calebwin

calebwin / pgclaw

Public

A "Clawdbot" in every row with 400 lines of Postgres SQL

206
11
100% credibility
Found Feb 13, 2026 at 124 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

pgclaw is an open-source PostgreSQL extension that introduces a 'claw' data type for embedding AI agents directly into database table rows, enabling automatic processing with various LLM providers.

How It Works

1
📖 Discover pgclaw

You learn about pgclaw, a clever way to put smart AI helpers directly into your database rows so they can work on your data automatically.

2
🛠️ Add it to your database

You easily add pgclaw to your existing database, getting everything ready in a few simple steps.

3
🤖 Link your AI helper

You connect a smart AI service, like one from Anthropic or OpenAI, so your row helpers can think and make decisions.

4
📊 Create a smart table

You make a table for things like customer tickets, adding a special helper column that holds AI instructions.

5
👀 Turn on watching

You tell pgclaw to keep an eye on your table, so it notices new or changed rows right away.

6
✨ Add your data

You drop in a new ticket, and in seconds, the AI helper reads it, thinks, and updates fields like priority or summary all by itself.

🎉 Smart database magic

Now every row in your table has its own AI brain, automatically handling tasks and keeping your data fresh and insightful.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 124 to 206 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is pgclaw?

pgclaw is a Rust Postgres extension that packs a "Clawdbot" AI agent into every row via a `claw` data type, using just 400 lines of Postgres SQL for triggers, queues, and agent configs. Add a claw column with a prompt or agent reference, call `claw_watch` on your table, and inserts/updates auto-trigger LLM calls to analyze row data and patch columns via JSON responses. It leverages Postgres ACID, joins, and backups while supporting any LLM provider through rig, from Anthropic to Ollama.

Why is it gaining traction?

It turns AI into a native DB primitive—no separate services or event buses needed—letting agents update rows, maintain memory, or even spawn Claude Code workspaces for file/tool access per row. Devs dig the zero-boilerplate flow: define reusable agents in `claw.agents`, route channels like Telegram via `claw.route`, or schedule heartbeats/crons. At 46 stars, the hook is embedding stateful AI directly in SQL workflows.

Who should use this?

Backend engineers on Postgres stacks building AI-driven CRUD apps, like auto-classifying support tickets, summarizing logs, or processing task queues. Suited for solo devs or small teams prototyping agentic databases without Kubernetes-scale infra.

Verdict

Intriguing for AI-in-DB experiments with good docs and tests, but 46 stars and 1.0% credibility score mean it's pre-prod—prototype now, monitor for stability.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.