No red flags found in any of the 11 categories — no credential harvesting, no data exfiltration, no curl-pipe-shell installer. Scanned against the SlowMist agent-security taxonomy, refreshed every 8 hours. Full audit →
by coleam00 · Agent Tool · ★ 116
Last updated: · Indexed by AgentSkillsHub · Auto-synced every 8h
🔒 Is dark-factory-experiment safe to install? View the security audit →
The Dark Factory Experiment (RAG YouTube Chat App) A public Dark Factory experiment. This repository is a working web application that is built, reviewed, and merged almost entirely by AI coding agents. Humans do two things: file issues and promote releases. Everything in between - triage, implementation, code review, testing, merging - is handled by Archon workflows running on a cron. Two honest caveats, because they are the design and not an asterisk. This runs at level 4, not level 5: the factory does not write its own issues. And there is a deliberate human-authored perimeter it is never allowed to touch - auth, rate limiting, the deploy configs, and the three governance files that define its own rules. The list is in , and a PR touching any of it is auto-rejected before anything else is evaluated. An autonomous system is only as trustworthy as the things it cannot change about itself. The application itself is a dark-mode AI chat app that lets you have grounded conversations about a creator's YouTube videos, with cited answers pulled from transcript passages. But the real point of this repo is the factory that builds it.
| Stars | 116 |
| Forks | 33 |
| Language | Python |
| Category | Agent Tool |
| Quality Score | 67.5089390234369/100 |
| Open Issues | 1 |
| Last Updated | 2026-08-14 |
| Created | 2026-04-02 |
| Platforms | claude-code, python |
| Est. Tokens | ~16k |
These tools work well together with dark-factory-experiment for enhanced workflows:
Looking for a dark-factory-experiment alternative? If you're comparing dark-factory-experiment with other agent tool tools, these 6 projects are the closest alternatives on Agent Skills Hub — ranked by topic overlap, star count, and community traction.
Spec-driven development with smart compaction. Claude Code plugin combining Ralph Wiggum loop with structured
Safe runtime for autonomous on-chain AI agents: isolated sandboxes, Library skills, encrypted secrets.
Self-hosted framework for orchestrating fleets of specialist AI agents — ensemble reasoning and a full agentic
The complete guide to Claude Code: CLAUDE.md, hooks, skills, MCP servers, and commands
One-stop handbook for building, deploying, and understanding LLM agents with 60+ skeletons, tutorials, ecosyst
SAW — SAFe Agentic Workflow AI Agent Harness for Multi-Agent Team Workflows Built on SAFe methodology (Scaled
Explore other popular agent tool tools:
dark-factory-experiment is A repository that ships its own code. AI workflows triage issues, implement them, review, and auto-merge with no human reading the diff. Runs on Archon. The app it maintains is a cited RAG chat over Y. It is categorized as a Agent Tool with 116 GitHub stars.
dark-factory-experiment is primarily written in Python. It covers topics such as agentic-coding, ai-agents, archon.
You can find installation instructions and usage details in the dark-factory-experiment GitHub repository at github.com/coleam00/dark-factory-experiment. The project has 116 stars and 33 forks, indicating an active community.
The top alternatives to dark-factory-experiment on Agent Skills Hub include smart-ralph, sandboxed.sh, captain-claw. Each offers a different approach to the same problem space — compare them side-by-side by stars, quality score, and community activity.
Grades come from a rule-based scan built on the SlowMist agent-security taxonomy, covering 11 red-flag categories including credential harvesting, data exfiltration, and curl | sh installers. It is a first-layer scan, not a manual audit — we say so rather than overstate it.
The scale of the problem is documented independently: Liu et al. (2026), in a study of 31,132 agent skills, report that 26.1% contain security vulnerabilities. Our own full-catalog census is published as a citable open dataset.
Sources & who's responsible: