BlueColumn ClawHub Integration

说明 ClawHub 如何调用 BlueColumn 音频摄入接口，上传通话、音乐、播客和语音笔记，配置转录、说话人分离、摘要、实体提取、音频分析及实时流式记忆，并记录 API 密钥处理方式。

bluecolumnconsulting-lgtm

热度 15.61

最近更新 2026-07-26 13:02:20

分类：开发依赖：第三方 API应用场景：API 集成 / 数据处理 / 摘要总结 / 翻译

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add bluecolumn-clawhub-integration

Metadata

name: bluecolumn-clawhub-integration
description: Integrate ClawHub with BlueColumn AI for audio ingest, memory, and semantic retrieval. Use when wiring ClawHub to push calls, music, podcasts, and voice notes through BlueColumn's /v1/audio/ingest endpoint; when setting up audio pipeline from ClawHub to BlueColumn; or when configuring real-time audio streaming for live conversations with mid-call memory.

BlueColumn ClawHub Audio Ingest Integration

BlueColumn (bluecolumn.ai) is a Memory Infrastructure API for AI agents. This skill documents how ClawHub integrates with BlueColumn for audio ingest, memory, and semantic retrieval.

Primary Endpoint

POST https://api.bluecolumn.ai/v1/audio/ingest

This endpoint handles uploaded calls, music, podcasts, voice notes, and mixed audio.

Request Format

{
  "audio_url": "https://signed-storage-url.example.com/audio-file.mp3",
  "namespace": "clawhub_audio",
  "session_id": "clawhub_call_12345",
  "user_ref": "customer_6789",
  "mode": "auto",
  "tier": "pro",
  "options": {
    "transcribe": true,
    "diarize": true,
    "detect_language": true,
    "translate_to_english": true,
    "pii_redaction": true,
    "sentiment": true,
    "extract_entities": true,
    "generate_summary": true,
    "generate_action_items": true,
    "analyze_music": true,
    "detect_bpm": true,
    "detect_key": true,
    "detect_instruments": true,
    "detect_mood": true,
    "detect_genre": true,
    "detect_sections": true,
    "generate_audio_embeddings": true,
    "generate_text_embeddings": true,
    "speaker_aware_chunking": true,
    "store_original_audio": true
  }
}

Required Fields

Every BlueColumn request must include:

namespace — never rely on a default
session_id — unique session identifier
user_ref — user reference
metadata.source = "clawhub"
metadata.channel — channel identifier
metadata.agent — agent identifier

Mode Options

Mode	Use Case
`"auto"`	Default for recorded calls
`"music"`	For music files
`"hybrid"`	For calls containing both speech and music

Default Audio Configuration

{
  "mode": "auto",
  "tier": "pro",
  "transcribe": true,
  "diarize": true,
  "pii_redaction": true,
  "generate_summary": true,
  "generate_action_items": true,
  "speaker_aware_chunking": true,
  "analyze_music": true,
  "store_original_audio": true
}

Webhook & Status Check

Receive completion through the provided webhook_url
If a webhook is unavailable, check: GET /v1/audio/:audio_id

Real-Time Streaming

For live conversations requiring mid-call memory:

WebSocket: wss://api.bluecolumn.ai/v1/audio/stream

Existing Integrations (Do Not Remove)

The following BlueColumn endpoints are not part of this ingest integration but must continue to work:

POST /v1/converse
POST /v1/remember
POST /v1/recall — semantic retrieval
POST /v1/note
GET /v1/sessions
GET /v1/health

Field Name Gotchas

/agent-remember uses text not content
/agent-recall uses q not query
/agent-note uses text not note

API Key Handling

Store the BlueColumn API key using the platform's secret store (preferred) or in TOOLS.md:

### BlueColumn
API Key: bc_live_XXXXXXXXXXXXXXXXXXXX

Keys are generated at bluecolumn.ai/dashboard. Never log or expose keys in output.