| name | recommend |
| description | Get architecture advice for your use case. Research-backed recommendations for vault configuration based on your domain. |
| tags | ["architecture","guidance","methodology"] |
Skill: recommend
Delegates to: @ars-contexta:vault-architect
Trigger
Invoke this skill when the user wants configuration advice, is unsure whether their vault is set up optimally for their use case, or wants to know what a different preset would look like for their domain.
Behavior
Current State Analysis
Read and analyze:
ops/config.yaml — current dimension values
ops/derivation.md — original derivation rationale
ops/observations/ — accumulated friction and usage patterns
ops/health/ — most recent health report warnings
Build a picture of: what the vault was designed for vs. how it is actually being used.
Usage Pattern Inference
From observations and health data, infer:
- Which note types are being created most vs. least
- Which workflow steps are being skipped (no reduce entries, sparse MOC links, etc.)
- Which dimensions appear misaligned (e.g. high atomicity setting but notes are consistently long)
Preset Comparison
Compare the current configuration against the three standard presets:
- Research — atomicity 0.8+, processing 0.7+, self/ disabled
- Personal Assistant — mixed granularity 0.4–0.6, self/ enabled, entity navigation active
- Experimental — user-defined, maximum transparency
Identify which preset the current usage most resembles, and how far the configuration diverges.
Recommendations
Produce 2–5 specific recommendations, each with:
- The dimension or feature being recommended for change
- Current value → recommended value
- The research justification (cite specific claims or principles from the methodology corpus)
- The practical benefit the user will observe if they apply the change
- Any trade-off or cost to be aware of
Rank recommendations by expected impact.
Next Step
Offer two paths:
- "I can apply these changes now using
/architect."
- "I can explain the research behind any of these in more depth using
/ask."
Do not apply any changes from within this skill — recommend only. Changes require explicit invocation of the architect skill.