AgentCall: Let Claude Code Join Google Meet, Zoom, or Teams Calls as a Teammate

Pattern AI Labs launched AgentCall, a skill that lets your existing coding agent (Claude Code, Codex, or Cursor) join video calls as a participant. The agent speaks, listens, screen-shares a localhost webpage, and can code live during the meeting — all within the same session with full file access.
How It Works
- Voice modes: Collaborative (sub-second latency via voice intelligence layer) or direct (~2s, the agent itself speaks with full reasoning).
- Screen sharing: Shares a URL or local port via per-call tunnel (temp subdomain + secret token, dies when the call ends). No desktop grab — only what the agent chooses.
- Meeting chat: Agent reads incoming messages and can reply (e.g., paste a PR link while talking).
- Participant awareness: Knows who joined/left, who's speaking, can raise hand, toggle mic, change voice/interruption settings.
- Leave/rejoin: “Leave for two minutes and come back” works.
Privacy & Pricing
- TTS/STT run on Pattern's own servers (English STT, multi-language TTS). Direct mode sends no data to third parties.
- No recordings or screen captures stored by default. Transcripts in-memory, wiped on disconnect unless opt-in retention (up to 7 days).
- Pricing: $0.35/min (volume discounts available).
Setup (2 minutes)
- Install the skill:
install join-meeting skill from https://github.com/pattern-ai-labs/agentcall - Get API key at agentcall.dev, paste when prompted.
- Paste any Google Meet / Zoom / Teams URL into your agent. Bot joins in 30–60 seconds.
Works best with Claude Code, also supports Codex and Cursor.
📖 Read the full source: HN AI Agents
👀 See Also

MemAware benchmark shows RAG-based agent memory fails on implicit context retrieval
The MemAware benchmark tests whether AI agents can surface relevant past context when users don't explicitly ask for it, revealing that current memory systems score only 2.8% accuracy on hard implicit queries versus 0.8% with no memory.

Claude-context-lint tool audits token overhead in Claude Code projects
A new tool called claude-context-lint scans Claude Code projects to show how much context window is consumed by CLAUDE.md files, skills, MCP servers, and system prompts before user input. The tool provides specific recommendations for reducing token usage.

CodeLedger: Open-source Claude Code plugin tracks token usage and background agents
CodeLedger is an open-source MCP server plugin for Claude Code that automatically tracks token usage across projects, identifies background agents, and provides cost optimization recommendations based on analysis of local JSONL session files.

Real-time stock analysis added to Claude Desktop via MCP server
A developer built an MCP server called agent-toolbelt that adds real-time stock analysis capabilities to Claude Desktop and Claude Code. The tool provides five specific analysis functions and works with a single command installation.