Score and sync AI-readiness context across dev tools

A skill for getting a project to 100% AI-readiness using .faf, an IANA-registered context format (application/vnd.faf+yaml) - a typed, portable, vendor-neutral file readable by any AI. faf-cli defines 21 context slots; the flow is three steps, only the last needs the user: the project's app type determines which slots are required (a CLI needs different slots than a full-stack web app, and irrelevant slots are marked slotignored and dropped from the denominator); faf auto detects the stack and language and seeds who/what/where from a written goal sentence; whatever's left empty (usually why and when, the "purely yours" Ws) is filled by answering the AI's questions. The six Ws are terse 3-4 word labels, not prose: who (audience), what (the thing being built), why (its reason to exist), where (where it runs/ships), when (timeline/stage), and how (how it's built) - who/what/where get seeded from the goal sentence and how gets sourced from stack detection, leaving why and when as the two typically hand-written answers. The fastest path is five commands: faf auto detects the stack and seeds context from the README, faf score shows the current percentage and which slots are empty, faf go guides confirming seeded Ws and filling the 1-2 remaining, faf score again confirms 100%, and faf sync optionally pushes context into CLAUDE.md or AGENTS.md. The honesty rule underpins all of it: the AI only seeds facts the goal sentence or README literally state, never invents or templates, and leaves anything it can't source empty rather than guessing - empty beats wrong.

Use this skill when you need your project at 100% AI-readiness fast, with minimum typing and maximum auto-detected context - it's the builder's quickstart path, feeding into faf-expert for depth. For a fully done-for-you one-click path use faf-wizard instead; to master the format's scoring internals, MCP config, and bi-sync use faf-expert instead; for driving a repo end-to-end with an agent use faf-go or faf-loop.

Input is one sharp, specific goal sentence (e.g. "A CLI that scores any repo's AI-readiness and syncs context to Claude, Cursor, and Gemini - for solo developers") plus answers to the 1-2 Ws the AI couldn't derive. Output is a completed .faf context file scored 0-100%, with irrelevant slots marked slotignored and excluded from scoring.

faf auto
faf score
faf go
faf score
faf sync

Produces a .faf file (application/vnd.faf+yaml) that can optionally sync into CLAUDE.md or AGENTS.md via faf sync for AI tools that read those files directly instead of the .faf format.

Time-conscious builders who want their project fully AI-context-ready without manually filling out a 21-slot spec - write one good goal sentence, confirm what auto-detection already seeded, answer the couple of questions only a human can answer, and reach 100% AI-readiness. Part of the FAF skill family (faf-context, faf-wizard, faf-expert), MIT licensed.