The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands to map the codebase and run vulnerability scanners.
Evidence includes shell pipes using find, grep, xargs, and wc to analyze project files.
The skill attempts to run external audit tools including pnpm audit, npm audit, pip-audit, safety, govulncheck, and cargo audit.
[CREDENTIALS_UNSAFE]: The skill uses automated patterns to search for and expose hardcoded secrets within the project files.
It executes grep commands targeting highly sensitive patterns such as sk_live (Stripe live keys), sk_test, AKIA (AWS Access Key IDs), and password assignments.
It explicitly searches for .env files that have been cached in git history using git ls-files --cached.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests the entire project codebase as untrusted data.
Ingestion points: The skill reads all files matching common programming extensions (.ts, .py, .go, .rs, etc.) into the agent context.
Boundary markers: No explicit delimiters or instructions are used to prevent the agent from obeying instructions embedded in the analyzed code.
Capability inventory: The skill has access to shell execution and sub-agent spawning, which could be exploited by malicious comments in the codebase.
Sanitization: There is no evidence of sanitization or filtering of the codebase content before it is processed by the analysis engine.

security-scan