OpenClaw is just not dangerous enough. I needed something else

Name: OpenClaw is just not dangerous enough. I needed something else
Availability: InStock
Author: kkovacs

by kkovacs·May 17, 2026·2 points·0 comments

Visit Project View on HN

AI Analysis

●●SolidCozyNiche Gem

Filesystem inject directory for cron automation is clever, but it's still a Telegram wrapper.

Strengths

•Inject directory pattern uses filesystem as automation hook—cron writes files, bot processes them.
•Single-user design enforces security boundary; no multi-tenant complexity to maintain.
•Bun compilation to single binary keeps deployment trivial for self-hosters.

Weaknesses

•Tied to Pi AI specifically—users of other agents get zero value from this.
•Telegram bot wrapper around CLI follows the Macabolic pattern—useful but not novel.

Post Description

I'm not saying the world needed another mini-claw, but I needed one. A minimalist self-hosted Telegram bot to talk directly to a [Pi](https://pi.dev) AI agent harness.

I'm not saying you should use it. But if you're a Unix guy like me, maybe you can get inspired, in a world where personal agents include the kitchen sink and a Minecraft server. :)

- Supports switching Pi sessions directly from your phone, to keep context on leash.

- Supports tasking Pi from shell (normal `cron` or `at`) and get response in Telegram.

- You get to run `bash` directly from Telegram, like `!rm -rf /`

- Bring your own Pi, configured exactly your way.

- It's intentionally minimalist like Pi itself. I'm actively removing any feature that can be done easier by bash or by the LLM.

- 100% TypeScript in Bun, compiles to single binary

Small is beautiful (and secure and maintainable).

Similar Projects

SaaS●Mid

Deploy OpenClaw in 60 seconds – one-click setup, no DevOps needed

Slashes OpenClaw setup from 60 minutes to one click, but relies entirely on the underlying open-source tool.

Ship It

Abhinayguptha

213mo ago

Security●●Solid

Clawsec - Open-source plugin for OpenClaw that blocks dangerous actions

Every tool call is caught by middleware, scored against built-in rulesets like destructive-commands, secrets/* and exfiltration/* in under 5ms, then enforced as block/confirm/allow via a clawsec.yaml — neat, pragmatic attack surface reduction. The demo and auto-generated config make onboarding trivial, but it currently reads as an OpenClaw-first solution; broader agent-framework integrations or stronger isolation guarantees would make this a must-install.

Solve My ProblemNiche GemSlick

subho007

114mo ago

SaaS●Mid