agentleFS
AnastasiyaWAnastasiyaW/codex-claude-code-config/skills/architecture/layer-new/SKILL.md

layer-new

Skill213 starsChanged 2 days agoMITView on GitHub

Scaffold a new layer in a project's docs/layers/ tree following the feature-layer architecture (principle 28). A layer is a bounded concern (security, data, ui, infrastructure, domain) with its own invariants, decisions, gotchas, patterns, and feature narratives. Use when: "create a new layer", "add security layer", "scaffold layer", "start tracking <concern> separately", "/layer-new", "add bounded concern". Operates on the kb-skeleton structure; idempotent -- will not overwrite existing layers. Do NOT use to scaffold an individual feature narrative inside an existing layer; use /feature-new for that (a layer is the container, not the per-feature doc).

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 4 are distinctive, and 10 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 10

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check the current working directory:

    line 54
  2. Very specific:

    Check if docs/layers/<layer-name>/ already exists.

    line 62
  3. Very specific:

    Must be lowercase kebab-case ([a-z][a-z0-9-]*).

    line 68
  4. Very specific:

    Must not start with _ (reserved for templates).

    line 69
  5. Very specific:

    Must not be a generic file-system name (src, tests, docs, build).

    line 70
  6. Distinctive:

    If invalid, refuse with a clear message and a suggested fix.

    line 72
  7. Very specific:

    Add a row to the ## Layer index table: | <layer-name> | <purpose> | active |

    line 121
  8. Distinctive:

    If a cross-layer Mermaid graph exists, add a node for the new layer with no edges (user will add edges as dependencies form).

    line 123
  9. Very specific:

    If the project has feature_list.json at repo root, leave it alone -- features get added by /feature-new.

    line 128
  10. Distinctive:

    If the project has AGENTS.md, suggest (but do not auto-edit) adding the new layer to the "Source-of-truth docs" table if multiple layers …

    line 132
  11. Very specific:

    Renaming a layer is not idempotent: If the user runs /layer-new wrong-name then realizes they wanted right-name, manually rename the …

    line 159
  12. Very specific:

    Layer name collision with existing directories: If docs/<layer-name>/ exists at the docs/ root (not under docs/layers/), refuse and ask the …

    line 163
  13. Very specific:

    Templates may have moved: If the template path ${CLAUDE_PLUGIN_ROOT}/templates/kb-skeleton/docs/layers/_LAYER-TEMPLATE/ does not exist (e.g.

    line 167
  14. Very specific:

    Encoding boundary: When writing files with Cyrillic content, follow the global rule from ~/.claude/rules/api-utf8-posting.md -- always …

    line 173

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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

/layer-new -- scaffold a project layer

Creates docs/layers/<layer-name>/ with the full template structure defined by principle 28.

When to use

  • Starting to track a new bounded concern in a long-running project
  • Refactoring sprawling cross-cutting code into a documented layer
  • Onboarding a new team member who needs the layer map

When NOT to use

  • One-off scripts or pet projects with <5 features (overhead not justified)
  • Layer name describes a directory (src is not a layer)
  • The "layer" is actually one feature in disguise -- use /feature-new inside an existing layer instead

Arguments

/layer-new <layer-name> [--purpose "..."] [--principles P-NN,P-MM]
  • <layer-name> -- kebab-case, single word preferred. Examples: security, data, image-processing, observability.
  • --purpose -- one-sentence purpose. If omitted, prompt the user.
  • --principles -- comma-separated Tier 1 principle IDs that govern this layer (e.g. P-02,P-21). If omitted, leave placeholder in README.

Direction (what to do, in order)

Step 1 -- Verify environment

Check the current working directory:

  1. Is it a git repository? Run git rev-parse --show-toplevel. If not, ask the user whether to initialize one (offer git init + private GitHub repo per global rule).
  2. Does docs/ exist? If not, create it.
  3. Does docs/layers/README.md exist? If not, copy from ${CLAUDE_PLUGIN_ROOT}/templates/kb-skeleton/docs/layers/README.md.
  4. Check if docs/layers/<layer-name>/ already exists. If yes, stop with a message -- do not overwrite. Suggest /feature-new <layer> <slug> instead.

Step 2 -- Validate layer name

  • Must be lowercase kebab-case ([a-z][a-z0-9-]*).
  • Must not start with _ (reserved for templates).
  • Must not be a generic file-system name (src, tests, docs, build).
  • If invalid, refuse with a clear message and a suggested fix.

Step 3 -- Copy the template

Source: ${CLAUDE_PLUGIN_ROOT}/templates/kb-skeleton/docs/layers/_LAYER-TEMPLATE/

Destination: <repo>/docs/layers/<layer-name>/

Copy the entire directory tree. Preserve subdirectory structure (kb/ and features/). Result:

docs/layers/<layer-name>/
├── README.md
├── history.md
├── kb/
│   ├── invariants.md
│   ├── decisions.md
│   ├── gotchas.md
│   └── patterns.md
└── features/
    └── _FEATURE-TEMPLATE.md

Step 4 -- Fill placeholders

In every file under the new layer, replace:

  • <layer-name> -> the actual layer name
  • <Layer name> -> Title Case of the layer name (e.g. "Security", "Image Processing")

In README.md specifically:

  • **Purpose:** <one sentence...> -> the --purpose argument value, or prompt the user
  • ## Governing principles list -> populate from --principles arg, or leave the placeholder bullets in place for the user to fill

In history.md:

  • Insert a "Layer created" entry at the top with today's date (YYYY-MM-DD) and the originating reason. Prompt the user for the reason if not provided.

Step 5 -- Register the layer

Update docs/layers/README.md:

  • Add a row to the ## Layer index table: | <layer-name> | <purpose> | active |
  • If a cross-layer Mermaid graph exists, add a node for the new layer with no edges (user will add edges as dependencies form).

Step 6 -- Wire to project state

If the project has feature_list.json at repo root, leave it alone -- features get added by /feature-new. Do not edit feature_list.json from this skill.

If the project has AGENTS.md, suggest (but do not auto-edit) adding the new layer to the "Source-of-truth docs" table if multiple layers exist.

Step 7 -- Confirm and suggest next step

Print a summary:

Layer created: docs/layers/<layer-name>/
Files: 1 README, 1 history, 4 kb/, 1 feature template

Suggested next steps:
1. Fill governing principles in docs/layers/<layer-name>/README.md
2. Write the first feature: /feature-new <layer-name> <slug>
3. Add the first invariant when it earns its place

Blueprints (files this skill writes from)

  • ${CLAUDE_PLUGIN_ROOT}/templates/kb-skeleton/docs/layers/_LAYER-TEMPLATE/ -- the source tree to copy
  • ${CLAUDE_PLUGIN_ROOT}/templates/kb-skeleton/docs/layers/README.md -- the layers index template (used only if missing)

Gotchas

  • Renaming a layer is not idempotent. If the user runs /layer-new wrong-name then realizes they wanted right-name, manually rename the directory and update references. This skill does NOT detect or fix duplicates.
  • Layer name collision with existing directories. If docs/<layer-name>/ exists at the docs/ root (not under docs/layers/), refuse and ask the user which they want -- there is no automatic merge.
  • Templates may have moved. If the template path ${CLAUDE_PLUGIN_ROOT}/templates/kb-skeleton/docs/layers/_LAYER-TEMPLATE/ does not exist (e.g. user is on a different machine), fall back to reading from https://github.com/AnastasiyaW/codex-claude-code-config/tree/main/templates/kb-skeleton/docs/layers/_LAYER-TEMPLATE via gh api.
  • Encoding boundary. When writing files with Cyrillic content, follow the global rule from ~/.claude/rules/api-utf8-posting.md -- always specify encoding="utf-8" explicitly.

Troubleshooting

SymptomCauseFix
"Layer already exists"Directory docs/layers/<name>/ presentUse /feature-new to add to it, or pick a different name
Template files missingPath moved or different machinePull templates from public repo via gh api repos/AnastasiyaW/codex-claude-code-config/contents/templates/kb-skeleton/docs/layers/_LAYER-TEMPLATE
Layers README not updateddocs/layers/README.md had no ## Layer index tableOpen file manually, add table per the kb-skeleton template
Validator warns about layervalidate_kb_links.py flagged broken linkLayer is fine; broken link is in a feature doc inside it. Run python scripts/build_kb_graph.py from the project root for the full health report.

Implementation note

The bulk of the work is file copy + placeholder replacement. No dynamic logic is needed; the template files do all the structural heavy lifting. Keep this skill deterministic and idempotent -- it must be safe to invoke twice on the same layer (second call should be a no-op with a clear message).

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.

Reviews can't be read right now.

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.