The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes repository source code and local documentation files, creating a vulnerability surface where instructions embedded in those files could influence agent behavior. (1) Ingestion points: The agent reads the local codebase and specific project files located in intent/ directories. (2) Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the ingested files. (3) Capability inventory: The skill is permitted to perform file system reads, write findings to context.md, and execute local git commits for documentation. (4) Sanitization: The skill does not describe any sanitization or validation of the data read from the repository. This surface is inherent to the skill's primary function as an SDLC thinking partner and is managed by explicit directives to isolate work and avoid unintended changes to the repository.

sdlc-explore