council-blackboard
Skill0 starsChanged 58 days agoApache-2.0View on GitHub
Visible OpenClaw–Open WebUI council rooms with typed blackboards and session handovers.
In this repository12 other files it gives its agents.
- blackboardSkill
- chatterbox-tts-recoverySkill
- check-notes-firstSkill
- invariant-guarded-debuggingSkill
- local-model-runtime-profilerSkill
- minimum-sufficient-changeSkill
- openwebui-regression-testSkill
- privileged-operationsSkill
- risk-aware-retrySkill
- session-handoverSkill
- symlink-space-saverSkill
- video-clip-editorSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use when the user asks OpenClaw and an Open WebUI model to confer, compare approaches, review one another, or form a multi-model council.
- Very specific: line 12
Create a fresh board and Open WebUI room for every new council: python skills/council-blackboard/scripts/council.py room --topic "..." …
- Distinctive: line 16
Do one bounded contribution per hop.
- Very specific: line 21
Stop on needs_user, a safety boundary, a failed board validation, or exhausted hops.
- Very specific: line 43
Create the room before models confer.
- Very specific: line 51
When UCS participates, pass the council's existing board path as blackboard_path.
- Distinctive: line 55
Check the saved UCS run ID and board revision before routing its result onward.
- Very specific: line 59
Use handover when switching lead model or pausing.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Council blackboard
Use when the user asks OpenClaw and an Open WebUI model to confer, compare approaches, review one another, or form a multi-model council.
Rules
- Create a fresh board and Open WebUI room for every new council:
python skills/council-blackboard/scripts/council.py room --topic "..." --user-id USER_ID - Give the user the returned
chat_urlimmediately. The WebUI room is the visible transcript; the JSON board is the auditable source of truth. - Validate before every hop. Read the goal, typed entries, revision, next capability, and unresolved questions.
- Do one bounded contribution per hop. Never rewrite another writer's entry.
- Record observations as facts, conclusions as inferences, user choices as user_decisions, and checks as evidence/test_results.
- The Open WebUI model uses the attached
Council Blackboardtool to read the board, contribute, and call an isolated OpenClaw council session. - A council handover must name the board path, WebUI chat ID/URL, current revision, decisions, unresolved questions, next capability, and receipt locators. Do not dump private memory into the room.
- The user may join either side at any time. Treat the newest authenticated user contribution as authoritative user context; preserve it as a user decision when it changes scope or constraints.
- Stop on
needs_user, a safety boundary, a failed board validation, or exhausted hops. - Completion still needs independent verification under the blackboard policy.
Configuration
The scripts default to ~/.openclaw/workspace and can be configured through environment variables rather than hard-coded machine paths.
Useful variables include:
OPENCLAW_WORKSPACEOPENWEBUI_BASE_URLOPENWEBUI_DBCOUNCIL_BLACKBOARD_SCRIPTCOUNCIL_ROOM_SCRIPTCOUNCIL_WEBUI_MODELCOUNCIL_WEBUI_MODEL_NAMECOUNCIL_OPENCLAW_MODELCOUNCIL_OPENCLAW_AGENTCOUNCIL_USER_WRITER
OpenClaw to WebUI
Create the room before models confer. Use council.py room; then work from the returned board. The browser-visible URL is derived from OPENWEBUI_BASE_URL and the created chat ID.
WebUI to OpenClaw
The WebUI model calls consult_openclaw with a bounded question and its current position. The tool appends the WebUI contribution, routes the board, invokes a dedicated OpenClaw session key, and returns the OpenClaw reply into the visible WebUI chat.
Handover
When UCS participates, pass the council's existing board path as
blackboard_path. Its imported constraints and appended inference/evidence use
the same typed board that the room reads. The host still owns route claims and
the visible room transcript; UCS does not post to WebUI or authenticate users.
Check the saved UCS run ID and board revision before routing its result onward.
Its four expert roles share the configured model callback and do not count as
four independent reviewers. Normal independent-verification rules still apply.
Use handover when switching lead model or pausing. Keep it operational and short: outcome, constraints, decisions, evidence, failures, open questions, exact next action.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

