fabric-design
Skill49 starsChanged 6 months agoMITView on GitHub
Read _projects/[name]/docs/discovery-brief.md for inferred signals, 4V's, and task flow candidates. Reference task-flows.md. For complex multi-pattern requirements, compose a hybrid using a base flow + overlays (document rationale inline in the handoff). Run decision-resolver.py --help for signal keys and fallback options. Write to _projects/[name]/docs/architecture-handoff.md. The runner pre-generates architecture-handoff.md (items, waves, ACs, decisions, diagram) before this skill is invoked. Your job: (1) verify accuracy against the discovery brief and decision-resolver output, (2) add project-specific rationale, trade-offs, and deployment strategy, (3) replace the…
In this repository8 other files it gives its agents.
- .github/copilot-instructions.mdCopilot instructions
- fabric-deploySkill
- fabric-discoverSkill
- fabric-documentSkill
- fabric-healSkill
- fabric-testSkill
- grill-with-docsSkill
- handoffSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
Read _projects/[name]/docs/discovery-brief.md for inferred signals, 4V's, and task flow candidates.
- Very specific: line 33
Run decision-resolver.py --help for signal keys and fallback options.
- Very specific: line 43
Before presenting the architecture for sign-off, verify EACH item below against the discovery brief AND the capability-mapper cache …
- Very specific: line 75
If the runner logged ⚠️ Decision resolver returned exit N or ⛔ Refusing to fast-forward, the pre-generated handoff is boilerplate — do not …
- Distinctive: line 80
Never deploy or create Fabric items
- Distinctive: line 81
Never skip "Alternatives Considered" or "Trade-offs"
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
Fabric Architecture Design (Architect Role)
Phase 1-design: Architecture Design
Step 1: Load Discovery Brief
Read _projects/[name]/docs/discovery-brief.md for inferred signals, 4V's, and task flow candidates.
Step 2: Select Task Flow
Reference task-flows.md. For complex multi-pattern requirements, compose a hybrid using a base flow + overlays (document rationale inline in the handoff).
Step 3: Resolve Architectural Decisions
python .github/skills/fabric-design/scripts/decision-resolver.py --discovery-brief _projects/[name]/docs/discovery-brief.md --format yaml- High confidence → accept the choice
- Ambiguous → read the referenced guide to resolve, present trade-offs to user
Run decision-resolver.py --help for signal keys and fallback options.
Step 4: Produce FINAL Architecture Handoff
Write to _projects/[name]/docs/architecture-handoff.md.
The runner pre-generates architecture-handoff.md (items, waves, ACs, decisions, diagram) before this skill is invoked. Your job: (1) verify accuracy against the discovery brief and decision-resolver output, (2) add project-specific rationale, trade-offs, and deployment strategy, (3) replace the <!-- AGENT: FILL --> marker in the ## Summary line (and anywhere else it appears) with a ≤20-word summary of the problem statement.
Step 4a: Pre-Handoff Verification Checklist (MUST run before showing user)
Before presenting the architecture for sign-off, verify EACH item below against the discovery brief AND the capability-mapper cache (_projects/<project>/docs/.capability-mapper-cache.json). If any check fails, revise the YAML items: and waves: blocks and regenerate the diagram via diagram-gen.py before user review.
Hard block — Capability coverage
Open .capability-mapper-cache.json. Every entry in required_items[] MUST appear in the handoff's items: block. The pre-generated handoff already unions them in (via --capability-cache); your job is to confirm they survived any edits.
- Every
required_items[*]from the cache has a matching item in the handoffitems:block - Every
required_capabilities[*]is satisfied by at least one item from itssatisfied_by_itemslist - If a required item was deliberately removed, the handoff's "Alternatives Considered" explains why
Velocity consistency
- If brief says "real-time" / "seconds latency" / "streaming" → handoff includes Eventstream + Eventhouse (or equivalent hot path)
- If brief says "batch" / "scheduled" / "nightly" → handoff includes Pipeline/CopyJob/Dataflow
- If brief says both → handoff includes both paths (hybrid)
- Trade-off rationale text does NOT contradict the brief's velocity (e.g., do not say "batch velocity" when brief is real-time)
Capability-driven signals (cross-check against .capability-mapper-cache.json)
python-ml-runtimerequired → handoff includes a Notebookhistorical-medallion-storerequired → handoff includes a Lakehouse with medallion structureconversational-queryrequired → handoff includes a DataAgentalerting-triggerrequired → handoff includes a Reflex (or OperationsAgent)mobile-field-intakerequired → handoff includes a CopyJob or DataPipeline path
Variety coverage
- Every source named in the brief (SQL DBs, sensors/SCADA, mobile forms, APIs, files) has an ingestion item routed to it
- Mobile/field-form intake → CopyJob or DataPipeline (not Eventstream alone)
- Historical SQL backfill → CopyJob or DataPipeline
Versatility alignment
- If brief says "low-code" preferred → ingestion/transforms use CopyJob/DataPipeline/Dataflow before Notebook/SparkJobDefinition
- If brief says "code-first for ML" → Notebook is present for ML workloads
If the runner logged ⚠️ Decision resolver returned exit N or ⛔ Refusing to fast-forward, the pre-generated handoff is boilerplate — do not trust its trade-offs or item list. Rebuild items/waves from the discovery brief + Step 3 resolver output manually.
Constraints
- Preserve YAML code blocks —
diagram-gen.pyparsesitems:andwaves:blocks to generate the architecture diagram. Do NOT convert them to markdown tables. - Never deploy or create Fabric items
- Never skip "Alternatives Considered" or "Trade-offs"
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.

