The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the vendor-owned commitshow package from the NPM registry using the command npx commitshow@^0.3.23.
[COMMAND_EXECUTION]: Executes shell commands to manage the audit workflow, including mkdir for creating the sidecar directory and npx for running the audit engine.
[DATA_EXFILTRATION]: Transmits repository data and signals to the remote API at api.commit.show for analysis, which is the documented primary function of the tool.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface:
Ingestion points: Data enters the agent's context from the .commitshow/audit.json file, specifically from the concerns[].bullet fields containing audit findings.
Boundary markers: Absent. The skill does not instruct the agent to use delimiters or ignore embedded instructions when reading tool output.
Capability inventory: The skill has the capability to execute shell commands and read/write local files.
Sanitization: Absent. There is no mention of validating or sanitizing the strings returned by the audit engine before the agent uses them to propose code changes.

production-audit