See delegated work without leaving OpenCode. Subagent Monitor is an MIT-licensed OpenCode TUI sidebar plugin that keeps running, completed, and failed subagents visible, with elapsed time and token/context usage when OpenCode provides it.
Add the package to your OpenCode TUI configuration:
{
"$schema": "https://opencode.ai/tui.json",
"plugin": ["opencode-subagent-statusline"]
}The configuration usually lives at:
~/.config/opencode/tui.jsonRestart OpenCode after saving the file. The package is published as opencode-subagent-statusline and requires Node >=22.13.
Delegating work is powerful, but child sessions can disappear into the background. Without a visible status surface, you have to guess:
- Is the review agent still running?
- Did the test agent finish?
- Which child session failed?
- How much context did a subagent use?
Subagent Monitor restores that visibility inside OpenCode, so you can keep working while still knowing what your delegated agents are doing.
The sidebar shows:
- running subagents;
- recent completed subagents;
- failed subagents;
- elapsed time;
- token/context usage when available.
When subagent activity is active, the plugin also adds a compact summary to the home/footer area.
Gentle AI offers Subagent Monitor as an optional OpenCode community plugin. Select and install it through Gentle AI to add the selected plugin to OpenCode's tui.json, or install this package directly using the configuration above.
This integration is optional. Subagent Monitor remains an independently installable OpenCode plugin.
Full OpenCode context with demo content in Spanish:
Focused sidebar view:
Run Subagents: Focus sidebar list from the OpenCode command palette, or press Alt+B, to focus the subagent sidebar list without using the mouse. List navigation shortcuts are handled only while the sidebar list is focused.
| Shortcut | Action |
|---|---|
Alt+B |
Toggle focus between the subagent sidebar list and the prompt. |
j / ArrowDown |
Move selection to the next visible subagent. |
k / ArrowUp |
Move selection to the previous visible subagent. |
Enter |
Open the selected subagent session. |
c |
Toggle retained completed history in the sidebar. |
h / ArrowLeft |
Collapse the subagent section. |
l / ArrowRight |
Expand the subagent section. |
Esc |
Leave list focus mode and return to the prompt. |
Opening a selected session is a no-op when there is no visible or navigable subagent.
Click Σ in the sidebar aggregate row to toggle completed history with the mouse. The toggle is not persisted; it resets when OpenCode or the plugin is reloaded. Completed history is bounded retained history, not a full database: terminal rows are kept for up to 3 days with a 1,500-row cap, and rows already pruned from state are not restored.
When a child session is opened from the sidebar, returning with OpenCode Up (session_parent) moves keyboard focus to the parent prompt so you can type immediately.
Stable 1.x public contract
For 1.x releases, the stable user-facing contract is:
- npm package name:
opencode-subagent-statusline; - TUI plugin entrypoints:
opencode-subagent-statuslineandopencode-subagent-statusline/tui; - OpenCode
tui.jsonplugin configuration; - visible sidebar and home/footer behavior;
- command palette entry,
Alt+B, and focused-list navigation; - local privacy and persistence behavior described in this README;
- Node, peer dependency, and install contract declared in
package.json.
Experimental or internal surfaces may change in 1.x without a SemVer-major bump:
opencode-subagent-statusline/runtime, intended for diagnostics and file-based runtime experiments;- diagnostic environment variables;
- exact
state.jsonschema andstatus.txtformat; - internal source modules and source-level exports.
Use the TUI plugin entrypoints for normal OpenCode usage.
For deeper installation, architecture, event-flow, state, rendering, TUI, configuration, testing, and troubleshooting details:
Check OpenCode logs:
grep -n "subagent-statusline\|failed to load tui plugin" ~/.local/share/opencode/log/*.logThen restart OpenCode after changing tui.json.
OpenCode may be using a cached package. Try clearing the cached package directory under:
~/.cache/opencode/packages/Then restart OpenCode.
OpenCode event payloads can vary by version and by event type. The plugin shows token/context usage when it is available and safely omits it when it is not.
The plugin persists a local JSON state file and status.txt snapshot under XDG_RUNTIME_DIR or the system temp directory by default. Those files can include OpenCode-derived subagent titles and summaries, which may contain short fragments derived from prompts or task descriptions. Files are written best-effort with owner-only permissions and atomic temp-file replacement where Node and the host filesystem support them.
OPENCODE_SUBAGENT_STATUSLINE_STATE overrides the state file path. Treat that environment variable as trusted local configuration because the plugin will write status data to the configured path.
For token/context backfill, the TUI reads recent local OpenCode SQLite/log data only from the user's OpenCode data directory. Very large log files are skipped to avoid blocking the TUI.
Development and testing
Install dependencies with lifecycle scripts disabled by default:
pnpm install --ignore-scriptsBuild the plugin:
pnpm buildTest a local TUI build by pointing OpenCode directly at dist/tui.js:
{
"$schema": "https://opencode.ai/tui.json",
"plugin": ["/absolute/path/to/sub-agent-statusline/dist/tui.js"]
}This project ships the OpenCode TUI sidebar plugin from src/tui.tsx. The TUI bundle is built with tsup and esbuild-plugin-solid in Solid universal mode for OpenTUI compatibility.
Package entrypoints:
opencode-subagent-statusline -> TUI plugin
opencode-subagent-statusline/tui -> TUI plugin
opencode-subagent-statusline/runtime -> experimental/diagnostic runtime modeUseful commands:
pnpm build
pnpm typecheck
pnpm test
pnpm test:watch
pnpm test:coverage
pnpm pack --dry-runAutomated tests use Vitest with @vitest/coverage-v8. For the testing strategy, file map, examples, and current TUI/e2e boundaries, see docs/testing.md.
Security hardening for maintainers
Recommended local npm/pnpm hygiene, following guidance from Gentle AI and Liran Tal:
- install project dependencies with lifecycle scripts disabled when possible, for example
pnpm install --ignore-scripts; - consider setting user-level
ignore-scripts=truefor npm/pnpm and temporarily opt in only when a trusted package needs scripts; - enable dependency age/cooldown policies where supported, for example
npm config set min-release-age 3or equivalent Renovate/Dependabot cooldowns; - block or review git, tarball, URL, and other exotic dependency specs, for example
npm config set allow-git nonewhere supported; - optionally screen new packages with tools such as
npqor Socket Firewall before adding them.
These are maintainer/developer controls, not runtime enforcement by this plugin.
Release maintainers should keep the repository NPM_TOKEN secret restricted, retain npm provenance, require npm 2FA on maintainer accounts, and protect the release branch in GitHub. See the release process for the tag and recovery gates.


