Skill: Discover Document Norms
Purpose
Help users define their project-specific artifact norms (paths, naming, lifecycle) for document governance. Projects may have their own documentation structure; this skill discovers and formalizes it so other skills (capture-work-items, brainstorm-design, assess-doc-readiness) can follow project norms.
Core Objective
Primary Goal: Produce docs/ARTIFACT_NORMS.md (and optionally .ai-cortex/artifact-norms.yaml) that declares the project's artifact paths, naming, and lifecycle for backlog, design, ADR, and calibration artifacts.
Success Criteria (ALL must be met):
- ✅ Project structure scanned: Existing
docs/structure and conventions inspected - ✅ User preferences confirmed: Paths for backlog, design, adr, doc-readiness confirmed via dialogue (or accepted from starting template)
- ✅ ARTIFACT_NORMS.md written: Human-readable norms file at
docs/ARTIFACT_NORMS.md - ✅ Optional YAML created: If user requests,
.ai-cortex/artifact-norms.yamlwritten with machine-readable schema - ✅ User confirmed: User explicitly approved the norms before final write
Acceptance Test: Can another skill (e.g. capture-work-items) read the output and correctly resolve paths for this project?
Scope Boundaries
This skill handles:
- Scanning project
docs/structure - Dialogue to confirm artifact paths and naming
- Generating
docs/ARTIFACT_NORMS.mdand optional.ai-cortex/artifact-norms.yamlper spec/artifact-norms-schema.md - Starting from AI Cortex default, project-documentation-template, or blank
This skill does NOT handle:
- Bootstrapping full project docs (use
bootstrap-docs) - Assessing doc readiness (use
assess-doc-readiness) - Validating existing docs against norms (use
validate-doc-artifacts)
Handoff point: When norms are written and confirmed, hand off to other document-producing skills or validate-doc-artifacts for compliance checks.
Use Cases
- New project: No artifact norms yet; user wants to establish them before using capture-work-items or brainstorm-design.
- Existing project: Project has custom
docs/structure; formalize it so skills align. - Migration: Adopting AI Cortex defaults or project-documentation-template; create norms file for consistency.
Behavior
Interaction Policy
- Defaults: Project path = workspace root; start from AI Cortex defaults if no norms
- Choice options: Starting point
[ai-cortex][template][blank]; path mapping per artifact type - Confirm: Before writing ARTIFACT_NORMS.md; all path mappings
Phase 1: Scan
- Inspect project root and
docs/(if exists) - Identify existing paths: backlog, design, adr, calibration, etc.
- Note any conventions (naming, front-matter) from sample files
Phase 2: Choose Starting Point
Ask user or infer from context:
- AI Cortex default: Use spec/artifact-contract.md as baseline
- project-documentation-template: Map template structure to artifact types
- Blank: Start with minimal table, user fills in paths
Phase 3: Confirm Paths
For each artifact type (backlog-item, design, adr, doc-readiness):
- Propose path_pattern and naming from starting point or existing structure
- Ask user to confirm or customize
- Document Path Detection rules for backlog-item if applicable
Phase 4: Write and Confirm
- Generate
docs/ARTIFACT_NORMS.mdper spec/artifact-norms-schema.md - Optionally generate
.ai-cortex/artifact-norms.yaml - Present summary; request user confirmation before writing
- Write files after confirmation
Input & Output
Input
- Project path (default: current workspace root)
- Optional: starting point (ai-cortex | template | blank)
Output
docs/ARTIFACT_NORMS.md: Human-readable artifact norms table.ai-cortex/artifact-norms.yaml(optional): Machine-readable schema
Restrictions
Hard Boundaries
- No overwrite without confirmation: Do not write or overwrite norms files without explicit user approval.
- Schema compliance: Output MUST follow spec/artifact-norms-schema.md.
Skill Boundaries (Avoid Overlap)
Do NOT do these (other skills handle them):
- Full docs bootstrap →
bootstrap-docs - Readiness assessment →
assess-doc-readiness - Compliance validation →
validate-doc-artifacts
Self-Check
Core Success Criteria (ALL must be met)
- Project structure scanned: Existing docs/ inspected
- User preferences confirmed: Paths confirmed via dialogue
- ARTIFACT_NORMS.md written: File exists at docs/ARTIFACT_NORMS.md
- User confirmed: User approved before write
Acceptance Test
Can another skill read the output and correctly resolve paths for this project?
If NO: Norms incomplete or inconsistent. Return to Phase 3. If YES: Handoff complete.
Examples
Example 1: New Project, AI Cortex Default
Context: Empty project, no docs/.
Steps: Start from AI Cortex default. Propose paths: docs/backlog/, docs/design-decisions/, docs/process-management/decisions/, docs/calibration/. User confirms. Write docs/ARTIFACT_NORMS.md and create docs/ subdirs as needed.
Example 2: Existing Custom Structure
Context: Project has docs/work-items/, docs/decisions/, no formal norms.
Steps: Scan structure. Propose mapping: backlog-item → docs/work-items/, adr → docs/decisions/. User confirms. Write norms file with custom paths.
Appendix: Output contract
This skill produces document-artifact outputs for project norms. Output MUST conform to:
| Element | Requirement |
|---|---|
| Primary output | docs/ARTIFACT_NORMS.md — human-readable artifact norms table (artifact_type, path_pattern, naming, lifecycle) |
| Optional output | .ai-cortex/artifact-norms.yaml — machine-readable schema per spec/artifact-norms-schema.md |
| Path mapping | Each artifact_type mapped to a path_pattern; user must confirm before write |