Your Complete Guide to Fluently

Learn how to use the 4D Framework, the fluent CLI, and contribute to the shared knowledge base.

The 4D Workflow CLI Commands AI Journey Contribute
The 4D Workflow

How a Fluently 4D Cycle Works

Every AI-assisted task goes through four dimensions. Fluently helps you apply them consistently.

1
🤝

Delegation

Who does what?

  • • Should AI do this fully?
  • • Should AI assist, human decide?
  • • What requires human judgment?
2
📝

Description

What does AI need to know?

  • • Provide context and scope
  • • Give examples and constraints
  • • Reduce ambiguity
3
👁️

Discernment

Is this output trustworthy?

  • • Spot hallucinations
  • • Check for overconfidence
  • • Verify against sources
4
✅

Diligence

Who is accountable?

  • • Define review process
  • • Create audit trail
  • • Set escalation triggers

Full Task Journey with Fluently

💡
Identify AI Task
Something you want AI to help with
🔍
Check Knowledge Base
fluent score "task"
⚡
Apply 4D Framework
Follow the guidance from similar cycles
🤖
Execute with AI
Use your briefed Delegation + Description
🔬
Evaluate Output
Apply Discernment checks
📋
Close the Loop
Document + approve (Diligence)
CLI Reference

Using the fluent CLI

Install once, use everywhere: npm install -g fluently

Typical Workflow

# 1. Browse what exists
fluent list
fluent list coding
# 2. Score your task
fluent score "automate bug triage"
# 3. Deep compare
fluent compare \
  --description "..." \
  --delegation "augmented"
list Browse cycles
fluent list [domain]

Lists all Fluently 4D cycles in the knowledge base. Filter by domain: coding, writing, research, etc.

fluent list coding → Shows 4 cycles
score Evaluate a task
fluent score "your task description"

Compares your task to the knowledge base and returns a 4D score (Delegation, Description, Discernment, Diligence) with top matching cycles.

Returns top 3 similar cycles + scores
compare Deep analysis
fluent compare --description "..." --delegation "augmented"

Detailed 4D comparison with specific delegation intent. Provides dimension-by-dimension suggestions and improvement tips.

contribute Add your cycle
fluent contribute

Interactive wizard to create a new Fluently 4D cycle entry. Validates your YAML against the schema before you open a PR.

sync Update knowledge
fluent sync

Updates your local knowledge base from the community repository. Run periodically to get the latest cycles contributed by the community.

--help Quick reference
fluent --help

Shows all commands and options. Also works per-command: fluent score --help.

Example: fluent score "automate code review triage"

✓ Loaded 16 knowledge entries
Top 3 matching Fluently 4D cycles:
1. Code Review Triage (coding)
Delegation: 24 | Description: 20 | Discernment: 18 | Diligence: 16
Overall 4D Score: 19.5
2. Bug Fix Prioritization (coding)
Delegation: 18 | Description: 15 | Discernment: 12 | Diligence: 10
# Suggestions follow for each dimension...
AI Collaboration Journey

Integrating Fluently into Your Daily AI Work

Fluently works alongside your existing AI tools. It is a thinking framework and guidance system, not a replacement.

1
Before
Frame the task
  • Run fluent list to see if a cycle exists for your task type
  • Decide your Delegation posture: full automation, augmented, or advisory?
  • Prepare a rich Description: context, constraints, examples, desired output format
  • Run fluent score "task" to validate your framing
2
During
Execute and observe
  • Use your prepared Description as the AI brief, and resist the urge to simplify it
  • Keep the Delegation boundary in mind and intervene when the AI exceeds its scope
  • Note any Discernment signals: hedging language, factual claims, overconfidence
  • Iterate: the Description-Discernment loop is normal and valuable
3
After
Close the loop
  • Apply your Diligence checklist: who reviews, approves, and signs off?
  • Log what AI did and what decisions humans made, to create an audit trail
  • Reflect: was the Delegation boundary right? Refine for next time
  • If you discovered a new cycle pattern, contribute it back!

Works Alongside Your Existing AI Tools

💬
Chat AI
Claude, ChatGPT, Gemini
Use 4D framing for prompts
💻
Code AI
Copilot, Cursor, Claude Code
Score your code review cycles
🔧
MCP Server
In-context knowledge access
AI reads cycles automatically
🤖
AI Agents
Autonomous workflows
Define delegation boundaries
Contribution Guide

Add Your Fluently 4D Cycle

Every contribution makes the knowledge base smarter for the entire community.

1

Find Your Cycle Using the Match Tool

Use the Knowledge Match tool on this site. Enter your Delegation and Description. If no strong match exists, you've found a new cycle worth contributing.

Try the Match Tool
2

Fork the Repository & Create Your YAML

git clone https://github.com/your-username/fluently.git
cd fluently
cp knowledge/coding-bug-fix-prioritization.yaml \
  knowledge/[domain]-[your-topic].yaml

Fill in all 4 dimensions: delegation, description, discernment, diligence, each with description, example, and antipattern fields.

3

Validate Locally

npm run build
node scripts/validate-knowledge.js
# Should show ✅ for your file

Your score_hints must sum to 1.0, all fields must be present, and the domain must be one of the supported values.

4

Open a Pull Request

git checkout -b knowledge/add-[your-topic]
git add knowledge/[your-file].yaml
git commit -m "docs: add Fluently 4D cycle for [topic]"
git push origin knowledge/add-[your-topic]

Open a PR on GitHub. CI will automatically validate your YAML, run tests, and post a 4D quality score as a comment.

5

CI Checks Run Automatically

Schema validation: All fields present, score_hints sum to 1.0
Test suite (30 tests): All must pass or PR is blocked
4D Score report: Bot comments your cycle's quality scores
Community review: Maintainers give feedback on quality

Once merged, your cycle is live in the knowledge base and available to all fluent CLI users worldwide.

Find Your 4D Match Full Contribution Docs