The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands including mdbook-listings, panschema, and mdbook build. It also directs the agent to run a repository-local shell script ./scripts/install-assets.sh to generate necessary project assets.
[EXTERNAL_DOWNLOADS]: The skill utilizes the panschema fetch command to download data (cqa contracts) from a remote source into a local cache.
[INDIRECT_PROMPT_INJECTION]: The authoring workflow relies on the agent reading and following instructions (CHARTER, OUTLINE, DEFERRALS, and CHECKLIST) found within HTML comments in the markdown files being edited. This establishes a surface where external content could influence agent actions.
Ingestion points: Markdown files in the book/ directory (e.g., introduction.md, ch01.md).
Boundary markers: Specific HTML comment labels are used as section headers, but no explicit instructions are provided to the agent to disregard potentially malicious content within those blocks.
Capability inventory: The agent has capabilities to write files, execute shell commands, and run local scripts.
Sanitization: No sanitization or validation of the content within the HTML comments is performed before the agent acts on the instructions.

ontology-authoring-advance-step