The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read various local resources before writing documentation, creating a surface for indirect prompt injection if those files contain adversarial content.
Ingestion points: The 'Read First' section in SKILL.md specifies reading the target doc, nearest sibling docs, source code, and a reference corpus of existing MDX files.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from obeying instructions embedded within the files it is tasked to read.
Capability inventory: The skill allows the agent to execute local Node.js scripts and pnpm commands (SKILL.md).
Sanitization: No sanitization or filtering logic is described for the content being read from the repository.
[COMMAND_EXECUTION]: The skill provides the agent with specific shell commands to execute within the project environment.
Evidence: Instructions in SKILL.md direct the agent to run node .agents/skills/autogoal/scripts/create-goal-scratchpad.mjs with several flags to initialize documentation tasks.
Evidence: The workflow includes running pnpm install to regenerate skill copies and pnpm --filter www build:source / pnpm --filter www check:docs for verification.

docs-creator