Send a one-shot prompt to a local Ollama model and return its text response.

Use for any handoff where the cloud model's full reasoning isn't needed: drafts, boilerplate, simple extractions, formatting, or quick lookups. Runs on the user's own GPU and consumes no cloud-LLM usage. Returns the model's raw text completion.

Hold a multi-turn chat against a local Ollama model.

Use instead of ask_local when the handoff needs more than one turn of context — a running conversation or a system + user + assistant history. Runs locally at no cloud cost. Returns the model's next assistant message as text.

Summarize a block of text using the local model.

Use to offload long files, logs, transcripts, or docs the cloud model does not need to fully ingest — call this instead of reading a large blob into the cloud context. Runs on the user's GPU at no cloud cost. Returns a concise prose summary; pass focus to bias it toward what matters.

Run a quick first-pass code review using the local coder model.

Use as a cheap pre-filter before asking the cloud model for a deeper review: it catches obvious bugs, style issues, and risky patterns. Runs locally at no cloud cost. Returns review notes as text.

Draft a conventional-style commit message from a diff using the local model.

Use for routine commits where the cloud model's analysis isn't needed — it is cheap and fast. Runs locally at no cloud cost. Returns a single commit message (subject plus optional body) as text.

Extract specific information from a text block using the local model.

Use to pull structured facts out of unstructured text — function names, URLs, error codes, TODO comments, dependency names — without spending cloud-model tokens. Runs locally at no cloud cost. Returns the extracted items as text, shaped by what_to_extract.

List the Ollama models installed locally and available to these tools.

Use to discover valid values for the model parameter before pinning a specific model. Returns a list of model-name strings.

Report the server's effective configuration.

Returns the default model, Ollama base URL, context size, and request timeout. Use to confirm which model the tools will use by default or to debug connectivity. Returns a JSON object.