The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external project files, including source code and feature description maps, which could contain adversarial instructions targeting the agent.
Ingestion points: Reads feature files and source code files from the target project directory during the 'Source wave' phase.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to isolate content read from source files from the agent's control logic.
Capability inventory: The coordinator agent has the capability to write to the skill's directory and execute harness scripts that interact with the system and application.
Sanitization: There is no evidence of sanitization or filtering of external content before it is processed by subagents or used by the coordinator to plan verification steps.
[COMMAND_EXECUTION]: The skill performs live verification passes by executing project-local harness scripts and driving application instances.
Evidence: Section 4 ('Live pass') and Section 5 ('Triage') describe driving the app via launch models (servers, UIs, or CLIs) and ensuring harness scripts are executable to verify feature behavior.

maintain-verification-skill