opencode-claude-auth

Self-contained Anthropic auth provider for OpenCode using your Claude Code credentials — no separate login or API key needed.

How it works

The plugin registers its own auth provider with a custom fetch handler that intercepts all Anthropic API requests. It reads OAuth tokens from the macOS Keychain (or ~/.claude/.credentials.json — or $CLAUDE_CONFIG_DIR/.credentials.json if that env var is set — on other platforms), caches them in memory with a 30-second TTL, and handles the full request lifecycle — no builtin Anthropic auth plugin required. On macOS, multiple Claude Code accounts are detected automatically and can be switched via opencode auth login.

It also syncs credentials to OpenCode's auth.json as a fallback (on Windows, it writes to both %USERPROFILE%\.local\share\opencode\auth.json and %LOCALAPPDATA%\opencode\auth.json to cover all installation methods). If a token is near expiry, it refreshes directly via Anthropic's OAuth endpoint (zero LLM tokens consumed), falling back to the Claude CLI if the direct refresh fails. Background re-sync runs every 5 minutes.

Prerequisites

• Claude Code installed and authenticated (run claude at least once)
• OpenCode installed

macOS is preferred (uses Keychain). Linux and Windows work via the credentials file fallback.

Installation

For Humans

Option A: Let an LLM do it

Paste this into any LLM agent (Claude Code, OpenCode, Cursor, etc.):

Install the opencode-claude-auth plugin and configure it by following: https://raw.githubusercontent.com/griffinmartin/opencode-claude-auth/main/installation.md

Option B: Manual setup

1. Add the plugin to ~/.config/opencode/opencode.json:

   {
     "plugin": ["opencode-claude-auth@latest"]
   }
   

   The @latest tag ensures OpenCode always pulls the newest version on startup. No manual npm install is needed — OpenCode automatically installs npm plugins using Bun at startup.

2. Use it — just run OpenCode. The plugin handles auth automatically using your Claude Code credentials.

For LLM Agents

See installation.md for step-by-step agent instructions.

Usage

Just run OpenCode. The plugin handles auth automatically — it reads your Claude Code credentials, provides them to the Anthropic API, and refreshes them in the background. If your credentials aren't OAuth-based, the plugin falls through to standard API key auth.

Supported models

13 supported models. Run pnpm run test:models to verify against your account.

Credential sources

The plugin checks these in order:

1. macOS Keychain (all Claude Code-credentials* entries — multiple accounts are detected automatically)
2. ~/.claude/.credentials.json (fallback, works on all platforms; if CLAUDE_CONFIG_DIR is set, reads $CLAUDE_CONFIG_DIR/.credentials.json instead)

Multiple accounts (macOS)

If you have multiple Claude Code accounts authenticated on macOS, the plugin detects all of them from the Keychain automatically. Each account is labeled by its subscription tier (Claude Pro, Claude Max, etc.).

To switch accounts:

opencode auth login

Select "Switch Claude Code account" and pick the account you want to use. Your selection is persisted across sessions.

If only one account is found, the switcher is hidden and the plugin uses it directly.

Troubleshooting

Diagnostic logging

If you're hitting auth errors that are hard to reproduce, enable debug logging to capture the full auth flow:

export CLAUDE_AUTH_DEBUG=1

Restart OpenCode and reproduce the issue. The plugin writes structured JSON logs to ~/.local/share/opencode/claude-auth-debug.log. All secrets (tokens, API keys) are automatically redacted — the log file is safe to paste into a GitHub issue.

To write logs to a custom path:

export CLAUDE_AUTH_DEBUG=/tmp/claude-auth-debug.log

Disable when done:

unset CLAUDE_AUTH_DEBUG

Long context (1M)

1M token context is supported natively — the API no longer requires a beta flag for it, so the plugin doesn't send the legacy context-1m-2025-08-07 header.

If your plan doesn't cover long context billing, requests beyond the standard window fail with "Extra usage is required for long context requests". When a long context error is caused by a beta flag (e.g. one added via ANTHROPIC_BETA_FLAGS), the plugin retries without the offending flag.

Validating OAuth refresh

To verify the direct OAuth token refresh works with your credentials:

pnpm run validate:oauth           # refresh + write-back (safe, keeps credentials valid)
pnpm run validate:oauth -- --dry-run  # show what would be sent without making the request

This reads your stored credentials, calls Anthropic's OAuth token endpoint, and writes the new tokens back to storage. Refresh tokens rotate on each use, so write-back is enabled by default to keep your stored credentials valid.

Environment variable overrides

All configurable parameters can be overridden via environment variables. If Anthropic changes something before we publish an update, set an env var and keep working:

Example:

export ANTHROPIC_CLI_VERSION=2.2.0

How it works (technical)

• Registers an auth.loader with a custom fetch that intercepts all Anthropic API requests
• Sets Authorization: Bearer with fresh OAuth tokens (cached in memory, 30s TTL, updated in-place after refresh)
• Translates tool names between OpenCode and Anthropic API formats (adds/strips mcp_ prefix)
• Buffers SSE response streams at event boundaries for reliable tool name translation
• Injects Claude Code identity into system prompts via experimental.chat.system.transform
• Sets required API headers (beta flags, billing, user-agent) with model-aware selection
• On macOS, enumerates all Claude Code-credentials* Keychain entries and labels them by subscription tier
• Provides an account switcher via opencode auth login when multiple accounts are found; persists selection to ~/.local/share/opencode/claude-account-source.txt
• Syncs credentials to auth.json on startup and every 5 minutes as a fallback; that same tick proactively refreshes once the token is within an hour of expiry
• On Windows, writes to both %USERPROFILE%\.local\share\opencode\auth.json and %LOCALAPPDATA%\opencode\auth.json
• Re-reads the credential source on every cache miss, so an account rotated by something other than this plugin — the claude CLI in another terminal, a second OpenCode instance, or a switcher like claude-swap — gets picked up mid-session without a restart. Bounded by the same 30s cache, so it adds at most about two source reads a minute under load. A stored token is adopted whenever it is usable, and when it isn't only if the one already held is also unusable — otherwise a failed write-back would resurrect the pre-refresh token it left behind
• Guards credential write-back with the access token the refresh started from, so a switch landing mid-refresh can't write one account's rotated tokens into another account's slot
• Retries API requests on 429 (rate limit) and 529 (overloaded) with exponential backoff, respecting retry-after headers
• On a 429 that outlives those backoff retries, re-reads the source once and retries only if the access token changed, so a rate limit another process has already resolved by switching accounts isn't surfaced. A changed token isn't proof of a switch — a routine refresh of the same account changes it too — so this costs at most one extra request
• On a 401, recovers in place rather than surfacing it: adopts an externally rotated token if the source now holds one, otherwise forces an OAuth refresh, then retries the request. Bounded at two attempts, so a rejected token costs at most three API calls. A 401 that survives recovery is returned unmodified, without SSE stream transformation, since it carries an error body rather than a stream
• Refreshes directly via POST https://claude.ai/v1/oauth/token using the runtime's own fetch (no LLM tokens consumed, no subprocess). Requests are triggered within 60 seconds of expiry on the API request path and within an hour on the background tick; concurrent refreshes of one account share a single request, since each rotation invalidates the previous refresh token
• Falls back to the claude CLI only within the 60-second window, the point at which Claude Code will actually rotate the token — running it earlier costs a real API request and returns the same token. New tokens are written back to Keychain (macOS) or credentials file (Linux/Windows) to keep stored credentials in sync with rotated refresh tokens
• If credentials aren't OAuth-based, the auth loader returns {} and falls through to API key auth
• If credentials are unavailable or unreadable, the plugin disables itself and OpenCode continues without Claude auth

Disclaimer

This plugin uses Claude Code's OAuth credentials to authenticate with Anthropic's API. Anthropic's Terms of Service state that Claude Pro/Max subscription tokens should only be used with official Anthropic clients. This plugin exists as a community workaround and may stop working if Anthropic changes their OAuth infrastructure. Use at your own discretion.

License

MIT