The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to manage the testing environment and verify functionality.
Evidence: Executes git diff main --name-only, git diff main --stat, and mktemp -d /tmp/kimi-smoke-XXXXXX for environment preparation.
Evidence: Runs the Kimi CLI using uv run python -m kimi_cli.cli with the --print flag, which automatically enables --yolo (auto-approval of all actions) for unattended testing.
Evidence: Executes a local helper script located at .agents/skills/feature-smoke-test/scripts/inspect_session.py to aggregate session results.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the repository and session outputs while possessing command execution capabilities.
Ingestion points: The agent reads code changes via git diff, source files specified in <source_paths>, and session logs (context.jsonl, wire.jsonl).
Boundary markers: Prompt templates in references/prompt-patterns.md provide structure for testing phases but lack explicit delimiters or warnings to ignore instructions embedded within the code or logs being analyzed.
Capability inventory: The skill can execute arbitrary shell commands and interact with the filesystem via the Kimi CLI and standard shell utilities.
Sanitization: No explicit sanitization or filtering of the ingested code or log content is performed before it is processed by the agent.

feature-smoke-test