A genuinely useful self-hosted AI outreach tool held back by sparse documentation and no clear way to verify it works before committing.

Clear positioning, compelling feature list, but the value prop assumes you already know what an 'AI SDR' means.

The README immediately explains what Linki is (open-source AI SDR), who it's for (B2B founders), and why it matters (no per-seat pricing, self-hosted, data stays private). The feature breakdown and demo link are prominent. However, the README doesn't explain what an 'SDR' is or why someone should care about 'AI agent writing messages'—it assumes familiarity with the sales motion and risk tolerance around AI-generated outreach.

Fast path to running the app, but missing prerequisites and deployment reality checks.

Docker setup is clear: `cp .env.example .env.local`, fill three fields, `docker compose up -d`—under 2 minutes. Manual Node.js path is also documented. The .env.local comments are helpful. However, the snapshot shows no .env.example file, meaning a newcomer copying that command will fail. The README also doesn't mention that LinkedIn authentication requires a browser with cookies (a material prerequisite) until step 2, and doesn't clarify whether Apollo, email accounts, or a LinkedIn Sales Navigator subscription are actually mandatory to test the core product.

The payoff is real but obscured by a 7-step setup gauntlet and no quickstart path to see it work.

The core promise is 'build campaigns and AI writes messages for each lead.' Steps 1–7 in the README show the intended path. But there's no minimal example: a first-timer can't import test leads and generate one AI message without setting up LinkedIn auth, email accounts, Apollo enrichment, and an OpenRouter API key. The YouTube demo exists but is not called out as the critical accelerator it should be. A 'get it running in 5 minutes with dummy data' path would close this gap.

Substantial code, but no tests, no example config, and no public issue/PR activity visible.

The codebase is large (TypeScript, Next.js, multiple integrations), suggesting real work. A hosted deployment option (Opsily) adds credibility. However, no test files are visible in the snapshot, no CONTRIBUTING.md, no example .env or sample campaign export, and the single open issue doesn't tell us if maintainers respond quickly. The '48 stars' and 'last pushed 2026-05-21' (future date, likely a clock error in the snapshot) don't convey ongoing health clearly.

Functional but with rough edges: missing .env.example, confusing email/LinkedIn as optional, and no error recovery docs.

The .env.example file is referenced in the README but not visible in the snapshot, breaking the first Docker command. The setup steps list email and Apollo as 'optional' but don't clarify that running a campaign without them means no outreach—confusing for newcomers. There's no troubleshooting guide for common failures (e.g., 'LinkedIn auth failed,' 'SMTP connection timeout'). The LICENSE type is 'Other,' which doesn't inspire confidence; it's actually a sustainable-use license, but that barrier is not explained upfront.

Project shows activity (recent push, hosted offering, homepage) but commit frequency and maintainer responsiveness are unclear.

The repository has a homepage, Docker deployment path, and a hosted partner (Opsily), all signs of active stewardship. The last push timestamp is listed as 2026-05-21 (likely a data error), so recency can't be verified from the snapshot. Only 1 open issue is visible, which could mean either rapid resolution or dormancy; PR response time and issue comment velocity are not evident. No changelog or release notes are present to show iteration.