The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses various shell commands to inspect the local development environment and execute testing suites.
Evidence includes calls to python3, xcrun simctl (iOS simulator control), adb (Android Debug Bridge), pgrep, and make commands.
These actions are consistent with the skill's stated purpose of orchestrating evaluations in a local development environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from test reports and transcripts generated during the evaluation process.
Ingestion points: Reads content from e2e/agent-evals/reports/ including report.md and transcript JSON files.
Boundary markers: None explicitly defined in the instructions for separating report content from agent instructions.
Capability inventory: The agent has the ability to execute shell commands (make eval, scripts/eval.sh).
Sanitization: The skill relies on the user to interpret the analyzed results; no specific sanitization of the report data is mentioned.

run-evals