# ProofGraph (Antigravity MCP)

> ProofGraph (Antigravity MCP) is an open-source MCP server for knowledge and memory. Evidence-first knowledge graph MCP server for verifying claims and tracing sources. It is available on CLI, API and self-hosted.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, API, self-hosted
- **Website:** https://aibuildinfra.com/proofgraph/
- **Repository:** https://github.com/ai-buildinfra/proof-graph
- **MCP Registry name:** io.github.AI-BuildInfra/proof-graph
- **Categories:** [knowledge and memory](https://ai-powered.com/categories/knowledge-and-memory)
- **Listing updated:** 2026-11-18

## About

The README says ProofGraph is built for the case where an agent needs to check a statement before repeating it. Instead of passing large blocks of scraped web text into a prompt, it resolves entities, looks up claims against stored evidence and returns a small packet with source URLs, retrieval timestamps and SHA-256 payload hashes.

The server exposes twelve MCP tools covering entity search, claim verification, contradiction comparison, multi-hop relationship traversal, a cross-platform footprint audit and graph export to JSON, JSON-LD, CSV or GraphML. A budget setting caps how much evidence comes back, and excerpts are deduplicated before they reach the model. It runs locally over stdio and blocks outbound requests to private IP ranges and cloud metadata endpoints.

The server is aimed at developers building AI applications or research tools that need to cite sources or map how organisations, packages and specifications relate. It does not establish real-world truth: when a claim has not been indexed, the output is UNVERIFIED rather than an invented source.

## Features

- **Claim verification:** Checks an assertion against stored evidence and classifies it as supported, contradicted, conflicting or unverified.
- **Source provenance trail:** Links a claim to its excerpt, source URL, retrieval timestamp and SHA-256 hash for audit.
- **Budgeted evidence packets:** Returns the smallest evidence set that answers a prompt within a configurable token limit.
- **Entity resolution:** Finds canonical entities by name, alias, keyword or domain, including GitHub org and package namespaces.
- **Relationship traversal:** Walks one-, two- and three-hop paths between entities with edge confidence weighting.
- **Graph export:** Exports the knowledge graph to JSON, JSON-LD, CSV or GraphML.

## Good for

- Adding a verification step that checks an agent's claims before it answers
- Mapping how organisations, packages and specifications connect without writing web crawlers
- Comparing conflicting statements from different sources side by side
- Auditing whether a company's name, domain and repository identifiers are consistent
- Reducing prompt size by replacing raw retrieved pages with deduplicated evidence

## Tools

- `get_evidence_packet`: Returns the smallest evidence packet sufficient to answer a prompt within a strict token budget.
- `verify_claim`: Checks an assertion against available evidence to determine support or contradiction.
- `trace_claim`: Constructs an audit chain linking claim, excerpt, source URL, timestamp and SHA-256 hash.
- `search_entities`: Multi-signal search for canonical entities by name, alias, keyword or domain.
- `get_entity`: Retrieves canonical profile, normalized metadata, connected edges and Schema.org JSON-LD.
- `find_relationships`: Traverses multi-hop relationship paths between entities with edge confidence weighting.
- `find_evidence`: Retrieves compact, ranked supporting evidence for a specific assertion.
- `compare_claims`: Surfaces discrepancies between conflicting statements with citation scoring.
- `explain_entity`: Synthesizes a structured profile backed by verified sources, structured edges and confidence scores.
- `analyze_web_presence`: Audits an entity footprint across domains, GitHub, package registries and docs.
- `audit_entity_consistency`: Verifies cross-platform identifier consistency across domains and repos.
- `export_graph`: Exports verified knowledge graphs into JSON, JSON-LD, CSV or GraphML.

## FAQ

### How does ProofGraph run?

The README says it runs locally over stdio and is started with npx -y @aibuildinfra/proofgraph.

### Which MCP clients are listed as tested?

The product page lists Google Antigravity, Claude Desktop and Cursor IDE, with the version and test date given for each.

### What license is it under?

The README links to an MIT license file in the repository.

### Does ProofGraph guarantee that a claim is true?

No. The product page says a source reference or content hash does not establish real-world truth, and that unindexed claims return UNVERIFIED.

### Is ProofGraph (Antigravity MCP) free?

Yes. ProofGraph (Antigravity MCP) is open source, so you can use and self-host it for free.

### What are the best alternatives to ProofGraph (Antigravity MCP)?

The closest alternatives to ProofGraph (Antigravity MCP) are Graqle, Veritasgraph, FactoGraph, Chaos Cypher and ProofMart.

## Where this comes from

- Description: written from the project's README, the project's website and its MCP Registry entry.

## Alternatives

- [Graqle](https://ai-powered.com/mcp/graqle-2): Graqle is a freemium MCP server for developer tools.
- [Veritasgraph](https://ai-powered.com/mcp/veritasgraph): Veritasgraph is an open-source MCP server for knowledge and memory.
- [FactoGraph](https://ai-powered.com/mcp/factograph): FactoGraph is a free MCP server for knowledge and memory.
- [Chaos Cypher](https://ai-powered.com/mcp/chaos-cypher): Chaos Cypher is an open-source MCP server for knowledge and memory.
- [ProofMart](https://ai-powered.com/mcp/proofmart): ProofMart is a freemium MCP server for news and media.
- [Ontoflow](https://ai-powered.com/mcp/ontoflow): Ontoflow is a freemium MCP server for knowledge and memory.
- [Knowledge MCP](https://ai-powered.com/mcp/knowledge-mcp): Knowledge MCP is an MCP server for knowledge and memory.
- [Graphicle](https://ai-powered.com/mcp/graphicle): Graphicle is an open-source MCP server for developer tools.

---
Source: https://ai-powered.com/mcp/proofgraph-antigravity-mcp
