agentleFS
ClickHouseClickHouse/loom/.claude/skills/loom-change/SKILL.md

loom-change

Skill87 starsChanged 3 days agoApache-2.0View on GitHub

Lists what else a Loom change must touch in the same PR (generated files, snapshots, erasure classification, docs, manifests) for a new or changed op, field, table, column, migration, setting, flag, worker, provider, CLI command, plugin hook or dashboard page. Use before editing for any of those.

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
  1. Distinctive:

    Read the rule for each area the change touches (AGENTS.md, "Rules by area"), and search for the helper you are about to write.

    line 31

≈ “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.

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

Making a Loom change

Find the row for the change, read its guide before the first edit, and finish its "also" column in the same pull request. The rules under .claude/rules/ add the area-specific traps.

ChangeGuideAlso in the same PR
Wire op (new, renamed, removed field)adding-an-operationpython -m scripts.codegen.sdk_py_gen, .sdk_py_client_gen, .sdk_ts_gen, .docs_reference, .op_reach_census; unit test + integration test for a read or write path; MCP listing budget if it is listed by default; its dashboard control, or a NOT_IN_DASHBOARD reason in server/operations/dashboard.py
Table or columnadding-a-tablemigration + make schema-dump; RMT row builders; loom.memory_versions + its view for a loom.memory column; erasure step + census row, or a NO_MEMORY_CONTENT / STEP_ONLY_TABLES entry with its reason; a new table's row in docs/architecture/storage-layout.md, its Today cell when the engine, sort key, partition key or TTL changes, its Erasure cell when its census classification changes, its row removed when the table is dropped (with the dump pending, the row test fails until it lands)
Migrationmigrationsnext free number on current origin/main; additive only; make schema-dump on a private ClickHouse (with none at hand, leave schemas/schema.sql unchanged and report the dump as pending; never hand-edit it); written values N-1 tolerates (rolling upgrades), make test-mixed-version
Settingextending-loomfield in its area's section of server/settings/catalog.py (models in Models) with a comment on what it controls and its values; python -m scripts.codegen.settings_reference, .settings_schema and .settings_example; read as settings.<field>; a rename or removal follows the settings rule
Retire a flag—delete the setting, its branch and any request field carrying it; the PR body's changelog entry names it
Workeradding-a-workerregistry row; LOOM_UPDATE_WORKER_SNAPSHOT=1 for an intended snapshot change
Providerextending-loommake bench-quick evidence before changing a default (ask first — see loom-bench)
Code-index languageadding-language-supportstay outside the extractor stamp; backfill plan in the PR
CLI command or flagcode-generationLOOM_UPDATE_CLI_SNAPSHOT=1 snapshot; python -m scripts.codegen.docs_reference; a command the chart's hook Jobs run (deploy/helm/loom/templates/hooks.yaml) changes there too
Plugin file (hook, lib/, bin/, skill, assets/, .mcp.json, Codex plugin-defaults.json, a manifest's listing fields)integrationshook scripts, lib/, bin/, skills and assets/: edit integrations/claude-code-plugin/, then python -m scripts.codegen.agent_plugin_sync; a manifest, hooks.json, .mcp.json or plugin-defaults.json: edit each plugin's own copy (plugins rule); for a behaviour change, bump both plugin manifests to a version no other open PR uses
Dashboarddashboard-design, local-developmentnpm --prefix frontend run test, npm --prefix frontend run build, bundle budget, make lint (Biome's accessibility rules)
Helm chartkuberneteshelm lint --strict and kubeconform on both overlays; a renamed setting as in the Setting row
Docsdocumentation-styleregenerate generated pages; never hand-edit them

Before the first edit

  • A new table, op group, auth-model change, external dependency or identity-tuple change is a material change: the PR body carries the design (what, why, the alternatives and their cost, the migration story), and the matching docs/architecture/ page describes the system as it now is (material changes).
  • Read the rule for each area the change touches (AGENTS.md, "Rules by area"), and search for the helper you are about to write.
  • Moving a module: shipped files resolve through server.core.paths, never Path(__file__).parents[N].

Then run loom-pre-push.

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.