Skip to content

narainio/context-daemon

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 

Repository files navigation

Context Daemon

An opinionated product development pipeline for builders who use AI agents. Structures the journey from idea to shipped product through seven sequential stages.

Research → Strategy → Gauntlet → Design → PRD → Build → Ship

Each stage is a guided divergent/convergent exercise. Context accumulates across stages through file artifacts. Bad ideas get killed early. Good ideas ship with full context.

Install

npx skills add narain/context-daemon

Or install manually by copying the skills/ directory contents into your agent's skills folder.

Agent Skills path
Claude Code ~/.claude/skills/
Claude Cowork Via skill/plugin install
OpenAI Codex ~/.codex/skills/
GitHub Copilot .github/skills/
Cursor .cursor/skills/ or project root
Google Antigravity .agent/skills/
Amp Code .agents/skills/
Mistral Vibe .vibe/skills/

What's included

context-daemon — The full pipeline skill. Detects your current stage, loads the right guide, and walks you through structured exercises at each step.

skills/context-daemon/
├── SKILL.md                # Router — detects stage, loads guides
├── guides/                 # Stage-by-stage guided exercises
│   ├── 01-research.md
│   ├── 02-strategy.md
│   ├── 03-gauntlet.md
│   ├── 04-design.md
│   ├── 05-prd.md
│   ├── 06-build.md
│   └── 07-ship.md
├── templates/              # Output templates for each stage
└── references/             # Pipeline overview, Double Diamond mapping

falsification-gauntlet — Standalone adversarial stress-testing for product ideas. Five phases: Thesis Inversion, Moat Analysis, Technical Feasibility, Competitive Landscape, Decision. Can be used independently or as part of the full pipeline.

skills/falsification-gauntlet/
├── SKILL.md                # Stress-testing framework
└── references/
    └── templates.md        # Phase output templates

How it works

  1. Start a conversation and tell your agent you want to take an idea through the pipeline
  2. The daemon scans your project folder for stage artifacts to determine where you are
  3. It presents the pipeline map and offers to proceed to the next stage
  4. Each stage produces a Markdown file that feeds the next — context accumulates, never degrades

The pipeline includes explicit kill points. If an idea isn't working, the process kills it early — before you've wasted build time on something that was never going to ship.

Learn more

License

Apache 2.0 — see LICENSE.

About

An opinionated product development pipeline for builders who use AI agents.

Topics

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors