g8-mcp-server

graph8 MCP server — GTM operations for AI agents.

Connects AI coding agents (Claude Code, Cursor, Windsurf) to graph8's GTM automation platform via the Model Context Protocol.

Install

# Run directly (recommended)
uvx g8-mcp-server

# Or install globally
pip install g8-mcp-server

Configure

Claude Code

Add to ~/.claude/mcp.json:

{
  "mcpServers": {
    "graph8": {
      "command": "uvx",
      "args": ["g8-mcp-server"],
      "env": {
        "G8_API_KEY": "g8_key_xxxxx",
        "G8_API_URL": "https://api.graph8.com"
      }
    }
  }
}

Cursor

Add to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "graph8": {
      "command": "uvx",
      "args": ["g8-mcp-server"],
      "env": {
        "G8_API_KEY": "g8_key_xxxxx",
        "G8_API_URL": "https://api.graph8.com"
      }
    }
  }
}

Get your API key at app.graph8.com/settings/api-keys.

Tools (370+)

The server exposes 370+ tools spanning CRM, prospecting, enrichment, sequences, campaigns, studio, voice, workflows, appointments, and more. Agents don't load every schema at once — they discover tools at runtime with g8_tool_search, which ranks the catalog by query. The highlights below are a starting point.

GTM Spine (11)

Tool	Description
`g8_connect_repo`	Connect a GitHub/GitLab repository
`g8_scan_repo`	Record the tech stack and GTM readiness you read from the code
`g8_get_scan_results`	Get scan results
`g8_status`	Get repo connection and install status
`g8_doctor`	Report the graph8-side state of a repo's GTM install
`g8_install_spine`	Generate GTM install patch set
`g8_apply_install`	Apply patches (requires confirmation)
`g8_list_campaigns`	List generated campaigns
`g8_get_campaign`	Get campaign details
`g8_launch_campaign`	Launch a campaign (supports dry_run)
`g8_search_kb`	Search the GTM knowledge base

Platform (5)

Tool	Description
`g8_search_contacts`	Search contacts in the CDP
`g8_search_companies`	Search companies by domain or industry
`g8_company_open_jobs`	What CRM companies are hiring for now (growth signal, not the marketplace)
`g8_lookup_person`	Instant person enrichment
`g8_lookup_company`	Instant company enrichment
`g8_add_to_sequence`	Add contacts to a sequence (requires confirmation)

Resources (90+)

The server exposes 90+ resources — read-only context agents can include in prompts. Examples:

g8://repos — connected repositories
g8://repos/{id}/scan — scan results
g8://repos/{id}/kb — knowledge base documents
g8://repos/{id}/kb/{doc_id} — individual KB document
g8://repos/{id}/campaigns — campaign list
g8://repos/{id}/campaigns/{id}/brief — campaign brief
g8://contacts — recent contacts
g8://companies — companies

Prompts (7)

Pre-built workflow templates include:

gtm_setup — full setup: connect → scan → install → launch
campaign_review — review and improve a campaign
icp_refinement — refine ICP profiles from feedback

Local stdio bridge

stdio_bridge.mjs proxies a local stdio MCP client (e.g. Claude Desktop) to the hosted graph8 MCP HTTP endpoint:

G8_API_KEY=g8_live_xxx node stdio_bridge.mjs

It targets production (https://be.graph8.com/mcp/) by default. Override the endpoint with G8_MCP_URL to point at another environment:

# QA
G8_MCP_URL=https://be.qa.graph8.com/mcp/ G8_API_KEY=g8_test_xxx node stdio_bridge.mjs
# local dev server
G8_MCP_URL=http://localhost:8000/mcp/   G8_API_KEY=g8_test_xxx node stdio_bridge.mjs

Env var	Default	Purpose
`G8_MCP_URL`	`https://be.graph8.com/mcp/`	Remote MCP HTTP endpoint the bridge proxies to
`G8_API_KEY`	—	Bearer token sent on every request

The bridge preserves the Streamable HTTP Mcp-Session-Id returned during initialization, so subsequent notifications and tool calls stay attached to the same remote session.

Development

cd mcp_server
pip install -e ".[dev]"
pytest

# stdio bridge unit tests (zero-dep, Node's built-in runner)
node --test tests/test_stdio_bridge.mjs

Development Status
- 4 - Beta
Intended Audience
- Developers
Programming Language

Metadata

Development Status
- 4 - Beta
Intended Audience
- Developers
Programming Language

Release files for g8-mcp-server 0.69.8

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for g8-mcp-server 0.69.8
File	Size	Uploaded	Provenance
g8_mcp_server-0.69.8.tar.gz	1.7 MB	Sep 11, 2026	With provenance	Details

Built distribution (wheel)

Table of built distributions (wheels) for g8-mcp-server 0.69.8
File	Interpreter	ABI	Platform	Reset
g8_mcp_server-0.69.8-py3-none-any.whl 985.6 kB Sep 11, 2026 With provenance	Python 3	none	any	Details

Total release size: 2.7 MB

Release files / g8_mcp_server-0.69.8.tar.gz

Download URL	g8_mcp_server-0.69.8.tar.gz
Size	1.7 MB
Tags	Source
SHA-256 checksum How to use checksums	`7abb5c748d19ffcec1b75721d2ac665dd92f0491cb115b993acc1a7823afb06b`
BLAKE2b-256 checksum How to use checksums	`0bbd5e8acccfdd9ee3fe47a0c743c3e03676b59fdd2abf30259c475cdbfa8a9c`
Upload date	Sep 11, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/graph8-com/g8
Publishing commit: github.com/graph8-com/g8/tree/198f09498a2f1ea0e956a8fa4367b45753f0dd6f

Workflow

Publishing configuration: github.com/graph8-com/g8/blob/198f09498a2f1ea0e956a8fa4367b45753f0dd6f/.github/workflows/publish-mcp-server.yml
Publishing logs: github.com/graph8-com/g8/actions/runs/34636167909/attempts/1

Artifact

Subject: g8_mcp_server-0.69.8.tar.gz
SHA-256 checksum: 7abb5c748d19ffcec1b75721d2ac665dd92f0491cb115b993acc1a7823afb06b Verifying attestations

Release files / g8_mcp_server-0.69.8-py3-none-any.whl

Download URL	g8_mcp_server-0.69.8-py3-none-any.whl
Size	985.6 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`786fd37633525a652c262a2267051e55c2045bb830370613017b8a01121931c3`
BLAKE2b-256 checksum How to use checksums	`88dc06ca81ed4067ec79582f73d5ce511de50817ce33d0e35e7464b062fe99ed`
Upload date	Sep 11, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/graph8-com/g8
Publishing commit: github.com/graph8-com/g8/tree/198f09498a2f1ea0e956a8fa4367b45753f0dd6f

Workflow

Publishing configuration: github.com/graph8-com/g8/blob/198f09498a2f1ea0e956a8fa4367b45753f0dd6f/.github/workflows/publish-mcp-server.yml
Publishing logs: github.com/graph8-com/g8/actions/runs/34636167909/attempts/1

Artifact

Subject: g8_mcp_server-0.69.8-py3-none-any.whl
SHA-256 checksum: 786fd37633525a652c262a2267051e55c2045bb830370613017b8a01121931c3 Verifying attestations

Security for g8-mcp-server 0.69.8

Report project as malware

Provenance

Full provenance

All 2 files in this release have been uploaded with provenance metadata.

Attestation for g8_mcp_server-0.69.8.tar.gz

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/graph8-com/g8
Publishing commit: github.com/graph8-com/g8/tree/198f09498a2f1ea0e956a8fa4367b45753f0dd6f

Workflow

Publishing configuration: github.com/graph8-com/g8/blob/198f09498a2f1ea0e956a8fa4367b45753f0dd6f/.github/workflows/publish-mcp-server.yml
Publishing logs: github.com/graph8-com/g8/actions/runs/34636167909/attempts/1

Artifact

Subject: g8_mcp_server-0.69.8.tar.gz
SHA-256 checksum: 7abb5c748d19ffcec1b75721d2ac665dd92f0491cb115b993acc1a7823afb06b Verifying attestations

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.