Agentic Workspaces

Browse / General-purpose / gasclaw

gasclaw

gastown-publish

Single-container deployment: Gastown + OpenClaw + KimiGas — multi-agent AI workspace with Telegram overseer

Open gastown-publish/gasclaw → Does not match the four rules.

ai-agentsclaude-codegastownkimimulti-agentopenclawtelegram-bot

Works with any agent

ships AGENTS.md, the cross-tool instruction file.

In its own words

README.mdgastown-publish/gasclaw

Gastown + OpenClaw + KimiGas in one container.

A single-container deployment that runs a Gastown multi-agent workspace, managed by an OpenClaw overseer bot on Telegram. All agents use Kimi K2.5 as their LLM backend through Claude Code's agentic interface.

Prerequisites

  • Docker Engine 24+ and Docker Compose v2 — Install Docker

How It Works

Agents run claude CLI with ANTHROPIC_BASE_URL pointed at Kimi's API. Permission bypass is via Claude config file — no --dangerously-skip-permissions flag.

Environment Variables

Optional: GT_RIG_URL, GT_AGENT_COUNT, MONITOR_INTERVAL, ACTIVITY_DEADLINE, DOLT_PORT, GATEWAY_PORT, LOG_LEVEL — see .env.example for the full list.

Documentation

  • Documentation Site — full guides and API reference
  • Wiki — quick reference

For AI agents: llms.txt — machine-readable project reference.

Contributing

See CONTRIBUTING.md and CLAUDE.md.

The whole README. Open the repo ↗

Read from the repository's git tree on 2026-08-17.