The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external codebases and configuration files, which presents a surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: Uses Read, Glob, and Grep to inspect target repositories and WebFetch to retrieve documentation from Cohere's official sources.
Boundary markers: The skill contains explicit instructions to maintain clear boundaries between retrieval and model processing, such as applying filters before Rerank and citations.
Capability inventory: The skill can perform file modifications using Write and Edit tools.
Sanitization: The instructions explicitly mandate redacting keys, authorization headers, customer identifiers, and retrieved documents from output to prevent data leakage. Additionally, it requires explicit owner approval for any side effects or changes to production environments.
[EXTERNAL_DOWNLOADS]: The skill references and fetches content from official Cohere domains and repositories. These are well-known services and the skill provides specific instructions to only use WebFetch for official primary documentation.

cohere-reference-architecture