The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze external project data, including code, architecture decision records (ADRs), and technical specifications, to identify deviations from established project "doctrine."
Ingestion points: Project-local files such as CLAUDE.md, README.md, .doctrine/adr/, and target implementation files are read into the agent's context.
Boundary markers: There are no explicit instructions to use delimiters or to disregard potentially malicious instructions embedded within the reviewed targets.
Capability inventory: The skill utilizes the doctrine CLI suite (backlog, review, harvest) to write structured findings to a ledger file (review-NNN.md).
Sanitization: The instructions do not specify any validation or sanitization steps for the data retrieved from external files before processing.
[COMMAND_EXECUTION]: The skill requires the use of a custom doctrine CLI tool for lifecycle management of reviews, including commands for creating backlog items, priming review caches, and concluding trials. It also utilizes a show CLI command to read project entities.
[PROMPT_INJECTION]: The skill mandates the adoption of a highly aggressive "Inquisitor" persona, using imagery associated with "medieval ecclesiastical zealots" and "executioners." While directed at technical review tasks, such extreme persona instructions can occasionally lead to unintended model behaviors or safety filter triggers.

inquisition