The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to inspect application manifests, middleware, and framework configurations to determine the monitoring strategy. This ingestion of untrusted local data, combined with the ability to modify source code and execute verification commands, creates a surface for indirect prompt injection.
Ingestion points: Application manifests, framework versions, middleware settings, and management endpoints (SKILL.md).
Boundary markers: The instructions do not specify any delimiters or safety markers to distinguish project data from instructions.
Capability inventory: The skill allows the agent to modify application code to add middleware/endpoints and execute shell commands (e.g., statlite inspect) for verification.
Sanitization: There are no requirements for sanitizing or validating the ingested project metadata before it influences the agent's actions.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to reference external documentation and implementation guides hosted at github.com/PVRLabs/statlite. These resources are used to guide the instrumentation of various frameworks like Spring Boot, Express, and FastAPI.
[COMMAND_EXECUTION]: The skill requires the agent to execute shell commands, specifically statlite inspect and other "focused checks," to verify the implementation of health endpoints and metrics collection.

lightweight-monitoring