Solfab
Skill philgoodvibe/aiai-mastermind-tools-and-skills/skills/model-router/codex/solfab
Hand any coding/build brief to Claude Fable, which orchestrates the build while GPT-5.6 Sol does the typing (the fabsol flow). Use whenever the user says "solfab", or asks to build, implement, create, or fix something in a codebase and wants the SolFab pipeline. You act as the remote control - you never build it yourself.From its SKILL.md
npx -y skills add philgoodvibe/aiai-mastermind-tools-and-skills --skill solfabAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
- runs commandsInstructs the agent to run 1 command, including `bash ~/.claude/model-router/scripts/solfab.sh "<the brief>" "<absolute project dir>"`.
SKILL.md
1.9 KB, 390 tokens by cl100k_base, as published. Nobody here has run it
SolFab - relay the brief to Claude Fable
You are the dispatcher, not the builder. Claude Fable is the orchestrator (plans, writes specs, verifies); it delegates 100% of the building back to GPT-5.6 Sol through a local proxy. Your whole job is one command and a faithful relay of the result.
The one command
bash ~/.claude/model-router/scripts/solfab.sh "<the brief>" "<absolute project dir>"
<the brief>: the user's request, verbatim, plus any context you have that Fable will need (paths, constraints, acceptance criteria). Fable runs headless and cannot ask questions - if the user's ask is ambiguous, ask THEM first, then send one complete brief.<absolute project dir>: where the work should happen. Omit to use the current directory.- Builds take minutes. Run the command in the foreground with a LONG timeout (at least 15 minutes). Do not kill it because it is quiet - Fable prints only at the end.
After it finishes
Relay Fable's final report to the user faithfully (what shipped, how it was verified, anything left open). Do not re-verify or edit the result yourself unless the user asks you to.
If it fails
claude CLI not found/ script missing -> tell the user to re-run the model-router installer (aiai-mastermind-tools-and-skills repo).- Proxy/auth errors in the report -> tell the user their provider login needs
a refresh:
cd ~/cli-proxy-api && ./cli-proxy-api -config config.yaml -codex-login - NEVER work around a failure by building it yourself unless the user explicitly asks you to take over.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.