From ollama
Internal helper contract for calling the ollama-companion runtime from Claude Code
How this skill is triggered — by the user, by Claude, or both
Slash command
/ollama:ollama-cli-runtimeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill only inside the `ollama:ollama-rescue` subagent.
Use this skill only inside the ollama:ollama-rescue subagent.
Primary helper:
node "${CLAUDE_PLUGIN_ROOT}/scripts/ollama-companion.mjs" task "<raw arguments>"The companion script talks directly to Ollama's HTTP API. There is no broker process and no app-server to start or stop — Ollama itself must be running at OLLAMA_HOST before the companion is invoked.
| Variable | Purpose | Default |
|---|---|---|
OLLAMA_HOST | Base URL for the Ollama HTTP API | http://127.0.0.1:11434 |
OLLAMA_PLUGIN_DEFAULT_MODEL | Default model when --model is not supplied | none (required to be set or passed explicitly) |
OLLAMA_PLUGIN_JOB_DIR | Override for background job storage directory | .ollama/companion-jobs/ inside workspace root |
OLLAMA_PLUGIN_LOG_LEVEL | Verbosity: silent, error, info, debug | info |
Pass --model <name> to override the default:
node "${CLAUDE_PLUGIN_ROOT}/scripts/ollama-companion.mjs" task --model qwen2.5-coder:14b "<task>"If --model is omitted and OLLAMA_PLUGIN_DEFAULT_MODEL is not set, the companion exits with a non-zero status and a message directing the user to /ollama:setup.
See the ollama-model-prompting skill for model selection guidance.
task once and return that stdout unchanged.git, direct Ollama API calls, or any other Bash activity.setup, review, adversarial-review, status, result, or cancel from ollama:ollama-rescue.task for every rescue request, including diagnosis, planning, research, and explicit fix requests.--model only when the user explicitly asks for one.--write unless the user explicitly asks for read-only behavior or only wants review, diagnosis, or research without edits.task invocation per rescue handoff.--background or --wait, treat that as Claude-side execution control only. Strip it before calling task, and do not treat it as part of the natural-language task text.--model, pass it through to task.--effort, pass it through to task. (The flag is accepted but currently inert — Ollama exposes no equivalent of Codex's reasoning-effort knob.)--emit-patch, pass it through. This forces the legacy patch-emit flow (the model returns a unified diff) instead of the default agentic tool-calling loop.--resume, strip that token from the task text and add --resume-last.--fresh, strip that token from the task text and do not add --resume-last.--resume: always use task --resume-last, even if the request text is ambiguous.--fresh: always use a fresh task run, even if the request sounds like a follow-up.--effort: accepted values are none, minimal, low, medium, high, xhigh.task --resume-last: internal helper for "keep going", "resume", "apply the top fix", or "dig deeper" after a previous rescue run.ollama:ollama-rescue unless the user explicitly asks for read-only behavior.task command exactly as-is.Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub darrylmorley/ollama-plugin-cc --plugin ollama