The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents Retrieval-Augmented Generation (RAG) patterns which are inherently susceptible to indirect prompt injection if the ingested documents contain malicious instructions. This is a standard risk for the described use case.
Ingestion points: The upsert and retrieve functions in SKILL.md process document text and user queries into the vector database.
Boundary markers: None identified; the examples use standard interpolation into prompt strings.
Capability inventory: The skill uses qdrant-client for network-based database operations (upsert, search, create_collection).
Sanitization: No explicit sanitization or validation of input data is demonstrated in the basic examples.
[COMMAND_EXECUTION]: The documentation includes standard shell commands for environment setup, package management, and service health checks.
Evidence: pip install, docker run, docker ps, and curl commands are provided in SKILL.md and references/troubleshooting.md for local development setup.
[PRIVILEGE_ESCALATION]: The troubleshooting guide mentions administrative commands for managing local service daemons.
Evidence: sudo systemctl start docker in references/troubleshooting.md is provided as a fix for common Docker connection issues.

qdrant-vector-search