Menu
Scaffold a complete knowledge system. Conversational onboarding derives configuration from your domain. Validates against 15 kernel primitives.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Add a new knowledge domain to an existing vault. Creates domain-specific MOCs, templates, and vocabulary mappings.
Research-backed vault evolution guidance. Toggle features (self-space, semantic search), rebalance MOCs, adjust dimensions with full rationale.
Query the Ars Contexta research methodology. Answers 'why does my system do X?' with research-backed explanations from 249 interconnected claims.
Analyze the vault's knowledge graph. Subcommands: orphans, density, bridges, synthesis, traverse, clusters, stats.
Run vault health diagnostics. 3 modes: quick (schema+orphans+links), full (all 8 categories), three-space (boundary violations).
Contextual guidance and command discovery. Shows available skills, agents, and intelligent suggestions based on vault state.
| name | setup |
| description | Scaffold a complete knowledge system. Conversational onboarding derives configuration from your domain. Validates against 15 kernel primitives. |
| tags | ["architecture","onboarding","derivation"] |
Delegates to: @ars-contexta:vault-architect
Invoke this skill when the user wants to create a new Ars Contexta vault, initialize a knowledge system, or run first-time setup.
Activate the vault-architect agent in setup mode. The agent runs the full derivation flow:
Screen 1 — Product Introduction Introduce Ars Contexta: a research-backed knowledge system that derives its structure from your domain rather than imposing a generic template. Explain that the system will ask a few questions, then generate a vault tailored to how you actually think and work.
Screen 2 — Three Presets Present the three starting points:
Screen 3 — What Happens Next Explain: "I'll ask you 2-4 questions about your work, then derive 8 configuration dimensions from your answers, propose a vault structure, and generate everything you need to start."
Conversation Phase (2-4 turns) Open with: "Tell me about what you want to track, remember, or think about."
Extract signals passively from the user's natural language. Listen for:
Derivation Phase Map extracted signals to the 8 configuration dimensions (0.0–1.0 scale):
Confidence scoring per dimension. Resolution threshold: cumulative confidence > 1.5 per dimension. For low-confidence dimensions, default to the preset baseline.
Proposal Phase Show the user what will be generated and explain every architectural choice in terms of their stated needs. Wait for confirmation before generating.
Generation Phase Generate the complete vault:
ops/derivation.md and ops/derivation-manifest.mdops/config.yaml with derived dimension valuesDual Context Files (always generate both)
copilot-instructions.md — GitHub Copilot CLI formatCLAUDE.md — Claude Code / OMC formatBoth are derived from the same configuration but use platform-specific syntax. This ensures vault portability across AI tooling.
Validation Phase Check all 15 kernel primitives: markdown-yaml, wiki-links, moc-hierarchy, tree-injection, description-field, topics-footer, schema-enforcement, self-space, session-rhythm, semantic-search, unique-addresses, discovery-first, operational-learning-loop, task-stack, methodology-folder.
Report PASS/FAIL per primitive. Surface any missing elements as a checklist for the user to resolve.