rust-faf-cli

FAF Compiler — AI-Readiness scoring engine for .faf files. Rust-native CLI.

Score any project's AI context quality. See exactly why. Fix it automatically.

This is the Rust-native edition. See also faf-cli (TypeScript, npm).

Install

Homebrew (macOS/Linux):

brew install Wolfe-Jam/faf/rust-faf-cli

Direct download:

# macOS ARM (Apple Silicon)
curl -L https://github.com/Wolfe-Jam/rust-faf-cli/releases/latest/download/faf-aarch64-apple-darwin.tar.gz | tar xz
sudo mv faf /usr/local/bin/

# macOS Intel
curl -L https://github.com/Wolfe-Jam/rust-faf-cli/releases/latest/download/faf-x86_64-apple-darwin.tar.gz | tar xz
sudo mv faf /usr/local/bin/

# Linux x86_64
curl -L https://github.com/Wolfe-Jam/rust-faf-cli/releases/latest/download/faf-x86_64-unknown-linux-gnu.tar.gz | tar xz
sudo mv faf /usr/local/bin/

Windows (PowerShell):

Invoke-WebRequest -Uri "https://github.com/Wolfe-Jam/rust-faf-cli/releases/latest/download/faf-x86_64-pc-windows-msvc.zip" -OutFile faf.zip
Expand-Archive faf.zip -DestinationPath .
Move-Item faf.exe "$env:LOCALAPPDATA\Microsoft\WindowsApps\"

What It Does

$ faf status

Mission Status Report
==================================================

  Project:  my-app
  Score:    73%
  Tier:     GO! Ready

Glass Hood Analysis
--------------------------------------------------
  Completeness   [===============     ]  75.0% x 40% =  30.0
  Clarity        [==============      ]  70.0% x 35% =  24.5
  Structure      [================    ]  80.0% x 15% =  12.0
  Metadata       [=============       ]  65.0% x 10% =   6.5
--------------------------------------------------
  TOTAL                                          73.0%

GO! 73 - Ready  |  ONE thing to improve:
  -> [DO] Add key_files to project.faf (3 items minimum)
  -> Projected: 73% -> 82%

Every score is transparent. Glass Hood shows exactly how it's calculated — weights, categories, projections. Nothing hidden.

Commands

Command	What it does
faf init <name>	Create project.faf + context files
faf status	Score + tier + top suggestion
faf status --verbose	Glass Hood deep-dive
faf status --card	Generate score-card.svg
faf status --json	Machine-readable JSON
faf sync	Sync to GROK.md, CLAUDE.md, GEMINI.md
faf sync --claude	Sync CLAUDE.md only
faf gold	Gap analysis to reach 100%
faf compile	Compile .faf to .fafb binary (288x faster)
faf bench	Race benchmark: FAF vs FAFB
faf embed grok	Embeddable context header
faf completions zsh	Shell completions

Scoring

Four categories, weighted by impact on AI comprehension:

Category	Weight	What it measures
Completeness	40%	Are all context slots filled?
Clarity	35%	Is the context unambiguous?
Structure	15%	Is it well-organized?
Metadata	10%	Version, timestamps, DNA tracking

Language bonuses: Rust +15, Go +10, TypeScript +5.

Tier System

Score	Tier	Status
100%+	Rocket	Mission-Ready
99%	Gold	Almost there
95%	Silver	Top tier
85%	Bronze	Production ready
70%	GO!	Solid foundation
55%	Caution	Needs improvement
<55%	Stop	Major work needed

FAFb Binary Format

Compile .faf (YAML) to .fafb (binary) for O(1) section lookup. 288x faster parsing. Priority truncation fits any context window.

IANA registered: application/vnd.faf+yaml

Links

• faf.one — Main site
• faf-rust-sdk — Rust SDK (MIT, crates.io)
• faf-cli — TypeScript CLI (npm)
• claude-faf-mcp — Claude MCP server
• All packages — Full ecosystem directory

Changelog

v1.1.0 (2026-03-05)

• Dual identity: xai-faf (xAI edition) + faf (universal edition) from one codebase
• Glass Hood scoring with 4-category weighted analysis
• Multi-target sync: GROK.md, CLAUDE.md, GEMINI.md
• Score card SVG generation (--card)
• Machine-readable JSON output (--json)
• DRS auto-fix suggestions (ONE thing to improve)
• FAFb binary compilation (288x faster than YAML)
• Cross-platform binaries: macOS ARM/Intel, Linux, Windows

License

Proprietary. See LICENSE.

---

team@faf.one | faf.one