6 Open-Source AI Lead Generation Agents on GitHub for 2026

An ai lead generation agent github project is free to fork, but running one still means paying for an LLM, a data source, and the hours it takes to keep it working. This guide compares six repos on what they actually need, what data they can reach, and how far they get from a prospect list to a sent email.

Every star count, license, and commit date below was read from GitHub on October 13, 2026. Setup details come from each project's README, and we did not run the code unless we say so. For the broader category beyond GitHub, start with our guide to the AI lead generation agent.

What a Free Lead Gen Agent Really Costs

Open-source lead generation agents are free software, not free pipeline. The code is a workflow; the leads still come from somewhere that charges for them.

Four costs show up in almost every repo.

• An LLM. OpenAI, Anthropic, OpenRouter, or a local model through Ollama. Local models cut the bill but not the setup.
• A data source. A licensed lead database, an email finder, or an enrichment provider. This is usually the real line item, and it is why waterfall enrichment across several providers beats one source.
• A sending path. A warmed mailbox or sequencer, plus the deliverability work that comes with it.
• Maintenance. Provider APIs change, rate limits move, and a repo with one maintainer can go quiet. Plan for that before you build a workflow on it.

If you would rather compare finished products than repos, see our roundups of AI lead gen software and free AI lead generation tools.

How We Checked Each Repo

We opened each repository on GitHub and recorded stars, license, language, and the date of the latest commit through the GitHub API on October 13, 2026. We excluded archived repos and anything without a commit in the last 12 months.

Setup requirements are quoted from each README. "README says" means exactly that: we read the documentation and did not install or run the project. Star counts move daily, so treat them as a snapshot.

1. SyncGTM MCP Server

The SyncGTM lead-gen and enrichment MCP repo is a documentation and reference repository for a hosted MCP server that gives Claude, Cursor, and other AI agents lead search, enrichment, and email finding. It is best for people who want an agent to pull real lead data without wiring up provider keys.

We should be clear about what is open source here. The GitHub repo is licensed Apache-2.0 and contains the tool reference, credit costs, and use-case examples; the server runs at https://api.syncgtm.com/mcp and requires a SyncGTM account with credits. With 0 stars, this is the least community-tested listing here.

The docs list 50+ tools. The ones that matter for lead generation are find_people, find_work_email, verify_email, search_linkedin_job_openings, and enrich_organization. Per the SyncGTM MCP docs, find_people costs 0.2 credits per profile found, find_work_email 1 credit, verify_email 0.1 credits, and enrich_organization 2 credits.

Pros

• No provider accounts or API keys to manage; auth is a browser sign-in.
• Works in any client with remote MCP support, including Claude Code, the Claude app, and Cursor.
• Covers search, work email, verification, hiring signals, and company enrichment from one connection.

Cons

• The server is hosted and paid, so it is not self-hostable open source.
• It stops at data. You still choose the sequencer and write or generate the emails.
• The GitHub repo has no stars and no community yet.

Best for: teams that already work in Claude or Cursor and want lead data on demand. Cost: free MCP connection, credit-based usage. See the setup walkthroughs for the B2B database MCP and the email finder MCP, or pricing on the SyncGTM pricing page.

2. OpenOutreach

OpenOutreach is a self-hosted Python CLI that takes a product description, finds matching people from a licensed data provider, writes down why each one fits, and emails them from your own mailbox. It is the most complete prospect-to-email flow in this list.

The README says it is browserless and uses no social-network account, so it has no LinkedIn scraping surface. It also ships a Claude Code plugin whose skill teaches Claude to run openoutreach find and read the CSV output.

Pros

• Goes from a one-sentence product description to qualified leads to sent email in one command.
• Every lead carries a plain-language reason, so you can audit the targeting.
• Search is free; the README says a paid lookup happens only to resolve a verified work email for the best-fit leads.
• Exports a CSV shaped for Instantly and Smartlead imports.

Cons

• GPL-3.0 is copyleft, which matters if you plan to build it into a product.
• Coverage depends on one data provider, BetterContact. The README itself notes the learning loop is an experiment not yet shown to beat random selection.
• Sends from your mailbox, so deliverability and sender reputation are on you.

Best for: a technical founder who wants a free, auditable find-and-email loop. Cost: free software plus LLM usage, email-finder credits, and your mailbox.

3. AI Sales Team for Claude Code

AI Sales Team for Claude Code is a set of 14 Claude Code skills and 5 sub-agents that research a prospect from a URL, score it with BANT and MEDDIC, map decision-makers, and draft outreach sequences. It is best for account research and message drafting, not for building a contact list.

The README's commands such as /sales prospect <url> write results to Markdown files, and it lists email sequences as drafts. It does not describe a lead database or a sending integration.

Pros

• Lowest setup effort here: one installer and Claude Code.
• Covers the research and qualification steps that most outbound agents skip.
• MIT license and a large star count for a skills repo.

Cons

• No verified contact data source is listed, so decision-maker results depend on what public pages and the model can find.
• Does not send email, so you still need an email finder, a verifier, and a sequencer.
• Last commit was about seven months before our check.

Best for: founders and reps who live in Claude Code and want research briefs per account. Cost: free repo plus your Claude usage. Pair it with a data source; our guides on building a Claude Code lead agent and lead generation with Claude show how.

4. AgentSDR

AgentSDR is a self-hosted Next.js and PostgreSQL workspace for email, LinkedIn, and WhatsApp outreach, with an AI CRM that classifies replies and drafts answers. It is the most complete outbound workspace here, and the heaviest to run.

Its enrichment tables connect to 15 providers, and the README names Apollo, Hunter, Lusha, Findymail, and ZeroBounce among them. You bring your own keys for each one you use.

Pros

• One workspace for leads, enrichment, sequences, and replies across three channels.
• MIT license, active commits, and no per-seat fees; you pay your server and providers.
• Reply classification and drafted responses go beyond send-only agents.

Cons

• Many moving parts: a database, a server, and several third-party accounts.
• LinkedIn and WhatsApp outreach runs through Unipile, which carries platform terms-of-service risk (see the compliance section).
• It works on leads you import or build; it is not a lead database itself.

Best for: a team that wants to own its outbound stack end to end. Cost: free software plus hosting and the providers you connect.

5. B2B SDR Agent Template

The B2B SDR Agent Template is a Markdown-and-shell template for an AI sales development rep that works WhatsApp, Telegram, and email, built for B2B export businesses. It runs on the OpenClaw agent gateway, and the README also offers a variant for Hermes Agent.

Instead of a lead database, it defines a 10-stage pipeline in seven context files: inbound lead capture, BANT qualification, CRM entry, web research, quotes, follow-ups, and cold email sequences.

Pros

• A full sales pipeline from first message to quote, not just prospecting.
• Active maintenance and a test and health-check script before you deploy.
• Strong fit for messaging channels that email-only tools ignore.

Cons

• Aimed at inbound chats and export sales; it is not a prospect-list builder.
• Needs a server, an agent runtime, and a WhatsApp number, so it is the most operational setup in this list.
• Automated WhatsApp outreach has its own platform rules and opt-in expectations.

Best for: exporters and distributors who sell over WhatsApp. Cost: free template plus server, LLM, and messaging costs.

6. Bricks

Bricks is a local-first enrichment spreadsheet with AI columns, a web research agent, an email finder, and a two-agent outreach writer. It is best for researchers who want to enrich a list they already have without a monthly subscription.

The README describes an email finder that generates likely patterns and validates them through Hunter, MillionVerifier, and QuickEmailVerification, and a Puppeteer engine that scrapes from your own IP. It also supports local models through Ollama.

Pros

• Close to free if you use Ollama and the free tiers the README lists.
• Data stays on your machine.
• The writer-versus-prospect agent loop is an unusual quality check on outreach copy.

Cons

• No license file, so you have no explicit right to reuse or redistribute the code.
• Pattern-guessing plus verification finds fewer emails than a licensed provider, and the README's volume figures are its own claim.
• Browser scraping can break site terms; it does not send email either.

Best for: solo operators enriching lists on a near-zero budget. Cost: free, plus whichever API tiers you exceed.

Comparison Table

Stars, license, and last commit are from GitHub on October 13, 2026. "Reaches sent email" means the README describes sending, not just drafting.

Which AI Lead Generation Agent GitHub Project Should You Pick?

Pick by the gap you need filled, not by star count. Most of these projects solve one slice of the pipeline.

• You want leads and a first email with the least setup: OpenOutreach.
• You want to own a full outbound stack: AgentSDR, if you can run Docker and PostgreSQL.
• You already work in Claude Code and need account research: AI Sales Team for Claude Code, plus a data source.
• You need verified contact data inside an AI agent: the SyncGTM MCP, accepting that it is hosted and credit-based.
• You sell over WhatsApp: the B2B SDR Agent Template.
• You have a list and no budget: Bricks, after checking the licensing risk.

You can also combine them. A research skill plus an MCP data source plus a sequencer covers what no single repo does; our AI lead finder guide covers the data side, and the small business AI lead tools roundup covers managed options.

Scraping, ToS, and Compliance Risk

Open source does not change what you are allowed to do with the data. Scraping and automating LinkedIn is the biggest risk across these repos.

LinkedIn's Prohibited software and extensions page says third-party crawlers, bots, and extensions that scrape or automate activity on its site violate the User Agreement and can lead to restricted accounts. AgentSDR's LinkedIn channel and Bricks' browser scraping both sit in that territory, so use a secondary account only if you accept the risk, or skip those features.

Cold email adds its own rules, including GDPR in the EU and CAN-SPAM in the US. For LinkedIn-sourced data without the automation risk, read how a LinkedIn scraper MCP handles it.

Conclusion

An open-source AI lead generation agent saves you license fees and costs you integration time. Match the repo to the step you are missing: research, data, or sending.

If your agent needs lead search, work emails, and verification without managing providers, connect the SyncGTM MCP and see how far one sign-in gets you. Credits and tools are listed in the SyncGTM MCP docs, and the pricing page has plan details.