grainulator

grainulation/grainulator: Research that compiles. · GitHub

by

Glitchie – Granulator mascot

version
demo
license
CI
Documents on DeepWiki

Research Sprint Orchestrator for Cloud Code.

Ask a question, get a ready-made brief for a decision. Each conclusion is tracked as a typed claim, adversarially challenged, confidence-graded, and compiled into a self-contained output. Zero third-party dependencies.

granulator.app – Interactive demo with in-browser AI. No install required.

Step 1 – Add Market

claude plugin marketplace add https://github.com/grainulation/grainulator/blob/main/.claude-plugin/marketplace.json

Step 2 – Install the Plugin

claude plugin install grainulator

Requirements: Node.js >= 20 (required for MCP servers running via NPX).

Troubleshooting: SSH permission denied

if you see git@github.com: Permission denied (publickey)The plugin installer is attempting to SSH but you do not have an SSH key set up with GitHub. Fix with one command:

git config --global url."https://github.com/".insteadOf "git@github.com:"

then try again claude plugin install grainulator. This tells git to use HTTPS instead of SSH for all GitHub repos.

Alternatively, clone manually:

git clone https://github.com/grainulation/grainulator.git ~/.claude/plugins/grainulator
claude plugin add ~/.claude/plugins/grainulator

For team-wide deployment, commit to your project .claude/settings.json: :

{
  "enabledPlugins": ["grainulator@grainulation-marketplace"]
}

Once installed, simply talk to the cloud:

  • “Research how our authentication system works” – Runs a multi-pass research sprint
  • “challenge r003” – adverse testing of a specific claim
  • “What are we missing?” – Blind Spot Analysis
  • “Write this” – prepares a compiled brief

No slash syntax required. Intent Router figures out what you want and runs the right workflow.

Claim It is a unit of knowledge. Each finding from research, challenges, witnesses, and prototypes is stored as a typed claim claims.json.

Claim Type What does it mean
constraint Strict requirements, non-negotiable limits
factual verifiable statements about the world
estimate Estimates, Approximations, Categories
risk Potential Failure Modes, Concerns
recommendation Proposed course of action
feedback Stakeholder Input, Opinion

evidence level Grade Confidence: stated → web → documented → tested → production.

compiler Runs 7 passes on your claims – type coverage, evidence strength, conflict detection, bias scan – and generates a confidence score. If there are any unresolved conflicts, it blocks the output until you resolve them.

Skill Description
/init Start a new research project
/research Multi-stage investigation with evidence collection
/challenge adverse trial of a claim
/witness Verify a claim against an external source
/blind-spot structural gap analysis
/brief Prepare a concise decision brief
/present Prepare a Presentation Deck
/status sprint dashboard snapshot
/pull Import knowledge from external sources (DeepWiki, Sangam)
/sync Publish artifacts to external targets
/calibrate Score prediction against actual results
/resolve adjudicate disputes between claims
/feedback Record stakeholder input

granulator subagent (agents/grainulator.md) Runs multi-pass research sprints autonomously. It reads the compiler output to decide which commands to run next – research, challenge, witness, blind-spot – until the sprint reaches decision-ready confidence.

Launch it from Cloud Code: "research X using grainulator".

granulator.app is a PWA that displays the Research Sprint concept in the browser:

  • Mobile-first chat interface – no shaders, dark background on mobile
  • In-browser AI via WebLLM (SmolLM2-360M, ~200MB download)
  • 50 pre-made demo themes with fuzzy matching (Instant Track A)
  • Live local estimation when model is downloaded (Track B)
  • Interactive Responses with Progressive Claim Disclosure
  • Compile flow with 7-pass confidence scoring
  • plugin manifest: : .claude-plugin/plugin.json
  • mcp server: Wheat (claims engine), Mill (format conversion), Silo (knowledge repository), DeepWiki (codebase research)
  • Skill: : skills//SKILL.md – 13 instant-engineered workflows
  • Representative: : agents/grainulator.md – Autonomous Sprint Subagent
  • hooks: auto-compile on claim mutation, write-guard on claims.json And compilation.json
  • Orchard: Via Multi-Sprint Orchestration orchard.json dependency graphs

Part of the granular ecosystem. Eight tools, each performing a function.

tool what does it do to install
wheat Research Engine – Develop Structured Evidence npx @grainulation/wheat init
Farmer Permissions Dashboard – Approve AI tasks in real-time npm i -g @grainulation/farmer
barn Shared Tools – Templates, Validators, Sprint Identity npm i -g @grainulation/barn
Mill Format conversion – export to PDF, CSV, Slides npm i -g @grainulation/mill
silo Knowledge Storage – Reusable Claim Libraries and Packs npm i -g @grainulation/silo
Crop Analytics – Cross-Sprint Patterns and Prediction Scoring npm i -g @grainulation/harvest
Orchard Orchestration – Multi-Sprint Coordination npm i -g @grainulation/orchard
granulation Unified CLI – Single Entry Point npm i -g @grainulation/grainulation

You don’t need all eight. start with claude plugin install grainulator. That’s it.

Three levels:

  1. team lead: Committed .claude/settings.json with enabledPlugins for your repo
  2. IT administrator: Deploy managed settings through MDM with pre-approved permissions
  3. air gap: Use CLAUDE_CODE_PLUGIN_SEED_DIR With plugins baked into container images

Each granulation tool runs only on Node built-in. No npm install waterfall. No left-pad. No supply chain worries. MCP servers download the first time you use them npx – No global installation required.

MIT



<a href

Leave a Comment