MCP server for multi-harness AI agent delegation
Wellknown found it in public sources; nobody has proven control of it yet. Claiming takes one click if the repository is under your GitHub account, or a small file on your domain otherwise. Verified owners get the badge, 15-minute checks, status alerts, edits that outrank crawled data, and a ranking boost.
Agents can do it too: POST https://wellknown.network/api/v1/claims with {"agent":"spindle-mcp","method":"well_known_file"} — machine-readable steps at claim.json, guide at /docs/claim.
Everything here was measured by our prober or read from a registry. Nothing is self-reported.
Attributed to the source that supplied each field. Treated as claims, not facts.
# Spindle <!-- Uncomment when published: [](https://badge.fury.io/py/spindle) [](https://github.com/spiritengine/spindle/actions/workflows/ci.yml) [](LICENSE) --> MCP server for multi-harness AI agent delegation. Spawn background agents (Claude Code, Codex, Gemini, Kimi) that run asynchronously, with optional git worktree isolation for safe parallel work. ## Features - **Async agent spawning** - Fire-and-forget pattern with spool IDs - **Optional blocking with gather/yield** - Wait for all results at once, or stream them as agents complete. Alternatively, agent can continue other work, spins are nonblocking by default - **Permission profiles** - Control what tools child agents can use (readonly, careful, full) - **Shard isolation** - Run agents in sandboxed git worktrees to prevent conflicts - **Model selection** - Route tasks to different models per-agent - **Session continuity** - Resume conversations with child agents (auto-recovers expired sessions) - **Rich querying** - Search, filter, peek at running output, export results - **`spindle doctor`** - One command to check the install: versions, paths, spool store, detected harnesses, and an optional live read-only smoke ## Requirements - Python 3.10+ - At least one harness CLI, installed and already authenticated. Spindle shells out to the CLI you already use and inherits its login — it never asks for an API key of its own: - [Claude CLI](https://docs.anthropic.com/en/docs/claude-code) (`claude`) — the default harness - [Codex CLI](https://developers.openai.com/codex/cli/) (`codex`) - [Gemini CLI](https://github.com/google-gemini/gemini-cli) (`gemini`) - [Kimi CLI](https://github.com/MoonshotAI/kimi-cli) (`kimi-cli`) - Git (for shard/worktree functionality) - `bwrap` (bubblewrap), on Linux, if you want s…
Mapped onto the structured taxonomy from declared text and observed tool names. Confidence shown for derived entries.
Every source is kept verbatim. Field changes are logged as events.