Monitor RSS feeds, websites, and GitHub repos, then get relevance scored summaries on Telegram.
Build multi step automation workflows with conditional branching and automatic retries.
Route each task to the cheapest capable AI model, including fully local models.
Let the agent plan and execute multi step goals autonomously, asking for input when stuck.
| thatsme/alexclaw | carterperez-dev/exs-cyberjob-scraper | elixir-vibe/vibe_kit | |
|---|---|---|---|
| Stars | 114 | 21 | 16 |
| Language | Elixir | Elixir | Elixir |
| Setup difficulty | hard | easy | easy |
| Complexity | 4/5 | 2/5 | 2/5 |
| Audience | developer | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires your own infrastructure including a PostgreSQL database with vector search support.
AlexClaw is a self hosted personal AI agent built on Elixir and OTP, a programming platform known for building highly reliable, concurrent systems. It runs entirely on your own hardware, so your data stays private, and it is designed for a single user, meaning you, rather than as a shared platform or service. The agent watches the world on your behalf across RSS feeds, websites, GitHub repositories, and other APIs. It gathers this information into a searchable knowledge base and runs scheduled workflows automatically without you needing to trigger them manually. You communicate with it through Telegram or Discord. One notable feature is its multi model routing system. Rather than always calling the same AI model for every task, AlexClaw classifies each task by complexity, light, medium, heavy, or local, and picks the cheapest model capable of handling it, including fully local models that run on your own machine with no internet connection required. Workflows are multi step pipelines you define to automate recurring tasks, for example fetching an RSS feed, scoring the items for relevance with an AI call, then sending a summary to Telegram. The system supports conditional branching between steps, retry logic, and circuit breakers, which automatically disable a step that keeps failing and turn it back on after a cooldown period. The agent also offers a reasoning mode where it plans and carries out multi step goals on its own, checks its own results against a rubric, and asks you for input when it gets stuck. Memory is stored in a PostgreSQL database with vector search for retrieving relevant context by meaning rather than exact keywords. An MCP server, a standardized protocol for AI tool integration, lets external tools such as Claude Code connect to AlexClaw's capabilities directly. The full README is longer than what was shown.
A self hosted, single user AI agent built on Elixir that monitors feeds and APIs, runs automated workflows, and reasons through multi step goals.
Mainly Elixir. The stack also includes Elixir, OTP, PostgreSQL.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.