agentleFS
Sign inStart free
BilLogicBilLogic/uno-blueprint/AGENTS.md

uno-blueprint

AGENTS.md4 starsChanged 14 days agoMITView on GitHub

Router for the ub service-blueprinting skills; the skills are plain markdown and run in any agent. - Vocabulary — before writing a domain word (lane, path, slice, finding) into data, a skill or a doc: CONTEXT.md, by its ## heading. - Routing — INDEX.md maps task to document. Generated from the docs' own frontmatter, so it states the tree as it is. - Panel labels — the word a field shows and the column behind it: references/interface-schema-map.md. Generated, with a…

  • Reads credentials

The file

Create a free account to mount
Files
AGENTS.md

ub — Uno Blueprint service blueprinting skills (agent router)

Router for the ub service-blueprinting skills; the skills are plain markdown and run in any agent.

Before the task

  • Vocabulary — before writing a domain word (lane, path, slice, finding) into data, a skill or a doc: CONTEXT.md, by its ## heading.
  • Routing — INDEX.md maps task to document. Generated from the docs' own frontmatter, so it states the tree as it is.
  • Panel labels — the word a field shows and the column behind it: references/interface-schema-map.md. Generated, with a reason on each divergence.
  • Running it — the local stack, and the guard set to run before pushing: SETUP.md § Before you push.

Skill routing

Invocation / intentRead and follow
ub:map — create, import, translate, present or evolve a service blueprint; a directory holding blueprint-workspace.json or a blueprint/ folderskills/map/SKILL.md
ub:slice — cut a stakeholder view (actor journey, moment, lane, cell brief) out of an existing blueprintskills/slice/SKILL.md
ub:audit — run the consistency-check roster, then record and triage findingsskills/audit/SKILL.md
ub:whatif — trace a hypothetical change, promoting only on acceptanceskills/whatif/SKILL.md

Rules that hold for every skill

  • A skill's SKILL.md is the contract: its ⚠ hard rules bind, and entry-state detection runs FIRST, always.
  • references/, agents/ and scripts/ paths inside a skill resolve against this repo root, or against a workspace clone's own copies. They are a published interface: a deployment imports them by fixed path from a pinned tag, so moving one is a version bump plus a matching consumer change — docs/adr/0004-reference-paths-are-a-published-interface.md.
  • Sub-agent dispatch (agents/) runs in a fresh context, which is what keeps a review or an audit blind. Where a runtime offers no sub-agent primitive, run the prompt in agents/ yourself in a fresh context and return only its structured output.
  • Validation is deterministic: scripts/validate_ir.py exit 0 gates every import, and that exit code is the whole of the verdict.
  • Secrets live in a gitignored .env and nowhere else. hooks/secret_guard.py enforces that automatically in Claude Code alone, so enforce it yourself in any other runtime.

The repo around the skills

  • Editing anything under skills/ or references/ means running npm run sync:skills afterwards: it refreshes src/lib/agent/skill/, the vendored copy the app bundles, and .agents/skills/, the SKILL.md mirror other coding agents discover; npm test fails on drift in either.
  • Workspaces scaffolded from here carry their own skills/, references/, agents/ and scripts/, and the routing above resolves against the workspace root. Workspace version = template version; the upgrade recipe is references/customization.md.
  • Template app — this repo also ships the blueprint app under src/ and its eval harness (scripts/agent-harness/, evals/); what lives where is docs/overview.md.

Agent skills

Config the mattpocock-skills engineering skills read. Open the file a skill names when it asks for it, rather than preloading.

  • Issues are GitHub Issues; the repository and the workflow are docs/agents/issue-tracker.md.
  • Labels are the canonical triage roles and the tracker strings they map to: docs/agents/triage-labels.md.
  • Domain docs are single-context — the layout and the consumer rules are docs/agents/domain.md.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free