
Connects your AI agent to the BuildPulse Platform API so you can surface flaky test inventories, CI run history, and coverage metrics without leaving your editor. Exposes five tools including find_flaky_tests with tag and recency filters, get_test_history for disruption events, and get_repo_coverage for the latest snapshot. Ships with four built-in prompts like triage_flaky_tests and whats_red. Every response includes deep links back to the BuildPulse web app. Supports both stdio and hosted streamable HTTP transports, so it works in Claude Desktop, Cursor, Cline, Continue, and other MCP clients. Requires a BuildPulse API token from your organization settings. Written in Go, distributed as an npm package that pulls the native binary for your platform.
Model Context Protocol server for the BuildPulse Platform API. Surface flaky tests, CI run history, and coverage health in Claude Desktop, Cursor, ChatGPT, Cline, Windsurf, Continue, Zed, VS Code Copilot, and any other MCP-aware AI agent.
npx -y @buildpulse/mcp
Or pin globally:
npm install -g @buildpulse/mcp
The package downloads the matching native binary for your platform on first install. Supported platforms: macOS arm64/x64, Linux arm64/x64, Windows x64.
Get a BuildPulse API token at https://buildpulse.io → Organization Settings → API Tokens.
~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"buildpulse": {
"command": "npx",
"args": ["-y", "@buildpulse/mcp"],
"env": { "BUILDPULSE_TOKEN": "your-buildpulse-api-token" }
}
}
}
.cursor/mcp.json (per-project) or ~/.cursor/mcp.json (global):
same JSON shape.
Cline, Continue, Windsurf, Zed, and VS Code Copilot all read an
mcpServers block in their respective config files. See the
install hub for copy-paste
snippets per client.
Your BuildPulse token may grant access to more than one organization.
Every repo-scoped tool takes an optional organization_id argument (the
org's id UUID, discoverable via list_my_organizations):
organization_id. It auto-defaults to your
one organization. Nothing changes; you never need to think about orgs.list_my_organizations returns 2+ orgs) — you
must pass organization_id on every repo-scoped call
(list_repositories, find_flaky_tests, get_test_history,
list_recent_submissions, get_submission_test_results,
get_recent_failures, get_repo_flakiness, get_repo_coverage). The
org is not auto-selected — omitting it returns an error that lists
every accessible organization and its UUID, so the agent can pick the
right one and retry. Call list_my_organizations first to enumerate
them.This avoids silently querying the wrong (often empty) organization and getting confusingly empty results.
| Tool | Purpose |
|---|---|
list_my_organizations | Enumerate the organizations this token can access; get the id (UUID) to pass as organization_id. |
list_repositories | List repositories in an organization. |
find_flaky_tests | Search a repository's flaky test inventory; filter by tags, recency, free-text. |
get_test_history | Recent disruption events for a specific test. |
list_recent_submissions | Recent test-result submissions (CI runs) for a repository. |
get_submission_test_results | Per-test results for one submission (one CI run). |
get_recent_failures | Tests that failed across the most recent submissions, aggregated by test identity. |
get_repo_flakiness | Current flakiness % over the last 14 days. |
get_repo_coverage | Current coverage % from the latest report. |
Repo-scoped tools accept an organization_id argument — required for
multi-org sessions, optional (auto-defaulted) for single-org tokens. See
Organizations above.
Every output that names a test or repo includes a web_url deep-link
back to the BuildPulse web app — the same polish Sentry / Atlassian
use in their MCP responses.
The server also ships four guided prompts (slash-pickable in clients that support them):
/triage_flaky_tests/ci_health_check/explain_test_failure/whats_red| Transport | Binary | Where it goes |
|---|---|---|
| stdio | cmd/mcp | npm → npx -y @buildpulse/mcp |
| Streamable HTTP | cmd/mcp-remote | hosted at https://mcp.buildpulse.io/mcp |
Same tool surface; same prompts; same resources. Pick whichever your client supports. The stdio path is universal; the hosted variant is the path to Claude.ai web and ChatGPT.
The server exposes two MCP resource templates so agents can pull state into context without a tool call:
buildpulse://repos/{repo}/flaky-testsbuildpulse://repos/{owner}/{name}/submissions| Variable | Required | Default |
|---|---|---|
BUILDPULSE_TOKEN | yes | — |
PLATFORM_API_URL | no | https://platform.buildpulse.io |
The hosted server (mcp-remote) will refuse to start unless
PLATFORM_API_URL is production or development Platform API. Local stdio
(npx @buildpulse/mcp) is unchanged. See SECURITY.md for
the threat model, tenant isolation, P1 rate limits (120 tool calls / token /
minute), the tool audit log, RFC 7009 /oauth/revoke, and what we
deliberately do not gate (HITL on reads, hiding tools, killing multi-step
triage).
git clone https://github.com/BuildPulseLLC/buildpulse-mcp
cd buildpulse-mcp
go build -o ./bin/buildpulse-mcp ./cmd/mcp
go build -o ./bin/buildpulse-mcp-remote ./cmd/mcp-remote
Requires Go 1.24+.
go test ./...
MIT — see LICENSE.
/docs/mcp — branded install hub with copy buttonsmain / feat/** / fix/** runs build-and-push + deploy (production vs development by branch).BuildPulseLLC/agents (dependabot-rollup workflow / DEV-85): open Dependabot PRs are folded onto feat/deps-dependabot-batch-* so development CI runs before any merge to main. See agents/scripts/dependabot-rollup.md.BUILDPULSE_TOKEN*secret40-character BuildPulse API token. Get one at app.buildpulse.io → Organization Settings → API Tokens.
PLATFORM_API_URLdefault: https://platform.buildpulse.ioOverride the platform API base URL. Default: https://platform.buildpulse.io