The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to discover and execute shell commands found within the repository (e.g., scripts in package.json, Cargo binaries, or Makefiles) to start the backend server.
[EXTERNAL_DOWNLOADS]: Uses networking tools such as curl and http to interact with API endpoints. While intended for testing the live server, these tools have the capability to make external network requests.
[DATA_EXFILTRATION]: The instructions guide the agent to identify and record authentication tokens, environment variables, and test account credentials into project documentation (project.md) to facilitate testing.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the repository, including project specifications and documentation, which could potentially influence agent behavior if those files contain malicious instructions.
Ingestion points: Reads docs/agents/project.md, repository specifications, and build configuration files (package.json, Makefile, etc.).
Boundary markers: None explicitly defined for separating instructions from repository data.
Capability inventory: Includes file system writes (project.md) and subprocess execution (starting the discovered server commands and running curl).
Sanitization: No specific sanitization or validation of the discovered commands or specification content is mentioned.

acceptance-api