The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from a local codebase (Phase 3 and 4) which could contain malicious instructions embedded in comments or strings.
Ingestion points: Local source files (.ts, .py, .go, etc.) analyzed via claudemem and the Read tool.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded prompts within the code results.
Capability inventory: The skill possesses extensive capabilities including arbitrary shell command execution via the Bash and Task tools, as well as file system read access.
Sanitization: No explicit sanitization or validation of the data returned by the AST analysis tool is implemented before the agent processes it.
[COMMAND_EXECUTION]: The skill relies on complex shell script templates for environment setup and result validation.
Evidence: Multiple scripts in the 'MANDATORY SETUP' and 'Result Validation Pattern' sections use the Bash tool to execute logic involving find, stat, grep, and wc to check file system state and index freshness.

developer-detective