The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from repository files such as README.md, product documentation, and package metadata to extract marketing facts. This creates an inherent surface for indirect prompt injection where malicious instructions embedded in these project files could influence the agent's behavior during context maintenance tasks.
Ingestion points: The logic in references/source-policy.md identifies authoritative project artifacts like README.md and package metadata as valid sources for repository inspection.
Boundary markers: The instructions do not define specific delimiters or boundary markers to isolate ingested file content from the system prompt during processing.
Capability inventory: The skill enables the agent to perform file-write operations to .agents/marketing-context.md and file-delete operations for legacy context files as documented in references/merge-and-conflicts.md.
Sanitization: There is no mention of technical sanitization for ingested content; the skill relies on the agent's logical ability to distinguish confirmed facts from hypotheses.
[NO_CODE]: The skill consists entirely of markdown instructions, YAML configuration, and SVG assets. It does not contain any executable scripts, binaries, or automated installation routines, which significantly limits its ability to perform unauthorized actions outside of the agent's prompted behavior.

a1-marketing-context