The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill directs the agent to base its routing decisions on external documentation, including other SKILL.md files, GLOSSARY.md, and Architecture Decision Records (ADRs). This establishes an ingestion surface for instructions that could be embedded in those files.
Ingestion points: External SKILL.md files, GLOSSARY.md, ADRs, and issue trackers located in the repository.
Boundary markers: The skill does not provide specific markers or safety instructions to the agent regarding the handling of content within these external files.
Capability inventory: The router itself is restricted via disable-model-invocation: true, but it facilitates transitions to other tools with capabilities such as file system modification, test execution, and bash script generation.
Sanitization: No explicit sanitization or content validation steps are defined for the data retrieved from the repository files.
[EXTERNAL_DOWNLOADS]: The documentation references aihero.dev, a well-known technical resource for AI development, for concepts related to context window management ('smart zone'). This reference is informational and links to a reputable source.

ask-matt