* fix(proxy): reuse upstream connections for OpenCode API requests
`createProxyMiddleware` was constructed without an `agent`, so `http-proxy`
fell back to `agent: false`. That disables connection pooling and forces
`Connection: close` on every proxied request, consuming one ephemeral port
per request.
Measured against a real `opencode serve` instance, 200 sequential requests
through the proxy created 201 TIME_WAIT entries (1.005 ports/request). With
a keep-alive agent the same load creates 0.
On macOS the ephemeral range is 16,384 ports and TIME_WAIT lasts 30s, so
sustained traffic around 546 req/sec exhausts the pool — after which every
process on the host fails to open outbound connections with EADDRNOTAVAIL.
`maxSockets: Infinity` preserves the unbounded concurrency of `agent: false`,
so this changes connection reuse only, not request throughput.
Partially addresses #2915.
* fix(proxy): derive proxy agent class from the target scheme
Addresses review feedback on #2916. The first commit created an
unconditional `http.Agent`, which regresses external OpenCode servers
configured over https via `OPENCODE_HOST` (accepted by env-config.js).
http-proxy dispatches through `https.request` when the target protocol is
`https:` (http-proxy/lib/http-proxy/passes/web-incoming.js:126), and
`http.Agent#createConnection` is plain `net.createConnection` — so an
http.Agent would open a plaintext socket to a TLS port and fail every
proxied request. `agent: false` previously worked for both schemes.
`createOpenCodeProxyAgent(target)` now returns an `https.Agent` for https
targets and an `http.Agent` otherwise, derived once from
`resolveProxyTarget()` at registration so the single shared instance is
preserved across `apiProxy` and `interactiveOAuthProxy`.
Guarded in both test layers, verified to fail when the selection is
reverted to an unconditional http.Agent. `https.Agent` extends
`http.Agent`, so the http cases assert `not.toBeInstanceOf(https.Agent)`.
* Round 2: fix: resolve the proxy agent lazily so cold starts honor https
Addresses the round-2 blocker on #2916. Deriving the agent class at
registration is too early: startup-pipeline-runtime.js calls setupProxy()
(line 104) before bootstrapOpenCodeAtStartup() (line 141), so on a fresh
process state.openCodePort is null, buildOpenCodeUrl() throws
(network-runtime.js:86-88), and resolveProxyTarget() returns the http
loopback fallback. An external server configured via OPENCODE_HOST=https://
only appears on state.openCodeBaseUrl after bootstrap, so it was still
getting a plain http.Agent — the regression the previous commit intended
to fix.
`agent` is now a getter backed by a per-scheme memoizing resolver.
http-proxy-middleware rebuilds per-request options with
`Object.assign({}, this.proxyOptions)` in prepareProxyRequest, which invokes
getters, so resolution happens at request time while still yielding one
shared pool per scheme.
Tests now model the production ordering — registration while the port is
null and buildOpenCodeUrl throws, then an https base URL appearing after
bootstrap — and fail against the eager implementation. A behavioral test
pins the http-proxy-middleware option re-read the fix depends on, so a
library change that froze options would fail loudly instead of silently
regressing https targets.
The resolver is module-private; `bun run dead-code` flagged it as an
unused export when it was exported.
* Round 3: docs(changelog): note upstream connection reuse under [Unreleased]
Repo precedent adds [Unreleased] bullets for comparable proxy/stability
fixes (1.18.4 Stability, 1.9.3 Reliability/Proxy). Non-blocker raised in
review on #2916.
* Round 3: docs(changelog): use repo-standard 'behavior' spelling
* Round 4: docs(changelog): don't imply a restart is the only recovery
The ephemeral port pool drains on its own once the exhausting traffic
stops (TIME_WAIT expiry), so a restart is sufficient but not necessary.
Optional nit raised in review on #2916.
* Round 5: fix: construct the proxy agent through one factory; widen the pool
Review found the https branch was mutation-uncovered: the resolver
re-implemented agent construction inline instead of calling the exported
`createOpenCodeProxyAgent(target)`, so replacing its https branch with
`new https.Agent()` — dropping OPENCODE_AGENT_OPTIONS, and with it
keep-alive — left the entire suite green. Since `createOpenCodeProxyAgent`
also had no production callers, its four tests were pinning dead code.
Delegating collapses both: the factory is now the single construction
path, and the mutation fails 2 tests including the live resolver path.
Also from review:
- maxFreeSockets 32 -> 256 (Node's own default). The lower cap evicted
pooled sockets under concurrency, reintroducing the churn this agent
exists to prevent: at 64 concurrent requests it left 303 sockets in
TIME_WAIT versus 0 at 256.
- Added `timeout` to OPENCODE_AGENT_OPTIONS. Free-socket eviction is
governed by agent.options.timeout, which was unset, so idle sockets
persisted until the peer closed them. `keepAliveMsecs` is the TCP probe
delay, not the idle lifetime.
- resolveProxyTarget() now checks openCodePort before calling
buildOpenCodeUrl instead of relying on it throwing. The port is nulled
on several runtime paths (health-check failure, failed restart), so a
degraded OpenCode made every proxied request pay for a thrown-and-caught
exception — and the getter added a second call per request.
- Test fixtures use :4096 rather than :443; WHATWG URL elides the default
port, so parseInt('') is NaN and env-config rejects that host. The
fixtures modeled a state that cannot reach production.
- The getter-read assertion is now exact (0 at construction, 1, then 2)
rather than >= 2, which would have passed if the getter were read twice
at construction and never per-request.
- listen() rejects on 'error' and servers start inside try/finally, so a
bind failure fails the test instead of hanging to timeout.
OpenChamber is an open-source workspace for running, supervising, and reviewing AI coding work across desktop, browser, editor, and mobile.
OpenChamber gives you one place to direct agent work, understand the changes, and move them toward release. Your projects stay available when you switch devices or step away.
More screenshots
What you can do with OpenChamber
Goals that continue on their own
Give a session a finish line with Session Goals. OpenChamber checks the result after every turn and keeps the agent working until the goal is complete, blocked, or reaches the limit you set — even after you close the app.
Compare and combine runs
Use Multi-run to give the same task to up to five models, each in its own session and optionally its own worktree. See what each one actually built, choose the best result, or use Fusion to combine the strongest parts into a new session.
Guided changes walkthroughs
Changes Walkthrough turns a large diff into an AI-guided tour of the change. It groups related edits into steps, puts them in the order the change makes sense, and explains how the pieces fit together.
Inspect a running app
Open your app beside the conversation with Preview. Point at an element and send the agent its screenshot, styles, position, and browser errors — all the context behind “this thing here.” Desktop brings the same workflow to any web page through its built-in browser.
GitHub context from issue to pull request
Start a session from a GitHub issue or pull request with its context attached. Send failed checks or review comments back to the agent, then update or merge the pull request from OpenChamber.
Continue on another device
Open the same projects and sessions from Desktop, Web/PWA, VS Code, iOS, or Android. Check progress, answer questions, review changes, and reattach to a running terminal.
Private remote access
Pair a device with a one-time QR code and connect through Private Relay without opening ports or exposing a public server. The connection is end-to-end encrypted and can be revoked at any time. Direct connections, LAN/VPN access, Cloudflare/Ngrok tunnels, and SSH are also supported.
Track work across projects
See which sessions are working, waiting, finished, or failed, along with approvals, scheduled tasks, provider limits, token use, and costs. Organize sessions into folders and keep notes, todos, and reusable project actions nearby.
Schedule recurring work
Run a prompt once, daily, weekly, or on a cron schedule. Scheduled tasks can use Session Goals, so they continue toward an outcome instead of stopping after one response.
Use it where you work
Surface
Role
Desktop
The complete workspace for macOS, Windows, and Linux, with multiple windows, Mini Chat, remote machines, SSH, and native notifications
Web / PWA
Open your workspace in a browser, install it as an app, and stay up to date through background notifications
VS Code
Keep sessions beside your code, send selections to the agent, open results in the editor, and compare parallel runs
iOS / Android
Review and steer work away from your desk, receive completion alerts, and use the terminal with touch controls
CLI / Server
Run OpenChamber on a workstation or server, schedule work, manage remote access, and keep it available after login
Quick start
Desktop — macOS, Windows, and Linux
Download the latest release from GitHub Releases. Desktop bundles the matching OpenCode CLI, so no separate OpenCode installation is required.
Linux releases are available as x86_64 and ARM64 AppImages. Make the downloaded AppImage executable and keep it in a writable location for in-app updates:
OpenChamber uses OpenCode to power its coding agents. We chose it because we believe it provides the best open-source agentic coding experience today: capable, extensible, and open by design.
Around that foundation, OpenChamber brings together the work that happens before, during, and after an agent run — deciding what to try, keeping it on track, reviewing the result, connecting from anywhere, and getting the change shipped.
OpenChamber is an independent project and is not affiliated with the OpenCode team.
Contributing
See CONTRIBUTING.md for development setup and contribution guidelines. Documentation authoring guidance lives in packages/docs.
Acknowledgments
Special thanks to:
OpenCode for its excellent API and extensible open-source architecture
Pierre for its fast diff viewer and syntax highlighting
Aaron Hogue
7611076436
