PushCI
AI-native, zero-config CI/CD. Detects 33 languages + 40 frameworks, generates pipelines, runs locally at $0 cloud cost, diagnoses failures with AI, and deploys to 20 targets.
README
pushci
AI-native, zero-config CI/CD that runs on your machine. Detects 33 languages and 40+ frameworks, writes the pipeline for you, runs locally at $0 cloud cost, and ships with a production MCP server so AI agents (Claude, Cursor, Windsurf, OpenAI, Gemini) can drive pipelines on your behalf.
- Website — https://pushci.dev
- Docs — https://pushci.dev/docs
- Dashboard — https://app.pushci.dev
- API — https://api.pushci.dev
- MCP discovery — https://pushci.dev/.well-known/mcp.json
- Issues — https://github.com/finsavvyai/pushci-cli/issues
Quick start
# Install (bundled binary, no network fetch on install)
npm install -g pushci
# One-command setup in your repo
cd your-repo
pushci init # detects stack → generates pushci.yml → wires pre-push hook
# Run the pipeline locally — free, no cloud minutes
pushci run
# Explain a failure with AI
pushci diagnose "ELIFECYCLE Test failed. See above for more details."
# Mutate your pipeline by plain English (v1.7.4+)
pushci extend "add e2e stage with playwright"
Works on macOS (Intel + Apple Silicon), Linux (amd64 + arm64), and Windows (amd64 + arm64).
MCP server — for AI agents
PushCI exposes its CLI surface over the Model Context Protocol so AI agents can plan, run, diagnose, and promote pipelines autonomously.
Connect
Launch the server over stdio — no daemon, no port:
npx pushci mcp
Claude Desktop / Cursor / Windsurf config
{
"mcpServers": {
"pushci": {
"command": "npx",
"args": ["pushci", "mcp"]
}
}
}
Tools exposed
| Tool | Description |
|---|---|
pushci_init |
Auto-detect stack and generate pipeline config for 33 languages, 40+ frameworks, 20 deploy targets |
pushci_run |
Execute the pipeline locally — free, no cloud compute |
pushci_status |
Inspect the most recent pipeline run |
pushci_doctor |
Diagnose environment issues (Docker, Node, Go, etc.) |
pushci_diagnose |
AI-powered error analysis for failed checks |
pushci_scan |
Security scan of pipelines (heuristic + AI analyzers, SARIF 2.1.0 export) |
pushci_recommend |
Best-fit deploy-target and integration suggestions for the current repo |
pushci_promote |
Register with AI registries (MCP list, OpenAI Apps, Claude directory) and search engines |
Discovery manifest: /.well-known/mcp.json.
Install options
# npm — recommended, ships 6 prebuilt binaries in the tarball (~8MB each)
npm install -g pushci
# Homebrew — macOS + Linux
brew install finsavvyai/tap/pushci
# curl — any POSIX shell, downloads from GitHub Releases
curl -fsSL https://pushci.dev/install.sh | sh
# npx — zero install, prints the CLI
npx pushci init
All paths resolve to the same signed, platform-specific binary. The npm tarball contains all six architectures so sandboxed environments (Claude Code sessions, Cursor, CI) work without network.
Features
- Zero-config init — scan repo, detect language + framework + build tool + test runner + deploy target, generate
pushci.ymlin seconds - Local-first runs — checks run on your laptop or self-hosted runner, not burned cloud minutes
- AI everywhere — Anthropic, Groq, DeepSeek, OpenAI, Gemini, local Llamafile (auto-selected by env var)
- Multi-CI bridges — GitHub Actions, GitLab CI, CircleCI, Jenkins, Bitbucket Pipelines, Travis, Buildkite, Drone
- Policy + scan — heuristic rule engine + optional Claude analyzer, SARIF 2.1.0 export for GitHub Security tab
- GitHub Actions runtime — runs your existing
.github/workflows/*.ymlvia embeddedact(no migration) - Deploy automation — 20 first-class targets (Cloudflare, AWS, Fly, Render, Vercel, Netlify, GCP, Azure, etc.)
- Audit chain — tamper-evident HMAC-SHA256 event log, SIEM streaming, 7-year retention on Enterprise
- SSO + SCIM — SAML 2.0 (SP-initiated + IdP-initiated) + SCIM 2.0 provisioning on Team/Enterprise
- Skill marketplace — community pipeline extensions installable with
pushci skill install <name>
Security model
- Transport — all API calls go to
api.pushci.devover TLS 1.3. No plaintext HTTP fallback. - At-rest secrets — pipeline secrets encrypted with AES-256-GCM using machine-bound keys derived from the OS keychain (macOS Keychain, libsecret, Windows Credential Manager). Keys never touch disk in plaintext.
- Auth — JWT with 1-year expiry, rotated on request. SAML + SCIM for Team / Enterprise. MFA-TOTP enrollment available from v1.7+.
- Audit logs — every administrative action recorded with HMAC-chained event hashes; tamper is detectable by replay. Export to Splunk HEC, Elastic, Datadog, or any SIEM.
- SBOM — every release ships with CycloneDX SBOM attached to the GitHub Release assets. Run
cosign verifyto check binary signatures. - CI self-dogfood — PushCI's own pipeline uses PushCI. Coverage: 90%+ line, 85%+ branch. SAST via
gosec, dependency audit viagovulncheck, secret scan viagitleaks. Release-blocking on any Critical or High finding. - No telemetry by default — opt-in via
pushci config set telemetry=on. No call-home, no usage metrics unless enabled. - File-size cap — every Go source file under 100 lines, enforced in CI. Reviewers never face 3,000-line files.
- Vulnerability disclosure —
security@pushci.dev. Response SLA 48h, patch SLA 14 days for Critical/High. See SECURITY.
Pricing
| Plan | Price | Core features |
|---|---|---|
| Free | $0 forever | Unlimited local runs, AI stack detection, 2 deploy targets |
| Pro | $9 / mo | AI diagnosis, 500 cloud minutes, 20 deploy targets, dashboard |
| Team | $29 / seat / mo | SSO/SAML, audit logs, governance, SLA, 2000 cloud minutes |
| Enterprise | from $25 / user / mo + optional $8k / mo Dedicated | SCIM, EU/US/APAC residency, 99.9% SLA, bridges, 7-year audit, self-hosted runners |
Full pricing page: https://pushci.dev/pricing. Enterprise / custom: https://pushci.dev/contact.
What's in this repo
| File | Purpose |
|---|---|
bin/pushci.js |
npm shim — resolves the platform binary from PUSHCI_BINARY env, local dev build, bundled binary, $PATH, GitHub Release download, or go build fallback |
LICENSE |
MIT for the shim contents |
| GitHub Releases | Binary tarballs + CycloneDX SBOMs for linux-amd64, linux-arm64, darwin-amd64, darwin-arm64, windows-amd64, windows-arm64 |
The PushCI product source (Go CLI, Cloudflare Workers API, React dashboard, landing) is proprietary and tracked privately at finsavvyai/pushci. This public repository exists so that:
- Install tools (Homebrew, curl, npm) can download binaries anonymously from GitHub Releases.
- Enterprise security reviews have a GitHub URL they can audit.
- The npm
repositoryfield and MCP registry entries have a stable public URL.
Release binaries are built from the private source via goreleaser on every tag and uploaded here automatically.
Reporting bugs
- Product bugs — runtime errors, wrong pipeline detection, deploy failures:
hello@pushci.devor https://pushci.dev/contact - Install-path bugs —
pushci: command not found, shim errors, Homebrew formula issues: open an issue on this repo - Security —
security@pushci.dev(PGP available on request). Responsible disclosure policy: https://pushci.dev/security
License
MIT for the contents of this repository (the bin/pushci.js shim + release binaries distributed under LICENSE).
The PushCI product itself — Go CLI source, Cloudflare Workers API, React dashboard, and landing page — is proprietary commercial software operated by FinSavvy AI Ltd. Commercial licensing / enterprise deployments: hello@pushci.dev.
Copyright © 2025-2026 FinSavvy AI Ltd.
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.