// Use OpenBiliClaw's adapter CLI to sync account signals, read profile summaries, fetch recommendations, submit feedback, and inspect runtime status.
Search for videos on Bilibili using keyword queries generated from user interests.
Browse Bilibili pages using agent-browser for visual exploration and DOM interaction.
Analyze video comment sections to discover recommended content and gauge content quality.
| name | openbiliclaw_adapter |
| description | Use OpenBiliClaw's adapter CLI to sync account signals, read profile summaries, fetch recommendations, submit feedback, and inspect runtime status. |
| user-invocable | true |
Use this skill when you are inside the OpenBiliClaw workspace and need current OpenBiliClaw state or want to push feedback back into the learning loop.
Choose deployment by target machine capability:
Run:
docker compose up -d --build
docker exec -it openbiliclaw-backend openbiliclaw init
Keep the repository checkout available so OpenClaw can discover this workspace skill.
If Docker is unavailable, bootstrap locally:
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
cp config.example.toml config.toml
Then initialize OpenBiliClaw once:
openbiliclaw init
If config.toml is still missing API Key or B 站 Cookie and the terminal is interactive, openbiliclaw init will guide the operator through setup. After init, verify the adapter bridge:
uv run python -m openbiliclaw.integrations.openclaw.cli doctor
For a longer setup guide, read docs/openclaw-quickstart.md.
Always call the adapter through the JSON CLI bridge:
uv run python -m openbiliclaw.integrations.openclaw.cli <command> [flags]
Supported commands:
sync-accountget-profileget-delight — check for a proactive surprise recommendationnext-probe — get the next speculative-interest hypothesis to ask the user aboutnext-avoidance-probe — get the next speculative avoidance hypothesis to ask aboutrespond-avoidance-probe --domain "..." --response confirm|reject|chat [--message "..."]chat --message "..." [--session openclaw] — send one Socratic dialogue turn, returns agent replyruntime-statusrecommend --limit 5recommend --limit 5 --refresh-if-neededsubmit-feedback --recommendation-id 7 --feedback-type like --note "很对胃口"listen — long-running WebSocket stream for real-time push events (see below)Instead of polling get-delight / next-probe, OpenClaw can receive real-time push notifications via WebSocket:
uv run python -m openbiliclaw.integrations.openclaw.cli listen
This connects to the runtime stream and outputs one JSON line per event:
{"ok": true, "data": {"status": "connected", "ws_url": "ws://127.0.0.1:8420/api/runtime-stream", "event_types": ["avoidance.probe", "delight.candidate", "interest.probe"]}}
{"ok": true, "data": {"type": "delight.candidate", "bvid": "BV1xxx", "title": "...", "delight_reason": "...", "delight_score": 0.92, "delight_hook": "深层共鸣"}}
{"ok": true, "data": {"type": "interest.probe", "domain": "建筑美学", "reason": "...", "question": "我从你最近的轨迹里嗅到你可能对【建筑美学】感兴趣——... 这个方向你自己认不认？"}}
{"ok": true, "data": {"type": "avoidance.probe", "domain": "浅层热点复读", "reason": "...", "question": "我猜【浅层热点复读】可能是你想避开的方向——... 这个判断准吗？"}}
Default event types: delight.candidate (surprise recommendation), interest.probe (interest hypothesis to confirm), and avoidance.probe (avoidance hypothesis to confirm). The command auto-reconnects on disconnection. Press Ctrl-C to stop.
Options:
--ws-url <url> — override the WebSocket endpoint--events <types> — comma-separated event types to forward (default: avoidance.probe,delight.candidate,interest.probe)OpenClaw can proactively ask the user to clarify or confirm their interests, then send the answer back into the learning loop.
uv run python -m openbiliclaw.integrations.openclaw.cli next-probe
Returns a ready-to-ask question plus raw hypothesis data (domain, reason, specifics, confidence). If no active hypothesis exists, probe is null.
uv run python -m openbiliclaw.integrations.openclaw.cli next-avoidance-probe
If the user confirms the hypothesis:
uv run python -m openbiliclaw.integrations.openclaw.cli respond-avoidance-probe \
--domain "浅层热点复读" \
--response confirm \
--message "对，这类我不想看"
uv run python -m openbiliclaw.integrations.openclaw.cli chat \
--message "嗯对，最近在看很多参数化设计的东西"
The agent replies in Socratic style (probing deeper, proposing hypotheses) and the dialogue automatically feeds back into the soul engine to refine the user's profile.
Use this order for routine work:
get-profilenext-probe — if a hypothesis is pending, ask the user and relay via chatnext-avoidance-probe — if a hypothesis is pending, ask and relay via respond-avoidance-proberecommend --limit <n>submit-feedbackruntime-statusget-delight or listen for proactive surprise recommendations and probessync-account when long-term signals need refreshing{ "ok": false, ... }, surface the error and stop.recommend --limit <n> for normal recommendation fetches. This is the fast path and does not trigger runtime refresh by default.--refresh-if-needed only when the user explicitly wants a heavier freshness check before recommendation fetch.comment feedback, always include --note.uv run python -m openbiliclaw.integrations.openclaw.cli get-profile
uv run python -m openbiliclaw.integrations.openclaw.cli recommend --limit 3
uv run python -m openbiliclaw.integrations.openclaw.cli recommend --limit 3 --refresh-if-needed
uv run python -m openbiliclaw.integrations.openclaw.cli submit-feedback \
--recommendation-id 12 \
--feedback-type comment \
--note "方向对，但我想看更深一点。"
uv run python -m openbiliclaw.integrations.openclaw.cli get-delight
uv run python -m openbiliclaw.integrations.openclaw.cli next-probe
uv run python -m openbiliclaw.integrations.openclaw.cli next-avoidance-probe
uv run python -m openbiliclaw.integrations.openclaw.cli respond-avoidance-probe \
--domain "浅层热点复读" \
--response confirm
uv run python -m openbiliclaw.integrations.openclaw.cli chat \
--message "嗯对，最近在看很多参数化设计的东西"
uv run python -m openbiliclaw.integrations.openclaw.cli listen