swarm-mcp
An MCP server that exposes Swarm's fact reconciliation to AI agents, allowing them to query scope, evidence gaps, and review requirements without being able to issue a final verdict.
README
swarm-mcp
An MCP stdio server that exposes Swarm's read + reconcile facts to agent clients (Claude Desktop, Cursor) — so an agent mid-task can ask Swarm "what's my scope?", "what evidence is missing?", "what should the reviewer not miss?" — without being allowed to declare itself done.
What it is — and what it is not
swarm-mcp is a thin adapter over the swarm CLI's --json contract. It spawns swarm <cmd> --json
with fixed arguments and reshapes the output into MCP tools, resources, and prompts. It does not
import swarm-cli's internals, run a model loop, write durable artifacts, or issue a verdict.
- Reconcile-only. Every tool result carries
noVerdictIssued: true; resources serve the CLI's--jsondata verbatim (also no verdict, just unwrapped). swarm-mcp surfaces facts (coverage gaps, out-of-scope changes, empty-evidence Pass rows, self-report mismatches) and a derived human-attention list; a human or an independent reviewer owns the Pass / Fail / Unverified / Blocked result. An empty or weak Evidence cell reads Unverified regardless of a clean reconcile. - Root-confined. It only reads inside a configured workspace root; every client-supplied input is
validated before any subprocess runs. File paths are realpath-confined (no
.., no absolute escapes, no symlink escapes); ids/stems must be a single safe segment; a git base must be a flag-free ref. - Many libraries, not a framework. It couples to swarm-cli only through the public
--jsoninterface, so swarm-cli keeps its minimal footprint and each piece stays useful on its own.
Run it
// Claude Desktop / Cursor MCP config
{
"mcpServers": {
"swarm": {
"command": "swarm-mcp",
"args": ["--workspace", "/path/to/your/swarm-workspace"]
}
}
}
Config: --workspace <path> / SWARM_WORKSPACE (the workspace root); --swarm-bin <path> / SWARM_BIN
(the swarm binary, default swarm on PATH). Requires the swarm CLI
installed.
The swarm-mcp command above resolves to this package's bin. To install from source until a published
build is available:
git clone https://github.com/jcosta33/swarm-mcp && cd swarm-mcp
pnpm install && pnpm build && npm link # exposes `swarm-mcp` on PATH (runs the built dist/)
Node: a published/built install needs Node ≥ 18.18; running from a source checkout (no dist/) needs
Node ≥ 22.6 (it strips types at runtime).
v0 surface (read-only)
- Tools (10). Reconcile/check:
swarm_get_status,swarm_check_workspace,swarm_check_file,swarm_scan_task,swarm_reconcile_review,swarm_validate_review_packet. Parsed-artifact loaders:swarm_get_task,swarm_get_spec,swarm_get_review,swarm_get_checks. - Resources (7). Fixed:
swarm://workspace,swarm://status,swarm://checks. Templated:swarm://tasks/{id},swarm://specs/{id},swarm://reviews/{id},swarm://findings/{id}. - Prompts (5).
swarm_task_briefing,swarm_before_done(the implementer — may not approve its own work),swarm_review_assistant(an independent reviewer — falsify, don't trust),swarm_evidence_rule,swarm_finding_candidate. The before-done / review-assistant asymmetry is deliberate: no prompt grants verdict authority.
Develop
pnpm install
pnpm gate # typecheck + lint + coverage (thresholds enforced) + build
Status: v0 — the full read + reconcile surface over the swarm CLI's --json contract. It couples to
swarm-cli only through that public contract (recorded in
swarm ADR-0085),
so swarm-cli keeps its minimal footprint and each piece stays useful on its own.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.