The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local commands to analyze code, run tests, and manage git state. Specifically, it calls npx tsx scripts/analyze-codebase.ts to build a dependency graph and npm run for type-checking, linting, and building.
[EXTERNAL_DOWNLOADS]: Commands like npx tsx and npx vitest are used, which may trigger downloads from the NPM registry if the packages are not present in the local environment or cache.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection from the source code it processes. An attacker with write access to the repository could embed malicious instructions in code comments.
Ingestion points: All files in the target repository directories (e.g., src/lib, src/api) are read and processed by worker agents.
Boundary markers: The orchestration prompt lacks explicit instructions to treat the ingested code as data only and to ignore embedded instructions.
Capability inventory: The skill possesses significant capabilities, including the ability to run arbitrary project scripts and perform git commits.
Sanitization: The input code is passed directly to the worker agents without filtering or sanitization.

simplify-parallel