free-claude-code-plus

A community fork of Alishahryar1/free-claude-code that adds first-class support for the MiniMax Token Plan as an LLM provider. Route Claude Code (CLI + VS Code) through MiniMax-M3 / M2.7 / M2.5 / M2.1 / M2, or mix MiniMax with NVIDIA NIM, OpenCode Zen, OpenRouter, and 14 other providers in the same project.

Fork mantido por @lpdigital.me — Philipe compartilha plugins, automações e IA toda semana no Instagram. Manda um follow se o proxy te ajudou.

What is this?

free-claude-code (FCC) is a proxy server that sits between the Claude Code CLI / VS Code extension and a backend LLM provider. By default, FCC points at NVIDIA NIM (free tier) — but you can configure it to talk to 18+ providers, including OpenCode Zen, OpenRouter, Groq, Cerebras, Fireworks, Z.AI, Kimi, DeepSeek, Mistral, Codestral, Gemini, and now MiniMax.

This fork exists because the upstream didn't yet have a MiniMax provider, and adding it correctly required more than a config line — it required a new AnthropicMessagesTransport subclass because MiniMax speaks Anthropic's wire format on a non-standard path.

┌────────────┐  Anthropic API  ┌──────────────────┐  Anthropic-compat    ┌────────────┐
│ Claude Code│ ──────────────► │  fcc-server      │ ───────────────────► │  MiniMax   │
│ (CLI/VSX)  │ ◄────────────── │  (this fork)     │ ◄─────────────────── │  Token Plan│
└────────────┘  SSE streaming  └──────────────────┘  x-api-key header     └────────────┘
                                            │
                                            │  same proxy can fan out to
                                            ▼
                                    ┌──────────────────┐
                                    │  NVIDIA NIM      │
                                    │  OpenCode Zen    │
                                    │  OpenRouter      │
                                    │  14+ more…       │
                                    └──────────────────┘

Why this fork exists

The upstream free-claude-code is excellent but moves slowly. MiniMax's Token Plan is one of the best value-for-money LLM subscriptions in the Brazilian market (R$120/month for M3 + M2.7 + M2.5 + M2.1 + M2 with 5h rolling windows), but adding it as a first-class provider required:

A new providers/minimax/ module (subclass of AnthropicMessagesTransport)
A new MINIMAX_DEFAULT_BASE constant (with the right /v1 suffix)
Wiring into the PROVIDER_CATALOG so it shows in the Admin UI
Env-var plumbing (MINIMAX_API_KEY) and Admin UI fields
A critical bug fix: the base URL must include /v1 or /messages resolves to a 404

All of that is in this fork. The bug fix alone is non-trivial to discover (see § Bug fix: missing /v1 suffix below).

What you get

With MiniMax enabled

# Set the API key
export MINIMAX_API_KEY=sk-cp-...

# Use any MiniMax model as if it were a native Claude model
claude --model minimax/MiniMax-M3 "explana o teorema de Bayes em PT-BR"
claude --model minimax/MiniMax-M2.5 "write a python web server"
claude --model minimax/MiniMax-M2 "summarize this diff"

The MiniMax-M3 is the flagship 1M-context model; the -highspeed variants trade some quality for latency.

With multi-provider routing

Configure the proxy to fan out across providers based on the request:

// ~/.claude/settings.json
{
  "env": {
    "ANTHROPIC_BASE_URL": "http://127.0.0.1:8082",
    "ANTHROPIC_AUTH_TOKEN": "dummy"
  }
}

Then the model id routes through fcc-server:

Model id	Provider
`minimax/MiniMax-M3`	MiniMax Token Plan
`nvidia_nim/meta/llama-3.3-70b-instruct`	NVIDIA NIM
`opencode/big-pickle`	OpenCode Zen
`open_router/anthropic/claude-sonnet-4`	OpenRouter
`groq/llama-3.3-70b-versatile`	Groq
`cerebras/llama-3.3-70b`	Cerebras
`deepseek/deepseek-chat`	DeepSeek
`gemini/gemini-2.5-pro`	Gemini
`mistral/mistral-large-latest`	Mistral
`kimi/kimi-k2-0711-preview`	Kimi (Moonshot)
`fireworks/accounts/fireworks/models/llama-v3p3-70b-instruct`	Fireworks
`zai/glm-4.5`	Z.AI
`codestral/codestral-2508`	Mistral Codestral
`ollama/llama3.2`	Ollama (local)
`llamacpp/...`	llama.cpp (local)
`lmstudio/...`	LM Studio (local)
`wafer/...`	Wafer (experimental)

Mix and match per conversation or per project. Use DISABLED_PROVIDERS and HIDDEN_MODELS (or the Admin UI's model visibility panel) to control which models appear in the /model picker.

Quick start

0. Install with an AI assistant (recommended if you're not technical)

Copy the prompt below and paste it into any AI chat (ChatGPT, Claude, Gemini, etc.). The assistant will install this wrapper end-to-end on your machine — no programming knowledge required on your side.

Click to reveal the install prompt — copy everything inside the box

You are helping me install "free-claude-code-plus", a Python wrapper
around the Claude Code CLI that adds support for MiniMax, OpenRouter,
OpenAI, DeepSeek, Mistral, and 13+ other LLM providers as first-class
backends (Claude Code natively only supports Anthropic). It runs
locally as `fcc-claude` / `fcc` and is normally installed via uv.

Your job: install it on my machine end-to-end and verify it works.
Do not ask me coding questions — make sensible decisions and tell me
what you did. If you hit a step that needs my input (e.g. choosing
a provider key), ask exactly one focused question and continue.

=========================================================
STEP 1 — Detect my environment
=========================================================
Run these commands and remember the output:

  uname -a                          # OS family (Darwin / Linux / Windows-bash)
  python3 --version                 # need 3.10+ (3.14+ recommended)
  command -v uv                     # need uv 0.11+ for `uv tool install`
  command -v claude                 # is Claude Code already installed?
  command -v fcc-claude             # is this wrapper already installed?
  echo "KEYS=${MINIMAX_API_KEY:-unset} ${OPENROUTER_API_KEY:-unset}"  # which keys exist?

If `uv` is missing, install it:
  curl -LsSf https://astral.sh/uv/install.sh | sh
  # restart the shell afterwards

If `python3` is missing, tell me to install Python 3 from
https://python.org and stop.

=========================================================
STEP 2 — Install the wrapper
=========================================================
Run:

  git clone https://github.com/philipecomputacao/free-claude-code-plus.git
  cd free-claude-code-plus
  uv tool install .

If `uv tool install .` fails with "no module named X", that's a
missing dev dep — install it with `uv tool install . --with X` and
retry. If it fails with "permission denied" on macOS, ensure
~/.local/bin is on PATH (`echo $PATH | grep -q "$HOME/.local/bin"`
should return true; if not, add `export PATH="$HOME/.local/bin:$PATH"`
to ~/.zshrc and restart the shell).

After install, verify the binaries exist:

  command -v fcc-server             # must print a path
  command -v fcc                    # must print a path

If either is missing, your PATH doesn't include ~/.local/bin — fix
the PATH and retry before continuing.

=========================================================
STEP 3 — Configure provider keys
=========================================================
Ask me ONE question: "Which provider do you want to use as backend?"
Options: MiniMax, OpenRouter, OpenAI, DeepSeek, Mistral, Groq,
Cerebras, Fireworks, ZAI, Kimi, NVIDIA NIM, Ollama (local), or
"skip for now" (you can add keys later).

After I answer, run:

  mkdir -p ~/.fcc
  cp .env.example ~/.fcc/.env
  # Now edit ~/.fcc/.env and set the matching env var to the key I paste.
  # NEVER invent a key — wait for me to paste it.

After I paste the key, restart my shell (or `source ~/.zshrc` /
`source ~/.bashrc`) so the new env var is visible to fcc-server.

=========================================================
STEP 4 — Start the local proxy
=========================================================
The wrapper runs a local HTTP proxy (fcc-server) that Claude Code
talks to. fcc-server has no subcommands — invoking it starts the
proxy directly in foreground. To run it in the background:

  # macOS / Linux — background
  nohup fcc-server > ~/.fcc/logs/server.out 2>&1 &

  # or, foreground (Ctrl+C to stop)
  fcc-server

Verify it's up:

  curl -s -o /dev/null -w "%{http_code}\n" -m 2 http://localhost:8082/
  # expect 404, 405, or 204 (any HTTP response means the proxy is up)

If curl times out, fcc-server didn't start — check the logs (the
proxy writes to ~/.fcc/logs/server.log via loguru; the uvicorn
stdout/stderr goes wherever you redirected it, e.g. server.out) and
tell me what they say.

=========================================================
STEP 5 — Run Claude Code through the wrapper
=========================================================
Now run Claude Code via the wrapper instead of the bare `claude`
binary. This makes Claude Code talk to the fcc-server proxy you
just started, which in turn talks to your chosen provider:

  fcc-claude                        # replaces `claude`
  # or with a specific session:
  fcc-claude --resume <session-id>

If you see "API key not set" or "provider not configured", go back
to STEP 3 — the env var didn't reach fcc-server (shell not reloaded,
or wrong env file).

=========================================================
STEP 6 — Verify end-to-end
=========================================================
In a fresh fcc-claude session, send a short prompt like "say hi".
Then exit the session. Run:

  cat ~/.cache/claude-llm-quota-bar/provider-quota.json 2>/dev/null \
      || echo "quota bar not installed yet — skip this check"

If the quota bar file exists, the wrapper is talking to the
provider correctly. If the file doesn't exist or is older than
5 minutes, something is off — paste me the most recent error from
~/.fcc/logs/server.log.

=========================================================
DONE — Tell me what you did
=========================================================
Summarise in 3-5 bullet points:
- which provider I picked and the matching env var you set
- the install path (output of `command -v fcc-server`)
- whether STEP 4 verification passed (curl status code)
- whether STEP 6 verification passed (quota cache fresh?)
- any caveats or things I should know

If anything failed, give me the exact error message and the command
that produced it. Don't try to fix it silently — surface it.

Prerequisites

Python 3.14+ (uv-managed)
uv 0.11+
An active subscription / API key for at least one provider

Install

git clone https://github.com/philipecomputacao/free-claude-code-plus.git
cd free-claude-code-plus

# Install as a uv tool (creates `fcc-server`, `free-claude-code`,
# `fcc-init`, `fcc-claude`, and `fcc-codex` commands)
uv tool install .

Configure

# Copy the example env and edit
cp .env.example ~/.fcc/.env
$EDITOR ~/.fcc/.env

Add the API keys for the providers you want to enable:

# ~/.fcc/.env

# Required for the server itself
# Defaults are HOST=127.0.0.1, PORT=8082 (localhost-only; safe default).
# Set HOST=0.0.0.0 ONLY if you need the proxy reachable from your LAN.
PORT=8082
# HOST=127.0.0.1

# MiniMax (this fork adds support)
MINIMAX_API_KEY=sk-cp-...

# Other providers (any subset)
NVIDIA_NIM_API_KEY=nvapi-...
OPENROUTER_API_KEY=sk-or-...
DEEPSEEK_API_KEY=sk-...
MISTRAL_API_KEY=ms-...
GROQ_API_KEY=gsk_...
# ...etc

# Model visibility (optional — see "Model visibility control")
# DISABLED_PROVIDERS=open_router,ollama
# HIDDEN_MODELS=deepseek/deepseek-reasoner,minimax/MiniMax-M2

Run

# Start the server in the background
nohup fcc-server > ~/.fcc/logs/server.out 2>&1 &

# Or, foreground (Ctrl+C to stop):
fcc-server

# Check health
curl http://127.0.0.1:8082/health

Wire into Claude Code

Edit ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "http://127.0.0.1:8082",
    "ANTHROPIC_AUTH_TOKEN": "dummy"
  }
}

Now every Claude Code request goes through fcc-server, which fans out to the provider configured for the model id.

Verify MiniMax works

# 1. List models
curl -s http://127.0.0.1:8082/v1/models | jq '.data[].id' | grep minimax

# 2. Send a test message
curl -s -X POST http://127.0.0.1:8082/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: $MINIMAX_API_KEY" \
  -d '{
    "model": "minimax/MiniMax-M3",
    "max_tokens": 100,
    "messages": [{"role": "user", "content": "diz oi em 1 linha"}]
  }' | jq '.content[0].text'

Expected output: "Oi!" (or similar, in Portuguese).

Admin UI

Open http://127.0.0.1:8082/admin in your browser. You'll see the provider catalogue, a button to test each provider, env-var fields, and a model visibility selector to toggle individual LLMs on/off per provider (see Model visibility control). The MiniMax entry was added in this fork.

Supported providers

Provider	Cost	Models	Notes
`minimax` (Token Plan)	Subscription	8	Added in this fork — see below
`nvidia_nim`	Free tier / paid	30+	Upstream default
`opencode` (Zen)	Free / paid	15+	OpenCode's curated catalogue
`opencode_go`	Free / paid	10+	OpenCode Go (faster routing)
`open_router`	Pay-as-you-go	200+	Gateway to every LLM
`deepseek`	Pay-as-you-go	2	Cheapest frontier model
`gemini`	Free tier / paid	3	Google
`groq`	Free tier / paid	4	Ultra-fast inference
`cerebras`	Free tier / paid	3	Wafer-scale silicon
`fireworks`	Pay-as-you-go	5+	Fast open models
`mistral`	Pay-as-you-go	5+	Mistral AI direct
`codestral`	Free for individual	1	Mistral code-specialised
`kimi` (Moonshot)	Pay-as-you-go	3	Chinese frontier
`zai`	Pay-as-you-go	5+	Z.AI / GLM family
`ollama`	Free (local)	100+	Local inference
`llamacpp`	Free (local)	∞	Local via llama.cpp server
`lmstudio`	Free (local)	100+	LM Studio local server
`wafer`	Per-chip	5+	Experimental Cerebras-like

free / paid means the provider has a free tier with rate limits and a paid tier. The Admin UI shows the current balance for paid providers that expose a balance API.

Model visibility control

By default, every model discovered by a configured provider is listed in /v1/models (and therefore visible in Claude Code's /model selector). Two env vars let you filter the list without removing provider keys:

`DISABLED_PROVIDERS`

Comma-separated list of provider IDs to hide entirely from the model listing. The provider remains functional for routing — if another tool or config references a model from a disabled provider, requests still go through. Only the listing is affected.

# Hide all OpenRouter and Ollama models from the model picker
DISABLED_PROVIDERS=open_router,ollama

`HIDDEN_MODELS`

Comma-separated denylist of individual model refs (provider_id/model_name) to hide from the listing. Everything not listed stays visible. Empty = show all.

# Hide specific models you don't use
HIDDEN_MODELS=deepseek/deepseek-reasoner,minimax/MiniMax-M2,minimax/MiniMax-M2.1

Admin UI: model visibility selector

The Admin UI (/admin) includes a Modelos visíveis panel in the Model Config view. It shows all discovered models grouped by provider, with checkboxes to toggle individual models on/off. Features:

Search: filter models by name across all providers
Per-provider toggle: "Marcar todos" / "Desmarcar todos" buttons
Global controls: mark all / clear all / refresh the discovered model list
Provider filter: only configured providers (with valid API keys) show the toggle
Newly configured providers show all models active by default

Changes are saved via the Apply button and take effect immediately on the server. Claude Code clients need to open a new window to pick up the updated model list (the client caches models at ~/.claude/cache/gateway-models.json).

MiniMax provider in depth

Models supported

Model	Context	Notes
`MiniMax-M3`	1M tokens	Flagship, best quality
`MiniMax-M2.7`	128K	Previous-gen, fast
`MiniMax-M2.5`	128K	Stable, well-tested
`MiniMax-M2.1`	128K	Older
`MiniMax-M2`	128K	Original
`MiniMax-M2.1-highspeed`	128K	Lower latency, slightly lower quality
`MiniMax-M2.5-highspeed`	128K	Same
`MiniMax-M2.7-highspeed`	128K	Same

Endpoint

POST https://api.minimax.io/anthropic/v1/messages

Wire format: Anthropic Messages API (system + messages + tools)
Auth header: x-api-key: ${MINIMAX_API_KEY} (not Authorization: Bearer)
SSE streaming: native, same as Anthropic
Model listing: GET https://api.minimax.io/v1/models (OpenAI-compat root)

Token Plan (subscription)

Unlike most providers that charge per-token, MiniMax sells a monthly subscription with 5-hour rolling windows of included usage. The plan includes all 8 models above. When you hit the limit, the API returns 429 until the window resets.

To see your current quota in the statusline, see claude-code-statusline — this fork's sister project adds a minimax quota adapter that calls the Token Plan API.

How the provider is implemented

The provider is a thin subclass of AnthropicMessagesTransport:

# providers/minimax/client.py (excerpt)
from core.transports.anthropic_messages import AnthropicMessagesTransport

class MiniMaxProvider(AnthropicMessagesTransport):
    DEFAULT_BASE = "https://api.minimax.io/anthropic/v1"

    def _auth_headers(self) -> dict[str, str]:
        return {"x-api-key": self._api_key}

    async def list_models(self) -> list[dict]:
        # OpenAI-compat root, not the Anthropic-compat one
        async with httpx.AsyncClient() as client:
            r = await client.get(
                "https://api.minimax.io/v1/models",
                headers={"Authorization": f"Bearer {self._api_key}"},
            )
            r.raise_for_status()
            return r.json()["data"]

Total new code: ~120 LOC (client + __init__.py + registry factory).

Bug fix: missing /v1 suffix

The AnthropicMessagesTransport base class appends /messages to the provider's base URL. If the base is https://api.minimax.io/anthropic, the request goes to https://api.minimax.io/anthropic/messages — which returns HTTP 404 page not found.

The correct path is https://api.minimax.io/anthropic/v1/messages. The fix is one character — change MINIMAX_DEFAULT_BASE from https://api.minimax.io/anthropic to https://api.minimax.io/anthropic/v1.

This is not documented in the MiniMax API docs. We discovered it by trial-and-error:

# Wrong (404):
$ curl -i -X POST https://api.minimax.io/anthropic/messages \
    -H "x-api-key: sk-dummy" -d '{}'
HTTP/2 404
not found

# Right (401 = path is valid, just auth failed):
$ curl -i -X POST https://api.minimax.io/anthropic/v1/messages \
    -H "x-api-key: sk-dummy" -d '{}'
HTTP/2 401
missing or invalid x-api-key

If you're integrating a new provider, always probe the path first. Many "Anthropic-compat" providers serve on a non-standard path (Kimi, Z.AI, MiniMax all need /v1).

This fix is in commit a2218a8 of this repo and is already upstreamed — see Contributing back.

Architecture

┌──────────────────────────────────────────────────────────────────┐
│  fcc-server (FastAPI)                                            │
│                                                                  │
│  ┌─────────────────┐  ┌──────────────────┐  ┌────────────────┐  │
│  │  /v1/messages   │  │  /admin/api/*    │  │  /v1/models    │  │
│  │  (Anthropic-    │  │  (Admin UI       │  │  (OpenAI-      │  │
│  │   compat proxy) │  │   JSON API)      │  │   compat list) │  │
│  └────────┬────────┘  └────────┬─────────┘  └────────┬───────┘  │
│           │                    │                     │          │
│           └────────┬───────────┴─────────────────────┘          │
│                    ▼                                             │
│           ┌─────────────────┐                                    │
│           │  ModelRouter    │ parses model id ("minimax/...")   │
│           │                 │ and dispatches to the right        │
│           │                 │ provider factory                   │
│           └────────┬────────┘                                    │
│                    │                                             │
│       ┌────────────┴────────────┐                                │
│       ▼            ▼            ▼                                │
│  ┌─────────┐ ┌──────────┐ ┌──────────┐                           │
│  │ minimax │ │  open_   │ │  nvidia  │  18 providers total        │
│  │ (NEW)   │ │  router  │ │   _nim   │                           │
│  └────┬────┘ └─────┬────┘ └────┬─────┘                           │
│       │            │           │                                 │
│       └────────────┴───────────┘                                 │
│                    ▼                                             │
│           ┌─────────────────┐                                    │
│           │  ProviderClient │ retry, rate-limit, stream-SSE      │
│           │  + QuotaTracker │ optional live quota check          │
│           └────────┬────────┘                                    │
└────────────────────┼─────────────────────────────────────────────┘
                     ▼
              Upstream LLM provider

Code layout

.
├── api/                    # FastAPI routes, Admin UI
│   ├── admin_routes.py     # /admin/api/* JSON endpoints
│   ├── admin_static/       # HTML/JS for the Admin UI
│   ├── routes.py           # /v1/messages, /v1/models
│   ├── model_router.py     # model id → provider dispatch
│   ├── model_catalog.py    # /v1/models response (applies denylist + disabled filters)
│   └── gateway_model_ids.py # gateway ID rewriting
├── cli/                    # `fcc` command, `fcc-server` launcher
├── config/                 # provider_catalog.py, settings
├── core/                   # cross-cutting (transports, retry, rate-limit)
├── providers/              # 18 provider implementations
│   └── minimax/            # ← this fork adds this
│       ├── __init__.py
│       └── client.py
├── messaging/              # Telegram + Discord bots (optional)
├── server.py               # FastAPI app factory
├── pyproject.toml          # 2.5.0 (model visibility + denylist)
└── smoke/                  # integration tests

Development

# Clone
git clone https://github.com/philipecomputacao/free-claude-code-plus.git
cd free-claude-code-plus

# Install in editable mode
uv tool install --editable .

# Run the server in dev mode (auto-reload)
fcc-server --reload

# Run smoke tests
pytest smoke/

# Type-check
basedpyright providers/ api/ core/

Adding a new provider

Create providers/<name>/{__init__.py,client.py} with a subclass of AnthropicMessagesTransport (for Anthropic-compat) or OpenAIChatCompletionsTransport (for OpenAI-compat).

Add a factory in providers/registry.py:

def _create_your_provider(settings: Settings) -> BaseProvider:
    return YourProvider(
        api_key=settings.your_provider_api_key,
        base_url=YOUR_PROVIDER_DEFAULT_BASE,
    )
PROVIDER_FACTORIES["your_provider"] = _create_your_provider

Add the constant in config/provider_catalog.py:

YOUR_PROVIDER_DEFAULT_BASE = "https://api.your-provider.com/v1"
PROVIDER_CATALOG["your_provider"] = {
    "default_base": YOUR_PROVIDER_DEFAULT_BASE,
    "auth_header": "Authorization",
    "auth_prefix": "Bearer ",
    "env_key": "YOUR_PROVIDER_API_KEY",
    "display_name": "Your Provider",
    "tier": "qwen-medium",
}

Add the env-var to Settings and .env.example.
Write a smoke test in smoke/prereq/test_provider_prereq_live.py following the existing pattern for cerebras or groq.
(Optional) Add a quota adapter in claude-code-statusline so the statusline shows live quota.

Running the Admin UI in dev

cd api/admin_static
python3 -m http.server 5173   # serves the static files
# Then point fcc-server's static mount at it

Syncing with upstream

This fork tracks Alishahryar1/free-claude-code:main and rebases the feat/provider-minimax branch on top. A weekly GitHub Action (.github/workflows/upstream-sync.yml) detects divergence and opens a tracking issue; nothing is auto-merged.

To sync manually:

cd ~/free-claude-code-plus

# Fetch upstream
git fetch upstream

# Update main from upstream
git checkout main
git merge upstream/main --ff-only  # or use a merge commit
git push origin main

# Rebase the feature branch on the new main
git checkout feat/provider-minimax
git rebase main
git push origin feat/provider-minimax --force-with-lease

# Run CI to verify
./scripts/ci.sh

Conflict hot-spots when rebasing:

providers/registry.py (new provider factories)
config/provider_catalog.py (PROVIDER_CATALOG dict)
pyproject.toml (version bumps)
uv.lock (regenerate with uv lock)

Contributing back to upstream

The MiniMax provider and the /v1 base-URL bug fix are intended to be upstreamed. To prepare a PR:

Squash the fork-specific commits into a single feature commit:

git checkout feat/provider-minimax
git rebase -i main
# Squash: 8890e7a (provider) + a2218a8 (bug fix) into one

Add a smoke test in smoke/prereq/test_provider_prereq_live.py — the upstream maintainer will require this.
Update pyproject.toml — keep the version bump, but the upstream might prefer a different version scheme.
Open a PR against Alishahryar1/free-claude-code:main with:
- Title: feat(minimax): add native MiniMax Token Plan provider
- Description: link to this fork's feat/provider-minimax branch for the full diff history
Update the upstream CLAUDE.md / AGENTS.md with MiniMax's specific gotchas (the /v1 bug is the most important).

Related projects

Alishahryar1/free-claude-code — the original FCC project that this fork builds on. Most provider code comes from here.
claude-code-statusline — the multi-provider statusline bar used with this fork. The minimax quota adapter was added in this fork's sister project.
opencode-llm-statusline — same statusline bar, but usable from OpenCode (not just Claude Code).

License

MIT — same as upstream. See LICENSE.

Fork mantido por @lpdigital.me.

Name		Name	Last commit message	Last commit date
Latest commit History 20 Commits
.github		.github
api		api
assets		assets
cli		cli
config		config
core		core
messaging		messaging
providers		providers
scripts		scripts
smoke		smoke
tests		tests
.env.example		.env.example
.gitignore		.gitignore
.python-version		.python-version
AGENTS.md		AGENTS.md
ARCHITECTURE.md		ARCHITECTURE.md
CLAUDE.md		CLAUDE.md
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
FORK_NOTES.md		FORK_NOTES.md
LICENSE		LICENSE
README.md		README.md
pyproject.toml		pyproject.toml
server.py		server.py
uv.lock		uv.lock

Folders and files

Latest commit

History

Repository files navigation