← 返回 Skills 市场
637
总下载
0
收藏
1
当前安装
5
版本数
在 OpenClaw 中安装
/install ask-council
功能描述
Ask LLM Council a question directly from Telegram/chat — get the chairman's synthesized answer without opening the web UI. Quick, headless access to multi-mo...
使用说明 (SKILL.md)
Ask Council — Quick Headless Access
Get LLM Council's synthesized answer without leaving your chat.
Usage
/council Should I invest in Tesla right now?
Returns the Chairman's synthesized answer after all models have debated.
How It Works
- Sends your question to the LLM Council backend
- Waits for Stage 1 (all models respond)
- Waits for Stage 2 (models rank each other)
- Returns Stage 3 (Chairman's final synthesis)
Takes 30-60 seconds — models need time to deliberate.
Prerequisites
LLM Council backend must be running:
/install-llm-council
Two Ways to Use LLM Council
| Mode | Best For | Command |
|---|---|---|
| Quick answer (this skill) | Fast decisions, mobile, casual questions | /council "question" |
| Full discussion (web UI) | Deep research, exploring disagreements, seeing all model responses | /install-llm-council then open browser |
Example
Input:
/council Is Python or Go better for a new microservice?
Output:
Council is deliberating... (this may take 30-60s)
................
═══════════════════════════════════════════════════════════════
CHAIRMAN'S ANSWER
═══════════════════════════════════════════════════════════════
Based on the council's deliberation, Python is recommended for rapid
prototyping and team velocity, while Go excels for high-throughput
services where performance is critical...
═══════════════════════════════════════════════════════════════
View full discussion: http://10.0.1.184:5173
Agent Instructions
When user says /council \x3Cquestion> or "ask council":
bash ~/.openclaw/skills/ask-council/ask-council.sh "\x3Cquestion>"
The script handles:
- Creating a conversation
- Starting the council run
- Polling until complete
- Extracting the chairman's answer
- Showing progress dots while waiting
Files
| File | Purpose |
|---|---|
SKILL.md |
Documentation |
ask-council.sh |
Main script — queries API and returns answer |
_meta.json |
Skill metadata |
Notes
- Timeout: 120 seconds
- If backend isn't running, suggests starting it
- Always includes link to full web UI for detailed exploration
- Creates a new conversation each time (no history)
安全使用建议
This skill is consistent with its stated purpose and appears to be low-risk, but check these before installing: 1) The script expects curl and python3 to be available — the skill metadata does not declare these; ensure your environment has them. 2) The script talks to a backend on localhost (127.0.0.1:8001) and constructs a local web-UI URL (port 5173). Only use this if you trust/run the LLM Council backend on your machine or network. 3) The package does not install the backend; SKILL.md suggests running /install-llm-council separately — confirm where that installer comes from. 4) The included script is small and readable (no obfuscated code), but because it will execute locally, review it yourself or run it in a safe environment if you are unsure. If you need higher assurance, ask the author for an explicit list of required binaries and a documented backend install procedure; resolving those would raise confidence to high.
功能分析
Type: OpenClaw Skill
Name: ask-council
Version: 1.0.4
The skill bundle is benign. The `ask-council.sh` script interacts exclusively with a local API endpoint (`http://127.0.0.1:8001`). User input is safely escaped using `python3 -c 'import json, sys; print(json.dumps(sys.stdin.read().strip()))'` before being embedded in JSON payloads for `curl` requests, preventing shell and JSON injection. The use of `hostname -I` is for displaying a local IP address in a URL for user convenience, not for malicious network activity. There is no evidence of data exfiltration, malicious execution, persistence mechanisms, or prompt injection attempts against the agent within the skill's instructions or code.
能力评估
Purpose & Capability
The skill's name/description (quick headless access to an LLM Council) matches the behavior in ask-council.sh: it creates a conversation, starts a run, polls until completion, and prints the chairman's answer. However, the script depends on curl and python3 (used for JSON parsing) and expects network access to localhost ports 8001/5173, yet the skill declares no required binaries or environment variables — this mismatch is operationally important though not malicious.
Instruction Scope
SKILL.md instructs the agent to run the included shell script with the user's question. The script only interacts with a local backend (API_BASE=http://127.0.0.1:8001), polls for status, and prints a local web-UI link; it does not reach out to arbitrary external hosts or attempt to read unrelated files or secrets.
Install Mechanism
This is an instruction-only skill with a bundled script (no install spec). That is low risk, but because the script will be executed on the host, users should confirm the script's contents (which were provided) and be aware there is no automated package installation. The SKILL.md references an external repo and a /install-llm-council step, but the skill package does not provide or install the backend.
Credentials
The skill requests no environment variables or external credentials and the script does not attempt to access secrets or unrelated config paths. Its only system interactions are network calls to localhost and a call to hostname -I to display a local IP.
Persistence & Privilege
always is false and the skill does not request persistent system privileges or modify other skills or system-wide settings. It runs a single script per invocation and creates ephemeral conversations on the backend.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install ask-council - 安装完成后，直接呼叫该 Skill 的名称或使用
/ask-council触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.4
Clean up - working short URL support, outputs conversation link
v1.0.3
URL now includes #synthesis anchor to jump directly to chairman's answer
v1.0.2
Short URL support — conversations get 7-character IDs for easy sharing
v1.0.1
Added conversation ID and run ID to output; notes about upcoming deep link support
v1.0.0
Initial release — quick headless access to LLM Council from chat/Telegram
元数据
常见问题
Ask Council 是什么？
Ask LLM Council a question directly from Telegram/chat — get the chairman's synthesized answer without opening the web UI. Quick, headless access to multi-mo... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 637 次。
如何安装 Ask Council？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ask-council」即可一键安装，无需额外配置。
Ask Council 是免费的吗？
是的，Ask Council 完全免费（开源免费），可自由下载、安装和使用。
Ask Council 支持哪些平台？
Ask Council 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Ask Council？
由 jeadland（@jeadland）开发并维护，当前版本 v1.0.4。
推荐 Skills