The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external documents and metadata for search and retrieval, creating a surface for indirect prompt injection attacks.
Ingestion points: Document addition via collection.add() in SKILL.md and vectorstore.from_documents() in references/integration.md.
Boundary markers: The provided examples do not demonstrate the use of delimiters or specific instructions to the model to ignore potential instructions embedded within retrieved data.
Capability inventory: The skill interacts with the local file system for database persistence (PersistentClient) and performs similarity searches.
Sanitization: No explicit input validation or sanitization logic is shown for the content being stored in the database.
[EXTERNAL_DOWNLOADS]: The documentation references standard package installations from public registries and official project resources.
Evidence: Installation instructions for chromadb via pip and npm, and links to the official project repository at github.com/chroma-core/chroma.
[COMMAND_EXECUTION]: The skill provides standard developer terminal commands for environment setup and service execution.
Evidence: Examples include pip install chromadb, npm install chromadb, and starting the database server using chroma run --path ./chroma_db.

chroma