Files
openclaw/docs/plugins/workboard.md
Peter Steinberger 401f278f11 feat: add Control UI plugin management (#103176)
* feat(ui): add plugin catalog management

* feat(gateway): add plugins.uninstall and richer plugin catalog metadata

Adds a plugins.uninstall gateway method (operator.admin, control-plane write)
backed by a lock-guarded uninstallManagedPlugin that mirrors the CLI flow:
config cleanup, install-record removal, managed file deletion, and registry
refresh. Bundled plugins stay disable-only. Catalog entries now carry a
manifest-derived category and a removable flag; ClawHub search results expose
download counts and verification tiers.

* feat(ui): redesign plugins page with inventory, store shelves, and cover art

Rebuilds /settings/plugins around three tabs: Installed (category-grouped
inventory with overview stats, state filters, uninstall for external plugins,
and inline MCP server management through the shared config seam), Discover
(featured/official shelves plus one-click MCP connectors and curated ClawHub
searches), and ClawHub (search with download counts and verification badges).
Every catalog entry renders bundled cover art or a deterministic gradient
monogram tile - no more empty boxes. Artwork generated with Codex CLI, shipped
as 512px WebP under ui/public/plugin-art.

* chore(ui): regenerate locale bundles for plugins manager strings

* docs: describe plugins manager tabs, uninstall, and MCP connectors

* fix(plugins): human catalog labels and un-pinned hosted fallback ids

listManagedPlugins now prefers manifest names over registry package-name
backfill, falls back to channel catalog labels and blurbs, and stops pinning
expectedPluginId when a hosted feed entry only exposes its package name
(which rejected every legitimate install of that package). Found via live
gateway testing against ClawHub.

* fix(ui): send minimal RFC 7396 merge patches for MCP server edits

config.patch merges rather than replaces, so key removal needs an explicit
null; sending the full config back made MCP server removal a no-op. Found
via live gateway testing.

* fix(ui): write explicit MCP transports for URL servers

The MCP runtime defaults URL-only servers to SSE, so streamable HTTP
endpoints saved by the add form or connector templates would fail at
connect time. Connector templates now declare their transport and the
add form infers streamable-http unless the URL follows the /sse
convention. Flagged by autoreview against the transport resolver.

* test(ui): wait for deferred plugin requests before resolving in e2e

* feat(ui): plugins detail view, action menus, and unified ClawHub search

Reworks the plugins page from PR #103176 feedback: merges the ClawHub tab into
Discover (typing searches ClawHub inline and appends a quiet From ClawHub
section, with Browse ClawHub demoted to a header text link), makes every row and
store card open a plugin detail overlay (hero art, primary enable/install
action, metadata table), and replaces enable/disable switches with a state chip
plus an overflow menu (Enable/Disable, Remove for external plugins, View
details) matching the ChatGPT-store install+menu pattern. MCP rows use the same
menu; refresh is now icon-only.

* chore(ui): regenerate locale bundles for plugins UI iteration

* feat(ui): vetted, grouped connector catalog for the plugins store

Expands Connect your world to 28 connectors organized into use-case shelves
(Work & productivity, Coding & infrastructure, Home & media, Everyday life).
Every entry passed a three-stage subagent review: official-docs verification
plus live endpoint probes for MCP servers, ClawHub result-quality and
malware/typosquat screening for curated searches, and an adversarial pass
that dynamically registered OAuth clients to prove one-click viability.

That review removed Figma (registration allowlisted, 403) and Atlassian
(OAuth issuer-mismatch bug upstream), downgraded GitHub to PAT-based setup
(no dynamic client registration upstream), fixed Linear (/sse retired) and
Home Assistant (/api/mcp, streamable HTTP) endpoints, retargeted poisoned or
dead searches (youtube, finance, hue dropped; calendar -> google calendar;
stocks replaces finance), and added Todoist, Airtable, Canva, Stripe,
Context7, DeepWiki, Hugging Face one-click MCP servers plus Jira, PDF,
transcription, Kubernetes, Reddit, maps, translation, and notes searches.
Keyless servers get a ready-to-use success message; new cover art included.

* chore(ui): regenerate locale bundles for connector groups

* fix(plugins): suppress hosted catalog rows once their package is installed

Hosted feed entries without a declared runtime id fall back to their package
name as catalog id, which never matches the installed runtime id, so the
Discover shelf kept offering an already-installed package. Installed package
names now also suppress official rows. Flagged by autoreview.

* fix(plugins): pin declared runtime ids and surface connector errors in place

The runtime-id pin now keys off explicitly declared catalog ids (plugin,
channel, or provider) instead of string-comparing against the package name,
so declared ids that equal their package name stay enforced while entry-id
fallbacks stay unpinned. Connector add failures on Discover now render on the
triggering card instead of the Installed tab's MCP section. Both flagged by
autoreview; regression tests included.

* feat(ui): full inventory artwork, pulse header, and two-column plugin list

Every bundled plugin now ships distinctive cover art (113 new Codex CLI
illustrations; 172 total, ~2.1MB WebP), so inventory rows and detail views
never fall back to monogram tiles. The four stat cards give way to a compact
inventory pulse: a segmented enabled/disabled/issues meter whose legend and
counts live inside the filter chips. Inventory, MCP, and search rows flow
into two columns when the panel is wide enough.

* chore(ui): regenerate locale bundles for pulse header

* fix(ui): omit stdio args from the MCP server row target

Stdio MCP args routinely carry tokens, and the inventory is visible to
read-only operators; mirror the config page and show only the command.
Flagged by autoreview; regression test included.

* fix(merge): point crestodian setup at relocated plugin commit/refresh modules

* fix(merge): add bootstrapToken to plugins page test gateway harness

* fix(plugins): name catalog install-action branches so Swift emits the union

* fix(ui): satisfy strict lint on plugins page form parsing and mocks

* chore(build): regen docs map, raise plugin-sdk declaration budget for new protocol surface

* fix(ui): type the plugins page patch mock with its real call signature
2026-07-10 11:56:44 +01:00

365 lines
24 KiB
Markdown

---
summary: "Optional dashboard workboard for agent-owned cards and session handoff"
read_when:
- You want a Kanban-style workboard in the Control UI
- You are enabling or disabling the bundled Workboard plugin
- You want to track planned agent work without an external project manager
title: "Workboard plugin"
---
The Workboard plugin adds an optional Kanban-style board to the
[Control UI](/web/control-ui): agent-sized work cards, assignment to agents,
and a link back to the card's task, run, and dashboard session.
Workboard is intentionally small: it tracks local operating work for one
OpenClaw Gateway. It is not a replacement for GitHub Issues, Linear, Jira, or
other team project management systems.
## Enable it
Workboard is bundled but disabled by default:
1. Open **Plugins** in the Control UI, or use `/settings/plugins` relative to
the configured Control UI base path. For example, a base path of `/openclaw`
uses `/openclaw/settings/plugins`.
2. Find **Workboard** and choose **Enable**. Because Workboard is included with
OpenClaw, it does not need an **Install** action.
3. If the UI reports that a restart is required, restart the Gateway.
The Workboard tab appears in the dashboard nav after the plugin runtime loads.
While it is disabled, the tab stays hidden from navigation. Opening the
`/workboard` route directly while the plugin is disabled or blocked by
`plugins.allow`/`plugins.deny` shows a plugin-unavailable state instead of card
data.
The equivalent CLI workflow is:
```bash
openclaw plugins enable workboard
openclaw gateway restart
openclaw dashboard
```
## Configuration
Workboard has no plugin-specific config. Enable/disable it with the standard
plugin entry:
```json5
{
plugins: {
entries: {
workboard: {
enabled: true,
config: {},
},
},
},
}
```
```bash
openclaw plugins disable workboard
openclaw gateway restart
```
## Card fields
| Field | Values |
| ----------- | ------------------------------------------------------------------------------------------------------------- |
| `status` | `triage`, `backlog`, `todo`, `scheduled`, `ready`, `running`, `review`, `blocked`, `done` |
| `priority` | `low`, `normal`, `high`, `urgent` |
| `labels` | free-form strings |
| `agentId` | optional assigned agent |
| linked refs | optional task, run, session, or source URL |
| `execution` | optional metadata for a Codex/Claude run started from the card (engine, mode, model, session, run id, status) |
Cards also carry compact metadata for attempts, comments, links, proof,
artifacts, automation settings, attachments, worker logs, worker protocol
state, claims, diagnostics, notifications, template id, archive state, and
stale-session detection, plus a recent-events list (`created`, `edited`,
`moved`, `linked`, `specified`, `decomposed`, `claimed`, `heartbeat`,
`execution_updated`, `attempt_started`, `attempt_updated`, `comment_added`,
`link_added`, `proof_added`, `artifact_added`, `attachment_added`,
`diagnostic`, `notification`, `dispatch`, `orchestration`,
`protocol_violation`, `archived`, `unarchived`, `stale`). This metadata lets an
operator see how a card moved through the board without opening the linked
session; it is local operating context, not a replacement for session
transcripts or GitHub issue history.
Cards are stored in the plugin's own Gateway state and move with the rest of
that Gateway's OpenClaw state (see [Storage](#storage)).
## Starting work from a card
Unlinked cards can start work directly:
- **Run Codex** / **Run Claude** starts a task-tracked agent run with an
explicit engine, sends the card prompt, and marks the card `running`. Codex
runs use `openai/gpt-5.5`; Claude runs use `anthropic/claude-sonnet-4-6`.
- **Open Codex** / **Open Claude** creates a linked dashboard session without
sending the card prompt or moving the card, for manual work that stays
attached to the board.
Autonomous starts use the Gateway's task-tracked agent run path (default agent
and model unless Codex/Claude is chosen explicitly); Workboard then links the
resulting task, run id, and session key back onto the card. Each linked
execution also records an attempt summary (engine, mode, model, run id,
timestamps, status, rolling failure count) so repeated failures stay visible.
The dashboard refreshes task status from the Gateway task ledger, matching
tasks to cards by task id, run id, or linked session key. A queued/running
task keeps the card's lifecycle active; a finished, failed, timed-out, or
cancelled task moves the card toward `review` or `blocked` using the same sync
rule as linked sessions (see [Session lifecycle sync](#session-lifecycle-sync)).
## Agent tools
| Tool | Purpose |
| ------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `workboard_list` | List compact cards with claim/diagnostic state; optional board filter. |
| `workboard_read` | Return one card plus bounded worker context (notes, attempts, comments, links, proof, artifacts, parent results, recent assignee work, active diagnostics). |
| `workboard_create` | Create a card with optional parents, tenant, skills, board, workspace metadata, idempotency key, runtime limit, retry budget. |
| `workboard_link` | Link a parent to a child card. Children stay `todo` until every parent reaches `done`, then dispatch promotion moves them to `ready`. |
| `workboard_claim` | Claim a card for the calling agent; moves `backlog`/`todo`/`ready` into `running`. |
| `workboard_heartbeat` | Refresh the claim heartbeat during a longer run. |
| `workboard_release` | Release the claim after completion, pause, or handoff; can move the card to a next status. |
| `workboard_complete` / `workboard_block` | Structured lifecycle tools for final summaries, proof, artifacts, and created-card manifests (must reference cards linked back to the completed card) or blocker reasons. |
| `workboard_attachment_add` / `workboard_attachment_read` / `workboard_attachment_delete` | Store small card attachments in plugin SQLite state, index on the card, expose in worker context. |
| `workboard_worker_log` / `workboard_protocol_violation` | Record worker log lines and block a card when an automated worker stops without calling `workboard_complete`/`workboard_block`. |
| `workboard_board_create` / `workboard_board_archive` / `workboard_board_delete` | Manage persisted board metadata (display name, description, archive state, default workspace). |
| `workboard_runs` | Return the persisted run-attempt history for a card. |
| `workboard_specify` | Turn a rough triage/backlog card into a clarified `todo` card; records the spec summary on the card. |
| `workboard_decompose` | Fan a parent orchestration card into linked children, inheriting board/tenant metadata; can complete the parent with a created-card manifest. |
| `workboard_notify_subscribe` / `workboard_notify_list` / `workboard_notify_events` / `workboard_notify_advance` / `workboard_notify_unsubscribe` | Manage notification subscriptions. Event reads are replay-safe; `advance` moves the durable cursor so callers resume without losing or double-reading completed/failed/stale card events. |
| `workboard_boards` / `workboard_stats` | Inspect board namespaces and queue stats. |
| `workboard_promote` / `workboard_reassign` / `workboard_reclaim` | Recover or hand off stuck work. |
| `workboard_comment` / `workboard_proof` | Add handoff notes or attach proof/artifact references. |
| `workboard_unblock` | Move blocked work back to `todo`. |
| `workboard_dispatch` | Nudge dependency promotion or stale-claim cleanup. |
Claimed cards reject agent-tool mutations from other agents unless the caller
holds the claim token returned by `workboard_claim`. Every card returned by an
agent tool or Gateway RPC call redacts `metadata.claim.token` to `[redacted]`
(the token itself is returned once, top-level, only from `workboard_claim`),
so dashboard operators and other agents can inspect claim state without ever
seeing a usable token. Recovery goes through
`workboard_promote`/`workboard_reassign`/`workboard_reclaim`, which do not
require the token.
## Dispatch
Dispatch is Gateway-local: it does not spawn arbitrary OS processes. Normal
OpenClaw subagent sessions still own execution. One dispatch pass:
1. Promotes dependency-ready cards.
2. Records dispatch metadata on ready cards.
3. Blocks expired claims or timed-out runs.
4. Marks board-configured triage cards as orchestration candidates.
5. Claims a small batch of ready cards and starts worker runs through the
Gateway subagent runtime.
Workers get bounded card context plus the claim token needed to heartbeat,
complete, or block the card through the Workboard tools.
### Worker selection
Each pass starts **at most 3 workers by default**. Ready cards are ordered by
priority, then position, then creation time. A pass starts only one card per
owner/agent and skips owners that already have running or review work on the
board. Archived cards, cards with an active claim, and cards not in `ready`
status are never selected for worker starts (they can still be affected by the
data side of dispatch: stale-claim cleanup, dependency promotion, timeout
cleanup).
Session keys are deterministic per board/card, so repeated dispatches route
back to the same worker lane instead of creating unrelated sessions:
- Assigned cards: `agent:<agentId>:subagent:workboard-<boardId>-<cardId>`
- Unassigned cards: `subagent:workboard-<boardId>-<cardId>` (Gateway resolves
the configured default agent)
If a worker cannot be started after a card is claimed, Workboard blocks the
card, clears the claim, records the run-start failure, and appends a worker
log line - visible in the dashboard, CLI JSON, agent tools, and card
diagnostics.
### Entry points
- Dashboard dispatch action
- `openclaw workboard dispatch`
- `/workboard dispatch` on a command-capable channel
All three use the Gateway subagent runtime when the Gateway is available. The
CLI has one operator fallback: if the Gateway call fails with a
connection/unavailable error (or an `unknown method` error for older
Gateways), and no explicit `--url`/`--token` target and no configured remote
Gateway (`OPENCLAW_GATEWAY_URL` or `gateway.mode: remote`) apply, the CLI runs
data-only dispatch against local SQLite state - it can promote dependencies,
clean stale claims, and block timed-out runs, but cannot start workers. Auth,
permission, and validation failures from a reachable Gateway are not treated
as unavailable; they surface as command errors, and so does any Gateway
failure when an explicit `--url`/`--token` target was given.
Board metadata can set `autoDecompose`, `autoDecomposePerDispatch`,
`defaultAssignee`, and `orchestratorProfile`. OpenClaw records this intent and
exposes it in worker context; actual specification/decomposition still runs
through the normal Workboard tools.
## CLI and slash command
```bash
openclaw workboard list [--board <id>] [--status <status>] [--include-archived] [--json]
openclaw workboard create "Fix stale card lifecycle" --priority high --labels bug,workboard
openclaw workboard show <card-id> [--json]
openclaw workboard dispatch [--board <id>] [--json]
```
`list` text output hides archived cards by default (`--include-archived`
overrides); `--json` always includes archived cards, matching the full-card
contract used by existing scripts. `show` accepts an unambiguous id prefix.
`list`, `create`, and `show` always read/write local plugin state directly.
Only `dispatch` calls the running Gateway, with the fallback described above.
See [Workboard CLI](/cli/workboard) for full flags, JSON output, Gateway
fallback behavior, id-prefix handling, dispatch selection rules, and
troubleshooting.
`/workboard list`, `/workboard show <card-id>`, `/workboard create <title>`,
and `/workboard dispatch` mirror the CLI. List and show are read operations
for any authorized command sender. Create and dispatch require owner status on
chat surfaces, or a Gateway client with `operator.write`/`operator.admin`.
## Session lifecycle sync
Cards can link to an existing dashboard session, or one created when you
start work from the card. Linked cards show the session lifecycle inline:
running, stale, linked idle, done, failed, or missing. You can also capture an
existing session from the Sessions tab with **Add to Workboard**; the card
links to that session, uses the session label or recent user prompt as title,
and seeds notes from the recent user prompt plus the latest assistant response
when available.
If the linked session goes missing, the card stays linked for context and
still offers start controls to restart into a fresh session. If an active
linked session stops reporting recent activity, Workboard marks the card
`stale` and stores that as metadata until the lifecycle clears it.
While a card is in an active work state, Workboard follows the linked session:
| Linked session state | Card status |
| ------------------------------------- | ----------- |
| active | `running` |
| completed | `review` |
| failed, killed, timed out, or aborted | `blocked` |
**Manual review states win.** Moving a card to `review`, `blocked`, or `done`
stops auto-sync for that card until you move it back to `todo` or `running`.
Starting a card uses normal Gateway sessions; Workboard only stores card
metadata and links. Conversation transcript, model selection, and run
lifecycle stay owned by the regular session system. Use **Stop** on a live
linked card to abort the active run - Workboard marks that card `blocked` so
it stays visible for follow-up.
New cards can start from Workboard templates (`bugfix`, `docs`, `release`,
`pr_review`, `plugin`). Templates prefill title, notes, labels, and priority;
the template id is stored as card metadata.
## Dashboard workflow
1. Open the Workboard tab in the Control UI.
2. Create a card with a title, notes, priority, labels, optional agent, and
optional linked session - or open Sessions and choose **Add to Workboard**
for an existing session.
3. Drag the card between columns, or focus its compact status control and use
the menu or ArrowLeft/ArrowRight.
4. Start work from the card to create or reuse a dashboard session.
5. Open the linked session from the card while the agent works.
6. Let lifecycle sync move running work into `review`/`blocked`, then manually
move the card to `done` when accepted.
## Diagnostics
Diagnostics are computed from local card metadata. Built-in checks flag:
| Kind | Condition |
| --------------------------- | ------------------------------------------------------------------------------ |
| `stranded_ready` | Assigned `todo`/`backlog`/`ready` card not updated in over 1 hour. |
| `running_without_heartbeat` | `running` card with no claim heartbeat or execution update in over 20 minutes. |
| `blocked_too_long` | `blocked` card not updated in over 24 hours. |
| `repeated_failures` | Card's tracked failure count reaches 2 or more. |
| `missing_proof` | `done` card with no proof, artifacts, or attachments. |
| `orphaned_session` | `running` card with a `sessionKey` but no `execution` metadata. |
## Permissions
Gateway RPC methods live under `workboard.*`:
| Scope | Methods |
| ---------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `operator.read` | `cards.list`, `cards.export`, `cards.diagnostics`, attachment list/get, notification event reads, `boards.list`, `cards.stats`, `cards.runs` |
| `operator.write` | `cards.diagnostics.refresh`, create/update/move/delete/comment/link/linkDependency/proof/artifact, attachment add/delete, worker log, protocol violation, claim/heartbeat/release/promote/reassign/reclaim/complete/block/unblock, `cards.dispatch`, `cards.bulk`, archive, `boards.upsert`/`archive`/`delete`, `cards.specify`/`decompose`, notification subscribe/delete/advance |
No RPC method requires `operator.admin`. Browsers connected with read-only
operator access can inspect the board but cannot mutate cards.
## Storage
Workboard stores durable data in a plugin-owned relational SQLite database
under the OpenClaw state directory: boards, cards, labels, lifecycle events,
run attempts, comments, dependency links, proof, artifact references,
attachment metadata and blobs, diagnostics, notifications, worker logs,
protocol state, and subscriptions all live in Workboard tables (not
plugin key-value entries). A card export preserves the board narrative
without inlining attachment blob contents.
Installations that used Workboard in the `.28` release can run
`openclaw doctor --fix` to migrate the shipped legacy plugin-state namespaces
(`workboard.cards`, `workboard.boards`, `workboard.notify`, and, if present,
`workboard.attachments`) into the relational database.
## Troubleshooting
**The tab says Workboard is unavailable**
```bash
openclaw plugins inspect workboard --runtime --json
```
If `plugins.allow` is configured, add `workboard` to it. If `plugins.deny`
contains `workboard`, remove it before enabling the plugin.
**Cards do not save**
Confirm the browser connection has `operator.write` access. Read-only operator
sessions can list cards but cannot create, edit, move, or delete them.
**Starting a card does not open the expected session**
Check the card's agent id and linked session, then open Sessions or Chat to
inspect the actual run state.
**Dispatch does not start a worker**
Confirm there is at least one `ready` card without an active claim:
```bash
openclaw workboard list --status ready
```
If the CLI reports data-only dispatch, start or restart the Gateway and
retry - data-only dispatch updates local board state but cannot start
subagent worker runs. Cards can also be skipped when another card for the
same owner or agent is already running or waiting for review; complete,
block, or release that active work before dispatching more for the same
owner.
## Related
- [Control UI](/web/control-ui)
- [Workboard CLI](/cli/workboard)
- [Plugins](/tools/plugin)
- [Manage plugins](/plugins/manage-plugins)
- [Sessions](/concepts/session)