fix(server): use CLI for OpenCode health check instead of spawning server

[UtkarshUsername]

What Changed

The OpenCode provider health check no longer spawns opencode serve to fetch inventory (models and agents). Instead, it runs two CLI commands directly:

• opencode models --verbose to get all connected models with full metadata
• opencode agent list to get agents with permission rules

External server configurations (where serverUrl is set) still use the existing HTTP SDK path unchanged.

CLI commands are retried once after 1s on transient failures (e.g. SQLite "database is locked" contention with concurrent serve processes). CLI failure degrades gracefully to empty inventory (warning status) instead of error status.

Why

The health check spawns a full opencode serve process just to call two HTTP endpoints (provider.list and app.agents). Under heavy load (git operations, port discovery, diagnostics queries all running concurrently), this server startup takes 15-20s+ (see #4132), causing the health check to time out and show an error to the user.

The CLI commands opencode models --verbose and opencode agent list return the same data without starting any server, and complete in under 2s regardless of system load.

Checklist

• This PR is small and focused
• I explained what changed and why
• I included before/after screenshots for any UI changes
• I included a video for animation/interaction changes

---

Note

Medium Risk
Changes provider status probing and model/agent discovery for local installs; external-server path is unchanged but parsing CLI output could diverge from SDK data if formats drift.

Overview
Local OpenCode provider health checks no longer start opencode serve or use the HTTP SDK to load models and agents. checkOpenCodeProviderStatus calls the new loadInventoryFromCli path, which runs opencode models --verbose and opencode agent list in parallel, with a one-second retry on transient CLI failures.

Runtime support adds CLI output parsers (parseModelsCliOutput, parseAgentListCliOutput) and maps results into the same inventory shape the SDK used. CLI inventory failures return empty providers/agents instead of failing the probe, so local installs can show warning with zero models rather than error. Configured serverUrl installs still use scoped server connect + loadOpenCodeInventory.

Tests are updated for no local server teardown on health check and for graceful degradation when CLI inventory fails.

^{Reviewed by Cursor Bugbot for commit e2f8d52. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Fix OpenCode health check to use CLI instead of spawning a local server

• For local installs (no serverUrl), checkOpenCodeProviderStatus now calls loadInventoryFromCli instead of spawning or connecting to a local server during health checks.
• loadInventoryFromCli runs opencode models --verbose and opencode agent list in parallel, retries once on transient failure, and degrades gracefully by returning empty inventory rather than failing.
• External server behavior is unchanged: connections still go through connectToOpenCodeServer with SDK and optional serverPassword.
• Behavioral Change: health checks for local installs no longer create a local server scope (closeCalls drops from 1 to 0); CLI failures now yield a warning status with zero models instead of an error.

^{Macroscope summarized e2f8d52.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 2dc24c30-27a5-461d-9f1b-d2c100c4f1fc

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR changes the health check mechanism from spawning a server to using CLI commands for local OpenCode installs. This is a significant runtime behavior change with new parsing logic and retry mechanisms, not a simple bug fix, warranting human review.

^{You can customize Macroscope's approvability policy. Learn more.}

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 2 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 3983e4a. Configure here.}