The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project files and code changes which could contain malicious instructions. \n
Ingestion points: Reads ./tasks/prd-[feature-slug]/techspec.md, ./tasks/prd-[feature-slug]/tasks.md, and the output of git diff commands. \n
Boundary markers: The instructions lack explicit delimiters or warning markers to help the agent distinguish its core logic from the content of the files being analyzed. \n
Capability inventory: The skill has the ability to execute shell commands (git), run test suites (bun run test), and perform type checks (bun run typecheck). \n
Sanitization: No validation or sanitization is applied to the content of the documentation or code files before analysis. \n- [COMMAND_EXECUTION]: The skill executes various shell commands to perform the review. \n
It runs git status, git diff, git log, bun run test, and bun run typecheck. These operations execute code and scripts present in the local repository, which is a standard operational requirement for this type of skill but represents a potential execution vector if the code is malicious.

executar-review