The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided fiction chapters and story metadata to generate persona-based feedback. This content is inherently untrusted and could contain instructions intended to override the agent's behavior or the specific persona constraints.
Ingestion points: The agent reads story.md, individual chapter files, and the output generated by the story context command.
Boundary markers: The workflow utilizes subagents to isolate personas from each other and uses the story context command to limit the background information provided to the persona, which provides some structural isolation.
Capability inventory: The skill can execute shell commands via the story CLI, write feedback files to the local filesystem, and spawn subagents.
Sanitization: There is no evidence of explicit sanitization or filtering of the prose content to detect or neutralize embedded instructions before it is processed by the AI personas.
[COMMAND_EXECUTION]: The skill workflow relies on the execution of several shell commands through a local CLI tool (story). These commands include parameters derived from the user's project state, such as chapter ranges and round numbers.
Evidence: The skill calls story build . --format html --stamp panel-round-{N}, story context chapter-{NN} --path ., story reindex ., story links ., and story validate .. While these are domain-specific project management commands, they represent a standard surface for command execution within the agent's environment.

reader-panel