The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its requirement to ingest project documentation for strategy sessions.
Ingestion points: Scans README, specs, and marketing docs in the repository to bootstrap context as specified in SKILL.md.
Boundary markers: No explicit delimiters are specified for processing external project documentation.
Capability inventory: The skill writes to .agents/positioning.md and POSITIONING.md, and utilizes platform tools for web search and fetching pages.
Sanitization: No content sanitization is described, although a privacy rule exists to avoid searching for unreleased product names.
[SAFE]: The skill uses network tools solely for market research and competitive analysis. It includes explicit privacy guardrails to avoid exposing unreleased product names to search engines.
[SAFE]: File access is limited to the skill's own artifact management and reading project documentation for context, with no access to sensitive system files or credentials.

positioning