colibri

experimentalplatform

sovereign frontier (MoE) text generation via a local colibri server (github.com/JustVugg/colibri)

pip install chp-adapter-colibri

Auto-discovered through the chp.adapters entry point — no host config needed.

Capabilities

What this adapter exposes.

chatv1.0.0

chp.adapters.colibri.chat

generatev1.0.0

chp.adapters.colibri.generate

generate_statusv1.0.0

Poll a background generate job (from generate background=True) by job_id — returns

chp.adapters.colibri.generate_status

installv1.0.0

Download + extract the prebuilt `coli` binary for this platform from GitHub

chp.adapters.colibri.install

list_modelsv1.0.0

List models served by the local colibri server (OpenAI /v1/models), via chp.adapters.http.

chp.adapters.colibri.list_models

pull_modelv1.0.0

Fetch a pre-converted colibri model from a HuggingFace repo into the models dir, by

chp.adapters.colibri.pull_model

pull_statusv1.0.0

Report background model pulls started by pull_model(background=True): pidfile

chp.adapters.colibri.pull_status

server_statusv1.0.0

Report tracked `coli serve` processes (pidfile + liveness) per port.

chp.adapters.colibri.server_status

start_serverv1.0.0

Start `coli serve` for a model (a local dir, or a HF repo id — pulled first if not

chp.adapters.colibri.start_server

stop_pullv1.0.0

Stop a background model pull started by pull_model(background=True) — SIGTERM the

chp.adapters.colibri.stop_pull

stop_serverv1.0.0

Stop a tracked `coli serve` by port (SIGTERM + clear the pidfile).

chp.adapters.colibri.stop_server