gossipcat

data-ai MCP Server

Multi-agent code review mesh — orchestrates AI agents from multiple providers to review code in parallel, cross-review each other's findings, and build accuracy profiles over time. Agents that catch real bugs get picked more often. Agents that hallucinate get deprioritized. MCP server for Claude Code, Cursor, and other IDEs.

Verified
data-aidata-ai
11 views41 stars3 forksv0.8.1MIT

Why This Matters

Discovered via github-topic:mcp and last synced 1w ago.

Verified
Source
github-topic:mcp
Stars
41
Last synced
1w ago
Install
Check source

Install

Install instructions not detected yet

Check the source repository for the latest setup steps.

View source instructions
56
Tools
0
Resources
0
Prompts
Standard I/O
Transport

Available Tools (56)

confirmed

D[reward signal] C -->

Step

What happens

Phase

Time

gossip_relay

Feed native agent results back into the pipeline

Provider

How

ANTHROPIC_API_KEY

Direct API access without the subscription path

gossip_status

System status, dashboard URL, agent list

Native

Relay

1s

Claude picks agents and dispatches them in parallel

Panel

What it shows

gossip_signals

Record or retract accuracy signals

OpenAI

`OPENAI_API_KEY` (+ optional `OPENAI_BASE_URL`)

gossip_dispatch

Multi-agent dispatch: `single`, `parallel`, or `consensus`

gossip_session_save

Save session context for the next session

GOOGLE_API_KEY

Built-in 429 watcher falls back to native on cooldown

gossip_setup

Create or update an agent team

gossip_bug_feedback

File a GitHub issue from an in-session bug report

OpenClaw

— (local gateway)

gossip_remember

Search an agent's cognitive memory

varies

Claude reads UNVERIFIED findings against the code, decides if they're real

Host

Native agents

gossip_relay_cross_review

Feed native cross-review results into consensus

gossip_tools

List all available tools

gossip_skills

Develop, bind, unbind, or list per-agent skills

gossip_update

Check for / apply gossipcat updates from npm

Windsurf

Relay-only (planned)

Yes

Yes

gossip_progress

Check in-progress task status

main_agent

Internal LLM for routing, planning, synthesis (set `provider: "none"` on Claude Code / Cursor to let the host classify natively)

Feature

Status

Cursor

No

gossip_verify_memory

Check a memory claim against current code (FRESH / STALE / CONTRADICTED) before acting on backlog

consensus_judge

Synthesis-only model that merges cross-review results (does not grade)

Field

Description

No

No

gossip_reload

Self-terminate so the host respawns with a fresh bundle (dev loop)

gossip_plan

Decompose a task into sub-tasks with agent assignments

gossip_watch

Stream signals as agents emit them (catches pipeline drops mid-round)

utility_model

Memory compaction, gossip, lens generation

None

Required per provider

DeepSeek

OS keychain via `key_ref`

Decompose

~1s

Cross-review

30s–1min

Capability

What it does

Tool

Purpose

gossip_run

Single-agent dispatch with auto agent selection

gossip_collect

Collect results with optional cross-review synthesis

gossip_scores

View agent accuracy, uniqueness, dispatch weights

gossip_resolve_findings

Mark consensus findings resolved/open

gossip_ask

Ask the dashboard a structured single/multi-select question

gossip_guide

Print the gossipcat handbook for humans

gossip_config

Manage runtime feature-gate flags

gossip_format

Return the canonical `<agent_finding>` output format block

_orchestrator-only_

`gossip_relay`, `gossip_relay_cross_review`, `reply` — used internally to feed native results + the chat bridge back into the pipeline; you never call these

Tag

Means

suggestion

insight` (invariant #8 in `docs/HANDBOOK.md`); the `gossip_signals` receipt surfaces the drop and a `finding_dropped_format` signal. If you see `&lt;agent_finding&gt;` instead of raw tags, a transport layer is entity-encoding output — pass agent output verbatim to `gossip_relay`. **Multiple IDE instances** — each gets its own dynamic port. For a stable port on one project, set `GOSSIPCAT_PORT=24420` in that environment. **Uninstall** — `npm uninstall -g gossipcat && claude mcp remove gossipcat -s user`; `rm -rf ~/.gossip` (global state) or `<project>/.gossip` (per-project). **Still stuck?** [Open an issue](https://github.com/gossipcat-ai/gossipcat-ai/issues) with the last 100 lines of `.gossip/mcp.log` + `gossip_status` output, or ask in-session *"file a gossipcat bug report about …"* (`gossip_bug_feedback` packages it). <br/> ## Configuration Most of `.gossip/config.json` is **auto-generated by `gossip_setup()`** — hand-edit only to change providers/models/endpoints. First-run defaults work for most projects. Config is searched: `.gossip/config.json` → `gossip.agents.json` → `gossip.agents.yaml`. ```json { "main_agent": { "provider": "google", "model": "gemini-2.5-pro" }, "utility_model": { "provider": "native", "model": "haiku" }, "consensus_judge": { "provider": "anthropic", "model": "claude-sonnet-4-6", "native": true }, "agents": { "sonnet-reviewer": { "provider": "anthropic", "model": "claude-sonnet-4-6", "preset": "reviewer", "skills": ["code_review", "security_audit", "typescript"], "native": true } } } ```