agentleFS
tatsoft-llctatsoft-llc/FrameworX-Projects/.claude/skills/uns-from-io-register/SKILL.md

uns-from-io-register

Skill1 starsChanged 27 days agoApache-2.0View on GitHub

Turn a project's Specifications/IO-Register.md into the solution's UNS foundation, UserTypes for repeated equipment, the Area/Equipment/Signal folder hierarchy, and one tag per register row, then verify the result by read-back and reconcile the register's "Target UNS Tag" column as built. The Pillar-1 counterpart to display-authoring-loop: the most consequential design layer, mechanized. Use when the user says "build the UNS", "create the tags from the IO register", "author the tag layer", "model the plant", "set up the UserTypes", "turn the point list into tags", or right after new-project-bootstrap hands off to authoring.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Ground first: Read Specifications/IO-Register.md + Specifications/Requirements.md and KnowledgeBase/Conventions/UNS-Naming.md before …

    line 29
  2. Very specific:

    TagProvider boundary: if an external source (OPC UA server, MQTT broker) publishes the catalog, the source is authoritative, surface it via …

    line 39
  3. Very specific:

    Update Specifications/IO-Register.md's Target UNS Tag column with the as-built paths.

    line 71

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

UNS from the IO Register: the foundation layer, traceable

Why this exists

The UNS is the first and most consequential design decision in a solution, every device point, alarm, historian entry, script, and display binds to it. Building it ad-hoc produces the two classic failures: tags that don't trace back to the register (orphan signals nobody commissions) and near-duplicate structures where a UserType should have been. This skill makes the register the single source: every row becomes exactly one tag, every tag traces to a row, and the register's Target UNS Tag column is updated as built, that column is the traceability contract commission-the-points later drives.

Hard rules

  • Ground first. Read Specifications/IO-Register.md + Specifications/Requirements.md and KnowledgeBase/Conventions/UNS-Naming.md before writing anything. The platform's UNS design rules come from the bootstrap Context Package: fetch, don't recall.
  • Naming: <Area>/<Equipment>/<Signal>, PascalCase members, / for hierarchy, . only for member access. Units live in the tag's engineering-unit property, never in the name.
  • UserType when instances repeat. Two or more pieces of equipment with the same member set (every Pump has Run, Speed, Fault…) = define the UserType once, instantiate per equipment. One-off signals stay plain tags. Don't force a UDT onto a single instance "for the future."
  • Never create tags for object-model state. Alarm counts, server time, device status live in the runtime object model (Alarm.Group.*, Server.*, Device.*), bind to them directly.
  • TagProvider boundary: if an external source (OPC UA server, MQTT broker) publishes the catalog, the source is authoritative, surface it via a TagProvider (load the platform's TagProvider skill via search_docs), never hand-author shadow tags of provider-published names.

Process

1. Derive the model from the register

Group rows by area and equipment. Propose the shape before writing: the UserTypes (name + members with data types), the folder tree, the plain tags, and the count ("42 rows → 3 UserTypes × 9 instances + 6 plain tags"). One short confirmation with the user on the shape: the UNS is the decision worth a question; everything after is mechanical.

2. Write in dependency order, in batches

get_table_schema('UnsUserTypes') / ('UnsTags') for the real field names (data type, engineering units, min/max, description, never guess), then write UserTypes first, then folders/instances/tags, related tables can share one multi-table write_objects. Keep batches modest (a dozen-odd objects on heavy tables); read every response's warnings and skipped counts. Carry the register's engineering columns (units, range, description) onto the tags now: they feed alarms, displays, and the historian later, and retrofitting them costs a full second pass.

3. Read back: counts and spot paths

  • get_objects / search_uns to confirm the created counts match the plan (UserTypes, instances, tags).
  • Spot-read full paths, including at least one member per UserType (Tag.Area1/Pump1.Run). Instance member access is the thing that breaks when a member name drifted.
  • On a large register, batch by area and verify each batch before the next. Track the batches in Claude Code's task list, so a long build stays legible and resumable.

4. Reconcile the register

Update Specifications/IO-Register.md's Target UNS Tag column with the as-built paths. Any register row without a tag, or tag without a row, is a finding to surface, not to silently absorb. Note the layer in your ChangeLogs/ChangeLog-<user>.md; a non-obvious modeling choice (why this UDT boundary, why that area split) goes to Decisions/.

5. What this skill does NOT do

Alarm limits, historian selection, and device addresses in the register are recorded here but built in their own layers (Pillar 2, first-device-channel for comms, alarm authoring per KnowledgeBase/Conventions/Alarm-Style.md). Finish and verify the foundation before fetching any Pillar-2 schema: pillar isolation is the platform's rule, not a preference.

Anti-patterns

  • Writing tags straight from the register without proposing the UDT/folder shape first. The shape IS the design decision.
  • Mirroring an external system's naming (snake_case columns, PLC symbol names) into member names. One convention, PascalCase; alias at the boundary (query aliases, provider mapping), never duplicate members.
  • A 200-tag single write. Batch by area; errors stay isolated and progress stays visible.
  • Leaving the Target UNS Tag column stale. An unreconciled register makes commissioning drive ghosts.
  • "The sibling instance works, so they all do." Each instance path you bind to later is its own verification (CLAUDE.md § Verify each path).

Composition

  • new-project-bootstrap: produces the register this skill consumes; hands off here.
  • first-device-channel: next layer: map the built tags to device points and prove comms.
  • ground-first / premise-check: the shape-confirmation step is a natural premise check moment.
  • run-tests: expected.json's structural assertions (UserTypes, tag counts) come from step 1's plan.

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.