The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends installing several security tools from public registries and repositories, such as pip-audit, cargo-audit, sigstore, slsa-verifier, pin-github-action, and osv-scanner. These tools are maintained by well-known organizations including Google and the SLSA framework.
[COMMAND_EXECUTION]: Provides numerous commands for auditing package registries, verifying artifact provenance, and managing container security. These commands are essential for the skill's purpose of incident response and supply chain security.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data such as lockfiles and dependency manifests.
Ingestion points: Processes files like package-lock.json, requirements.txt, and Cargo.lock (found in SKILL.md and references/details.md).
Boundary markers: Includes a mandatory confirmation gate requiring user authorization before executing active security probes or changes.
Capability inventory: Uses shell commands for package management (npm, pip, cargo), network requests (curl, gh), and system configuration (kubectl).
Sanitization: Advises the use of isolated environments, such as sandboxes or disposable VMs, and manual verification of digests and signatures.

supply-chain-attack-response