The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a script scripts/forge-exit.js designed to execute standard development commands (e.g., npm run lint, pytest, cargo test) to verify the implementation of project slices. These operations are restricted to established development tools and local project context.
[DYNAMIC_EXECUTION]: The script scripts/ddl-exec.js optionally uses @electric-sql/pglite (a well-known WASM-based PostgreSQL engine) to validate SQL schema definitions provided in documentation. This allows the agent to verify that constraints and triggers are syntactically and logically correct through actual execution in a sandboxed database environment.
[INDIRECT_PROMPT_INJECTION]: Forge processes data from potentially untrusted project files (PRDs, roadmaps, architecture). To mitigate risks of malicious instructions embedded in these files, the skill enforces a multi-perspective review protocol (Quality Analyst, Red Team) and requires explicit human validation at every phase gate before proceeding.

forge