The Agent Skills Directory

[DATA_EXFILTRATION]: Accesses sensitive configuration files. The skill directs the agent to read ~/.bashrc, ~/.zshrc, .env.local, and .env files to find API tokens. These files are highly sensitive and commonly contain private keys, database credentials, and other system secrets.
[DATA_EXFILTRATION]: Bypasses user oversight for credential retrieval. The instructions specify that if a token is found in environment files, the agent should "use it; don't ask," which circumvents the user's ability to monitor or approve access to their local environment data.
[COMMAND_EXECUTION]: Shell command usage. The skill instructs the agent to use curl for API interactions and provides both bash and Python snippets for local data processing and identifier calculation.
[EXTERNAL_DOWNLOADS]: Outbound API communication. The skill connects to api.servicegraph.co to perform law firm searches, validate filters, and handle authentication.
[PROMPT_INJECTION]: Indirect prompt injection risk. The skill processes untrusted data from external API responses and integrates it into the agent's context without adequate security boundaries or sanitization.
Ingestion points: Responses from the ServiceGraph API endpoints (/v1/tags, /v1/search, /v1/get).
Boundary markers: Absent. The skill provides no instructions for the agent to distinguish between valid data and potential instructions embedded within API results.
Capability inventory: The skill has the capability to execute shell commands and read sensitive local files.
Sanitization: Absent. There is no requirement for validating or escaping content retrieved from the remote API.

find-law-firm