一键导入
setup
Connect Claude Code to the Voicenter MCP server using OAuth authentication
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Connect Claude Code to the Voicenter MCP server using OAuth authentication
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Designs the structural skeleton of a Voicenter Bot via interview. Use this skill when the user wants to create, design, scope, or modify a Voicenter voice/chat bot — phrases like "design a bot", "create an agent spec", "build a Voicenter bot", "patch this bot", "add an intent", "change the bot's persona", "modify the flow graph", or any reference to the Agent Spec Designer / Skill 1 in the Voicenter bot generation pipeline. Produces an Agent Spec markdown file (sections 1-4, 4.5, section 5 stubs, section 6 initial, section 7 init). Two named entry modes — greenfield (no spec attached) and patch (spec attached). Does NOT author per-intent language content (validationPrompt, post-execution intentInstructions) — that's Skill 2 (Intent Detail Author). Does NOT emit wire-format JSON — that's Skill 3 (JSON Assembler).
Assembles a fully-detailed Voicenter Agent Spec into Bot JSON wire format — the final mechanical step in the three-skill pipeline. Use this skill when an Agent Spec exists with all section 5 entries marked `[detailed]` and the user wants the deployable JSON. Trigger phrases include "run Skill 3", "assemble the JSON", "emit the bot JSON", "publish the bot", "build the wire-format", "Skill 3 (JSON Assembler)", or any direct continuation from Skill 2's completion handoff. Produces a single `bot-<name>-<date>.json` file plus a banner identifying every fail-loud sentinel and any drift between spec section 6 and what Skill 3 regenerated. Refuses to assemble if any intent is still `[structural]` or `[detailed-revisit]`, or if the spec deviates from the strict template (Doc 2 §3.7). Runs the §15.4 cross-reference pass — 23 checks (8 §15.4 + 3 Compass + 3 botIntents-role + 1 duplicate-global-intent + 8 field-placement doctrine), checks 1–7, 11–13, 15, and 16–21 blocking. Does NOT author any text content (Skills 1 and
Authors the per-intent language content of a Voicenter Agent Spec — slot descriptions, validationPrompt, post-execution intentInstructions, and RT-specific Configuration text. Use this skill when an Agent Spec exists with section 5 entries marked `[structural]` or `[detailed-revisit]`, and the user wants to fill them in. Trigger phrases include "run Skill 2", "detail the intents", "fill in the per-intent fields", "Skill 2 (Intent Detail Author)", or any direct continuation from Skill 1's handoff hint. Walks intents in user-confirmed batches with a checkpoint after each batch. Reactivable — invoke as many times as needed; spec state is the resume point. Does NOT modify the structural skeleton (sections 1, 2, 3, 4, 4.5.1/.2/.4) — that's Skill 1 (Agent Spec Designer). Does NOT emit wire-format JSON — that's Skill 3 (JSON Assembler).
Pull call detail records (CDR) from Voicenter using the Call Log API
Receive and handle CDR push notifications from Voicenter after every call ends
Initiate or terminate outgoing calls using the Voicenter Click2Call API
| name | setup |
| description | Connect Claude Code to the Voicenter MCP server using OAuth authentication |
Language. Reply in the user's language: detect what they write — Hebrew→Hebrew, English→English — and mirror it, switching if they switch mid-conversation. This shapes your prose, your questions, and your
AskUserQuestionoption labels only. It does not change the artifacts you produce — identifiers, JSON keys, BCP-47 language codes, API field names, and other data stay exactly as specified.
Opening. Your first message greets bilingually so the user knows both languages are available — e.g. "נוכל להמשיך בעברית או באנגלית — מה נוח לך? / We can continue in Hebrew or English — whichever you prefer." Then mirror whatever language the user replies in.
One question per turn. Ask exactly one question per message and wait for the answer before asking the next — never present multiple questions in a single turn. When the answer is a closed set (pick-one / yes-no / pick-from-list), use the
AskUserQuestiontool rather than plain text; it automatically adds an "Other" free-text escape, so don't hand-roll one. Reserve plain free-text questions for genuinely open inputs (names, descriptions, URLs, numbers).
Help the developer connect Claude Code to the Voicenter MCP server at mcp01.voicenter.co/mcp.
The Voicenter MCP server uses OAuth — not a static API token or environment variable. You do not need to configure any credentials manually. Claude Code handles the entire OAuth flow for you automatically when you first connect.
There is nothing to set in .env files, shell profiles, or environment variables. The voicenter-mcp plugin only needs the server URL, which is already configured in plugin.json.
In Claude Code:
/plugin install voicenter-mcp@voicenter
The first time Claude Code tries to use the Voicenter MCP server, it will detect that authentication is required and launch the OAuth flow automatically. You will be prompted to:
Claude Code stores the OAuth token securely and refreshes it automatically. You will only need to go through this flow once per device (or when your session expires).
Once authenticated, ask Claude:
List all my Voicenter extensions
Claude will call the MCP server live. If it responds with your extension list, you are connected.
| Prompt | What happens |
|---|---|
| "Initiate a click2call from extension SIPSIP to 0501234567" | Calls the Click2Call API |
| "Show me all calls from the last 7 days" | Queries the Call Log API |
| "Add 0501234567 to the blacklist" | Calls the Blacklist API |
| "List all active campaigns in the dialer" | Calls the Dialer GetCampaignList API |
| "Log in agent user 123456 on extension SIPSIP" | Calls the Login/Logout API |
| "What calls are active right now?" | Calls the Active Calls API |
If your OAuth session expires or you switch Voicenter accounts, Claude Code will prompt you to authenticate again the next time it tries to use the MCP server. This is handled automatically — you do not need to clear any tokens manually.
Claude does not launch the OAuth flow — Try explicitly asking Claude to use a Voicenter tool (e.g. "list my extensions"). This triggers the connection attempt and the OAuth prompt.
Authorization page shows an error — Make sure your Voicenter account has API / MCP access enabled. Contact api@voicenter.com if it does not.
Connection works but Claude cannot perform an action — The OAuth scope granted may not include the required permission. Disconnect and re-authenticate to get a fresh token with the correct scopes.