tdenton8772

Da Boss — Development skill for Claude Code

Development community

Web-based manager for spawning, monitoring, and controlling multiple Claude Code agent instances.

How to install Da Boss

This entry records only its repository, not the path inside it, so there is no exact command to give. Open tdenton8772/da_boss and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Da Boss does

Web-based manager for spawning, monitoring, and controlling multiple Claude Code agent instances.

Alternatives in Development

OpenAI Codex CLI — (Rust implementation) 67.8k ★
Research Codebase — You are tasked with conducting comprehensive research across the codebase to answer user questions by spawning 10k ★
Agent Of Empires — Manage multiple Claude Code, OpenCode agents from either TUI or Web for easy access on mobile 3.1k ★

README

da_boss

An open-source control plane for fleets of [Claude Code](https://docs.anthropic.com/en/docs/claude-code) agents. Built on the `@anthropic-ai/claude-agent-sdk`.

da_boss runs each agent in its own Kubernetes pod on the dispatching user's own Claude credentials, then shepherds the agent's work through a full change lifecycle: branch → draft PR → tests → agent-driven code review → gated merge → gated deploy. A supervisor watches everything, unsticks stuck agents, sizes pods, heals dead ones, and re-tests `main` when it moves.

It is domain-neutral: da_boss knows nothing about your application. Target repos declare their own toolchain and pipeline in a `.daboss/` directory, and deploy identity is injected via service accounts — never baked in.

How it works

Browser (React)  ──REST + WebSocket──▶  Boss (Express :3847)  ──▶  Kubernetes API
MCP clients      ──POST /mcp────────▶       │                        ├─ agent pods (1 per agent, per-user creds + workspace PVC)
                                            │                        ├─ pipeline pods (tests, deploys)
                                        Postgres  ◀──LISTEN/NOTIFY── ├─ kaniko pods (image builds)
                                            │                        └─ optional native sidecar (heartbeat, git telemetry, leases)
                                       Supervisor (cron + dispatch queue)

One pod per agent. The worker (server/src/worker/) clones the target repo into a per-user workspace volume, runs the SDK loop, and on each turn end auto-commits, pushes, and opens/updates a draft PR. Permissions, steering commands, and live events flow between pod and boss exclusively through Postgres LISTEN/NOTIFY — no shared memory.
Per-user everything. Each user stores their own Claude credential (API key or OAuth token) and git token, encrypted at rest (AES-256-GCM). Agents run and bill on the credentials of whoever dispatched them.
**The pipeline is the repo's contra