The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown documentation defining architectural and behavioral contracts. It does not contain executable scripts, binaries, or network-invoking commands.
[SAFE]: The guidelines in review-contract.md promote defensive security practices by instructing agents to identify and block critical vulnerabilities such as hardcoded secrets, lack of ownership checks on public writes, and insecure data handling (e.g., non-deterministic queries, un-awaited writes).
[SAFE]: execution-context.md enforces a stateless execution model, explicitly prohibiting agents from creating hidden registries, status files, or persistent runtime caches unless explicitly authorized by the user.
[SAFE]: pr-ship.md encourages the use of repository-defined linting and testing scripts before code is committed, ensuring that standard quality and security checks are executed.
[SAFE]: No patterns of prompt injection, obfuscation, data exfiltration, or unauthorized privilege escalation were detected across the analyzed files.

pack-shared