The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructs the agent to automatically scan local configuration files, specifically .env.local and .env in the current working directory, to extract the SERVICEGRAPH_TOKEN credential. It explicitly tells the agent to use the found token without asking the user for permission.
[DATA_EXFILTRATION]: The instructions require the agent to check shell environment variables (e.g., $SERVICEGRAPH_TOKEN) to obtain authentication secrets.
[COMMAND_EXECUTION]: The skill utilizes shell commands including curl for interacting with the ServiceGraph API and openssl for generating firm identifiers via SHA-256 hashing.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from an external API (api.servicegraph.co) and integrates it into the agent's context.
Ingestion points: Data is retrieved from the /v1/search and /v1/get endpoints of the ServiceGraph API as described in SKILL.md.
Boundary markers: There are no defined delimiters or instructions to the agent to disregard any natural language instructions that might be embedded in the API responses.
Capability inventory: The agent maintains the ability to execute shell commands (curl, openssl) and read local filesystem data.
Sanitization: The skill does not provide mechanisms for sanitizing or validating the content returned from the external catalog before it is processed by the agent.

find-ai-consultancy