The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted dependency manifest files (e.g., package.json, requirements.txt, pom.xml) to perform its audit functions. This creates a potential surface for indirect prompt injection if those files contain malicious metadata designed to influence the agent's behavior.
Ingestion points: The DependencyDiscovery class in resources/implementation-playbook.md parses various project configuration files to inventory dependencies.
Boundary markers: No explicit delimiters or instructions to ignore embedded instructions within the parsed files were found.
Capability inventory: The skill can perform network requests to external APIs, execute shell commands for dependency updates, and generate pull requests.
Sanitization: The provided script templates do not explicitly sanitize package metadata before it is used in subsequent operations or prompts.
[COMMAND_EXECUTION]: The implementation playbook includes templates for shell scripts to automate security updates and dependency management. These scripts use standard tools such as npm audit fix, pip-compile, and git commands to manage project state.
[DATA_EXFILTRATION]: The skill performs network operations to fetch metadata and security advisories from established package registries and services. These operations are essential for the skill's primary function and target well-known, trusted domains.
External references: Vulnerability data is fetched from official registries including registry.npmjs.org, pypi.org, and rubygems.org. Bundle size analysis is performed via bundlephobia.com.

dependency-management-deps-audit