amplifier-app-opencode

Launch the opencode TUI on top of amplifier-agent with one command:

amplifier-opencode

This binary discovers what models your local amplifier-agent serves, writes a working opencode config from that discovery, and launches opencode — every time, no manual config to maintain.

What this gives you

opencode is a fast terminal coding assistant. amplifier-agent is Microsoft's modular agent framework with a multi-provider OpenAI-compatible HTTP face.

Without this adapter, integrating them requires you to manually mirror amplifier-agent's /v1/models output into your opencode config and re-edit it whenever your provider mix changes.

With this adapter:

One command (amplifier-opencode) handles the whole flow
The model list in opencode's picker is always live — re-discovered on every launch
amplifier-agent's server is auto-started in the background if it isn't running
Drop-in opencode TUI: /models, /connect, slash commands all work normally
Includes a doctor subcommand that diagnoses any setup issue before you ask

Prerequisites

You need a few system tools and three Amplifier components installed. This README walks through the official install for each so a brand-new machine can get set up start-to-finish.

0. System tools — git, curl

Most macOS installs already have these via Xcode Command Line Tools. On a fresh Linux container you'll need:

# Ubuntu/Debian
sudo apt-get update && sudo apt-get install -y git curl

# Fedora/RHEL
sudo dnf install -y git curl

# Arch
sudo pacman -S --noconfirm git curl

Why: git is required because amplifier-agent and amplifier-app-opencode are installed via git+https://... URLs (neither is on PyPI yet). curl is required by the uv and opencode one-line installers.

1. amplifier-agent — the backend server (>= 0.8.0 required)

amplifier-agent is the OpenAI-compatible HTTP server this adapter talks to. Use the official one-line installer — it pulls the latest released binary and primes the bundle cache so the first run is instant:

curl -fsSL https://raw.githubusercontent.com/microsoft/amplifier-agent/main/install.sh | bash

# to pin a specific version instead of latest:
#   curl -fsSL https://raw.githubusercontent.com/microsoft/amplifier-agent/main/install.sh | bash -s -- --tag v0.9.0

# ensure ~/.local/bin is on PATH, then verify:
amplifier-agent version --json   # → {"version":"0.9.0","protocolVersion":"0.3.0"}

The installer needs uv and curl on PATH and will tell you exactly what to install if either is missing — it will not bootstrap them silently.

Version requirement: amplifier-agent >= 0.8.0 is mandatory. Versions below 0.8.0 do not ship the serve chat-completions HTTP face, multi-provider routing, or the auth subcommand — amplifier-opencode will fail to spawn the server, populate /v1/models, or read persisted credentials.

If the version reported is older than 0.8.0, upgrade in place with amplifier-agent update and re-run amplifier-opencode doctor to confirm.

For full install options (source builds, manual uv tool install --from git+…, installer flags) see the amplifier-agent README.

2. opencode — the TUI

The official one-line installer downloads the platform-native opencode binary:

curl -fsSL https://opencode.ai/install | bash

It places opencode in ~/.opencode/bin/ and appends that directory to your shell PATH by writing an export line into ~/.bashrc or ~/.zshrc.

Open a new terminal (or source ~/.bashrc / source ~/.zshrc) and verify:

opencode --version

Heads-up if you're running this in a container, headless server, or any non-interactive shell: the opencode installer only updates your shell's rc file. Non-interactive shells (systemd services, container exec scripts, sub-shells launched by other tools) do NOT source ~/.bashrc or ~/.zshrc, so they won't see opencode on PATH. In those environments, add ~/.opencode/bin to PATH explicitly — for example:
export PATH="$HOME/.opencode/bin:$PATH"
Or place that export in /etc/profile.d/opencode.sh for system-wide coverage. amplifier-opencode doctor will flag this with a clear error if opencode isn't on PATH when it runs.

For other install methods (Homebrew, manual download, package managers) see opencode.ai/docs/intro.

3. amplifier-app-opencode — this adapter

Same uv-tool pattern as amplifier-agent:

uv tool install --from git+https://github.com/microsoft/amplifier-app-opencode amplifier-app-opencode

# Verify
amplifier-opencode --help

(Once published to PyPI, this becomes uv tool install amplifier-app-opencode.)

4. At least one provider credential

amplifier-agent talks to upstream model APIs (Anthropic, OpenAI, etc.) and needs credentials for at least one of them. Easiest: set an environment variable that amplifier-agent already knows about:

# Pick at least ONE that you have access to:
export ANTHROPIC_API_KEY="sk-ant-..."
export OPENAI_API_KEY="sk-..."
export AZURE_OPENAI_API_KEY="..."
export OLLAMA_HOST="http://localhost:11434"   # if running ollama locally

# To persist across all terminals, add the line to ~/.zshrc (or your shell's rc)

Alternative: use amplifier-agent's persistent credential file. Run once and your key is stored under ~/.amplifier-agent/credentials.json (mode 0600), available to every future invocation from any directory:

amplifier-agent auth set anthropic sk-ant-...
amplifier-agent auth list                       # confirm it's stored

The amplifier-agent server uses env-first resolution: shell env vars win over the credentials file, so existing shell-rc workflows continue working unchanged.

Provider Credentials

amplifier-opencode auto-detects which AI provider you have credentials for and configures amplifier-agent to serve those providers. Set one or more of the following environment variables before running amplifier-opencode launch:

Provider	Env var
Anthropic (Claude)	`ANTHROPIC_API_KEY`
OpenAI (GPT)	`OPENAI_API_KEY`
Azure OpenAI	`AZURE_OPENAI_API_KEY` or `AZURE_OPENAI_KEY`
Ollama (local models)	`OLLAMA_HOST`

Run amplifier-opencode doctor to see which providers will be available.

Advanced: custom host_config.json

If you want fine-grained control (custom MCP servers, approval policies, per-provider config overrides), write your own host_config.json and pass it with --host-config:

amplifier-opencode launch --host-config /path/to/host_config.json

See amplifier-agent's host_config documentation for the full schema.

First run

amplifier-opencode

You'll see something like:

[1/4] Starting amplifier-agent (port 9099, workspace='opencode')
      amplifier-agent ready at http://127.0.0.1:9099/v1
[2/4] Discovering models via GET http://127.0.0.1:9099/v1/models
      - anthropic   claude-haiku-4-5-20251001            Claude Haiku 4.5
      - anthropic   claude-opus-4-8                      Claude Opus 4.8
      - anthropic   claude-sonnet-4-6                    Claude Sonnet 4.6
[3/4] Wrote /Users/you/.config/opencode/opencode.jsonc  (global config)
[4/4] Configuration complete.

✓ opencode is configured to use Amplifier. Pick how you want to drive it:

    TUI       run `opencode` in any directory
    Desktop   open the opencode desktop app — it picks up the global config
    Headless  opencode run "your prompt here"

To jump straight into the TUI next time: amplifier-opencode launch

The bridge is now set up. Run opencode in any terminal and pick a model under the Amplifier section — or open the opencode desktop app. The global config applies everywhere.

If you'd rather have one command that sets up the bridge AND drops you into the TUI, use amplifier-opencode launch instead.

Daily usage

Refresh the bridge

amplifier-opencode

Default action: check the server, start it if needed, re-discover models, and update opencode.jsonc. Doesn't launch anything — just keeps the bridge in sync with whatever amplifier-agent is currently serving. Run this any time you change providers, add credentials, or restart amplifier-agent.

Set up the bridge AND launch the TUI

amplifier-opencode launch

Same setup flow plus an exec into the opencode TUI. From any directory.

Pass arguments through to opencode

Anything after -- is forwarded to opencode:

amplifier-opencode launch -- run "summarise this codebase"

Write to a project's opencode.json instead of the global config

cd my-project
amplifier-opencode --project-dir .          # bridge only, into ./opencode.json
amplifier-opencode launch --project-dir .   # bridge + TUI from this directory

The generated config lives in ./opencode.json. opencode walks up from cwd to find it.

Point at a different amplifier-agent

# Server running on another machine or port
amplifier-opencode --base-url http://my-server:9099/v1 --api-key my-token

# Server already running — don't auto-start
amplifier-opencode --no-start

Troubleshooting

Run the doctor before you ask anyone for help. It reports on every prerequisite in one shot:

amplifier-opencode doctor

Output:

amplifier-opencode doctor

  [ OK ]  amplifier-agent     amplifier-agent found at /Users/you/.local/bin/amplifier-agent
  [ OK ]  opencode            opencode found at /Users/you/.opencode/bin/opencode
  [ OK ]  server              amplifier-agent server running at http://127.0.0.1:9099/v1
  [ OK ]  credentials         Credentials available (env: anthropic=ANTHROPIC_API_KEY)
  [ OK ]  opencode config     opencode config has provider.amplifier with 3 models
  [ OK ]  live models         Discovered 3 model(s): claude-haiku-4-5-20251001, ...

All required checks passed.

Exit code is 0 when everything passes, 1 if any FAIL. Suitable for CI scripts.

Common FAILs and their fix:

FAIL message	Fix
`amplifier-agent not on PATH`	Re-run the install one-liner above, then open a new terminal
`opencode not on PATH`	`curl -fsSL https://opencode.ai/install \| bash` (then open a new terminal)
`No provider credentials found`	Export `ANTHROPIC_API_KEY`, OR run `amplifier-agent auth set anthropic <key>`
`opencode config ... is malformed JSON`	Open `~/.config/opencode/opencode.jsonc`, fix or delete it, retry

CLI reference

amplifier-opencode [GLOBAL OPTIONS] [SUBCOMMAND] [SUBCOMMAND OPTIONS]

Global options

Flag	Env var	Default	Purpose
`--base-url`	`AMPLIFIER_AGENT_BASE_URL`	`http://127.0.0.1:9099/v1`	amplifier-agent endpoint
`--api-key`	`AMPLIFIER_AGENT_API_KEY`	`local-dev-secret`	wire-level bearer token

`launch` (default)

Discover models, write opencode.json, exec opencode. Run when no subcommand is given.

Flag	Env var	Default	Purpose
`--workspace`	`AMPLIFIER_AGENT_WORKSPACE`	`opencode`	Server-side workspace name (only used when starting the server)
`--host-config`	`AMPLIFIER_AGENT_HOST_CONFIG`	—	Path to a host_config.json to use verbatim; overrides auto-generated config (only used when starting the server)
`--project-dir`	—	(use global)	Write opencode.json into this directory instead of global config
`--no-start`	—	false	Don't auto-start amplifier-agent; require server already up
`--no-launch`	—	false	Don't exec opencode; just write the config
`--amplifier-agent-bin`	`AMPLIFIER_AGENT_BIN`	autodetect	Override the amplifier-agent binary path
`--provider-id`	—	`amplifier`	Provider ID under `provider.<id>`
`OPENCODE_ARGS...` (after `--`)	—	—	Pass-through to opencode

`doctor`

Run all prerequisite checks. No flags; honours the global --base-url and --api-key.

Where things live

Path	What
`<project-dir>/opencode.json` OR `~/.config/opencode/opencode.jsonc`	Generated config (overwritten on every launch)
`/tmp/amplifier-agent.log`	stdout/stderr from the spawned amplifier-agent process
`/tmp/amplifier-opencode-agent.pid`	PID of the spawned amplifier-agent
`~/.amplifier-agent/state/workspaces/`	amplifier-agent's session storage
`~/.amplifier-agent/credentials.json`	Persistent credentials (mode 0600) — set via `amplifier-agent auth`
`~/.local/share/opencode/log/opencode.log`	opencode's own log file

Design notes

opencode's documented contract for custom OpenAI-compatible providers (the "Atomic Chat" pattern) requires a static models block listing every model ID the upstream server serves. opencode does not auto-fetch /v1/models at runtime for custom-config providers.

Rather than maintain the model list by hand or attempt fragile runtime monkey-patches against opencode's internals, this binary materialises the static config from the live /v1/models response before opencode launches. Every invocation re-syncs. The model list is therefore always live; the "static" nature of opencode's config is just an implementation detail.

opencode itself is unmodified. The adapter lives entirely outside both upstream projects — no plugins, no patches, no npm packages, no JavaScript.

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.

License

MIT

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

amplifier-app-opencode

What this gives you

Prerequisites

0. System tools — git, curl

1. amplifier-agent — the backend server (>= 0.8.0 required)

2. opencode — the TUI

3. amplifier-app-opencode — this adapter

4. At least one provider credential

Provider Credentials

Advanced: custom host_config.json

First run

Daily usage

Refresh the bridge

Set up the bridge AND launch the TUI

Pass arguments through to opencode

Write to a project's opencode.json instead of the global config

Point at a different amplifier-agent

Troubleshooting

CLI reference

Global options

`launch` (default)

`doctor`

Where things live

Design notes

Contributing

Trademarks

License

Uh oh!

FilesExpand file tree

README.md

Latest commit

History

README.md

File metadata and controls

amplifier-app-opencode

What this gives you

Prerequisites

0. System tools — git, curl

1. amplifier-agent — the backend server (>= 0.8.0 required)

2. opencode — the TUI

3. amplifier-app-opencode — this adapter

4. At least one provider credential

Provider Credentials

Advanced: custom host_config.json

First run

Daily usage

Refresh the bridge

Set up the bridge AND launch the TUI

Pass arguments through to opencode

Write to a project's opencode.json instead of the global config

Point at a different amplifier-agent

Troubleshooting

CLI reference

Global options

launch (default)

doctor

Where things live

Design notes

Contributing

Trademarks

License

`launch` (default)

`doctor`