The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs the agent to read sensitive configuration files and extract plaintext secrets. Specifically, it uses cat .env | grep -E "(OPENAI|ANTHROPIC|GOOGLE)_API_KEY" to verify keys, which directly exposes these credentials to the LLM context.
[COMMAND_EXECUTION]: The skill's core functionality relies on executing shell commands such as cat and grep to perform static analysis on the local filesystem, including dependencies, environment variables, and TypeScript configurations.
[PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection (Category 8) because it ingests data from untrusted files across the project directory.
Ingestion points: The agent reads src/mastra/index.ts, package.json, tsconfig.json, .env, and all files matching src/**/*.ts.
Boundary markers: No delimiters or instructions are provided to the agent to ignore potentially malicious content or instructions embedded within the scanned code files.
Capability inventory: The agent has the ability to execute shell commands (cat, grep) and is instructed to generate code fixes based on the files it reads.
Sanitization: The skill lacks any sanitization, escaping, or validation steps for the content retrieved from the filesystem before it is processed by the LLM.

mastra-system-check