The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes high-privilege system commands including docker-compose, npx, and flutter to manage application infrastructure and run automated tests.
[COMMAND_EXECUTION]: Uses curl to interact with local API endpoints, allowing it to verify health status and perform functional testing of the backend server.
[PROMPT_INJECTION]: Exhibits a significant indirect prompt injection surface (Category 8) by processing local project code and runtime API responses to autonomously diagnose and apply code fixes. Maliciously crafted data within the ingested files or API responses could theoretically influence the agent's logic to introduce vulnerabilities during the automated remediation phase.
Ingestion points: Reads various project configuration and source files (e.g., routes.ts, schema.prisma, package.json) and captures output from curl requests during testing.
Boundary markers: Absent; there are no clear delimiters or instructions to prevent the agent from being influenced by instructions embedded within the data it processes.
Capability inventory: The agent has the authority to execute shell commands and autonomously modify project source code across multiple files.
Sanitization: There is no evidence of input validation or sanitization of the ingested data before it is used to generate and apply code fixes.
[COMMAND_EXECUTION]: Implements a persistent logging mechanism by appending execution telemetry to ~/.claude/projects/skill-telemetry.md if the directory exists.

qa