The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash via run.sh to execute a local Python script ops_qdrant.py using uv run. This is a standard execution pattern for this environment.
[EXTERNAL_DOWNLOADS]: The skill references official documentation and model cards for Qdrant and Jina AI (e.g., qdrant.tech, huggingface.co). These are trusted well-known services used for informational retrieval, not runtime code execution. All library dependencies in pyproject.toml (typer, httpx, loguru, pydantic) are standard, well-known packages from public registries.
[DATA_EXFILTRATION]: The skill interacts with local endpoints (127.0.0.1) for Qdrant, a memory daemon, and an embedder. It performs read-only operations to report health status. No sensitive file access or exfiltration to external domains was observed.
[SAFE]: The assess command functions as a static analysis linter to enforce architectural boundaries (preventing direct Qdrant mutation from other skills). The sanity.sh script includes safety checks to ensure the skill itself remains read-only by scanning its own source code for mutation verbs.

ops-qdrant