The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from untrusted sources to inform its recommendations.\n
Ingestion points: Reads local project files including README.md, CLAUDE.md, and dependency manifests in Phase 1a, and processes results from WebSearch in Phase 2.\n
Boundary markers: The instructions do not specify the use of clear delimiters or instructions to ignore embedded commands within the ingested content.\n
Capability inventory: The skill has the ability to execute shell commands (git log, mkdir, and path resolution scripts) and perform network requests via the WebSearch tool.\n
Sanitization: There is no mention of sanitizing or filtering external content before it is interpolated into the agent's context or the final report output.\n- [COMMAND_EXECUTION]: The skill utilizes shell commands for project exploration and file system organization.\n
Evidence: Phase 1a executes git log --oneline -20 to understand project history.\n
Evidence: Phase 3 uses a bash script to create the ./ideas directory and ensure unique file naming for the generated reports.\n- [DYNAMIC_EXECUTION]: The skill generates and runs a shell script at runtime to resolve output paths.\n
Evidence: Phase 3 contains a bash snippet that dynamically constructs a file path using a date string and a user-influenced SLUG variable. This is a low-risk pattern for managing file output.

compare-tools