The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an external source (docs.phatysd.me) in both Markdown and JSON formats during the template discovery phase. This data is used to match project evidence and generate a recommendation.
Ingestion points: Template metadata and authored guidance are fetched from https://docs.phatysd.me/content/:slug/markdown and https://docs.phatysd.me/content/:slug/json.
Boundary markers: The skill contains a dedicated 'Research boundaries' section and mandates that a recommendation is always a 'Proposal' until the user explicitly confirms it. It also requires validation of representations to ensure identity consistency.
Capability inventory: The agent can read local documentation (AGENTS.md, docs/requirement.md, CONTEXT.md) and has the capability to update the CONTEXT.md file after receiving user confirmation. It also performs network operations to fetch template data.
Sanitization: The instructions explicitly forbid generating production code, creating prototypes, or invoking downstream skills based on the research. It treats external content as read-only research input.
[EXTERNAL_DOWNLOADS]: The skill is designed to fetch content from docs.phatysd.me. These network operations are necessary for the skill's primary function of researching published templates from the vendor's library.

setup-template