The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests data from test scenario files located in the temp/ directory and uses them to generate prompts for the copilot-cli-agent. This configuration presents an indirect prompt injection surface where a malicious scenario file could influence agent behavior.\n
Ingestion points: Phase 1 and 2 in SKILL.md read and process scenario JSON files.\n
Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between its own system instructions and the content of the $SCENARIO_PROMPT.\n
Capability inventory: The agent has access to file system read/write tools and the ability to execute shell commands via Bash.\n
Sanitization: No sanitization or safety checks are performed on the content of the scenario prompt before it is passed to the execution agent.\n- [COMMAND_EXECUTION]: The skill executes multiple shell commands for artifact verification, including grep, wc, and find. It also runs a local Python script scripts/experiment_log.py to persist test results to a durable log. While these are functional requirements for a testing tool, they involve the execution of external code and shell processing.

os-evolution-verifier