The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local CLI tool named doctrine to fetch data and record results.
Evidence includes shell commands for fetching questions: doctrine compare elicit --json [--depth K] [--kind comparison|anchor-review].
Evidence includes shell commands for recording answers: doctrine compare record <A> <B> --prefer a|b | --equal | --incomparable --rater human --by <name> [--frame F] [--audience T] [--note ...].
Evidence includes a nomination command: doctrine estimate set.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from an external source (the output of the doctrine command).
Ingestion points: Data enters the agent's context through the JSON output of the doctrine compare elicit command, specifically fields such as reasons, ask, and subject.anchor (found in SKILL.md).
Boundary markers: The instructions do not define explicit delimiters or instructions for the agent to ignore potentially malicious content embedded within the tool's JSON output.
Capability inventory: The agent has the capability to execute subprocesses and write to the comparison ledger via the doctrine compare record and doctrine estimate set commands (found in SKILL.md).
Sanitization: There is no evidence of sanitization, validation, or escaping of the strings returned by the CLI tool before they are presented to the user or used to frame the next steps in the elicitation session.

elicit