package.json, pyproject.toml) and source code. This introduces a surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: Manifest files, source code in the src directory, and existing documentation files (docs/adr/*.md, docs/standards/*.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between file data and control instructions during analysis.
Capability inventory: Filesystem inspection (ls, find, grep) and the ability to trigger the arch-manage skill to write new files.
Sanitization: There is no mention of sanitization or filtering of the content read from the project files.

arch-discover