using-superlite
Skill249 starsChanged 14 days agoMITView on GitHub
Use when starting any conversation - explains superlite, how to size a request, and which skill fits which task
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- commands/agents.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .cursor/rules/feature-worktrees.mdcCursor rule
- .cursor/rules/no-cursor-commit-notes.mdcCursor rule
- diagnosisSkill
- draftingSkill
- feature-prototypeSkill
- inventingSkill
- mvp-prototypeSkill
- proofSkill
- shippingSkill
- swarmSkill
- testingSkill
- writing-specsSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
Who orchestrates?: Ask once, in that same batch: human orchestrator (approves the plan before it runs) or AI orchestrator (hands off).
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Superlite: parallel work, short reports.
Start menu. First message, no mode named, not trivial: superlite-cli suggest_flow, then /superlite:start.
Size the request first.
- Trivial: just do it (human orchestrator: after a yes in chat).
- Bounded (change to an existing flow): short design in chat, build.
- Architectural: batched questions (one AskUserQuestion, up to 4), one combined design+plan doc.
Who orchestrates? Ask once, in that same batch: human orchestrator (approves the plan before it runs) or AI orchestrator (hands off). Skip when the request settles it ("decide yourself", away, headless): AI, as a Ruling. superlite-cli orchestrator set human|ai. AI stops only for destructive, outward or security-sensitive actions.
Skill map
- build -> superlite:inventing
- plan -> superlite:drafting
- execute -> superlite:swarm
- bug -> superlite:diagnosis
- tests first -> superlite:testing
- claim done -> superlite:proof
- finish -> superlite:shipping
Builds with 2+ tasks always run through superlite:swarm; load it before writing the plan. No general-purpose agents for build or review. Build inline only when small (a few files, one chain); say so.
Final review always applies. A non-trivial build ends with a fresh superlite:final-reviewer on the review packet before claiming done.
Tools. bash <skill-dir>/../swarm/scripts/superlite-cli <tool>, never python. Incidents go to .superlite/incidents.jsonl; Graphify only if superlite-cli graphify_gate allows.
Platform adaptation. Off Claude Code, read the matching references/<harness>-tools.md (antigravity, codex, cursor, gemini, hermes, muse, pi); it maps tool names and says how to run agents/*.md.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.