Brain Vault Align

Step 1: Load Source of Truth

Read the following files:

1. brain.yaml -- master configuration
2. Machine/Rules/active-rules.md -- current behavioral rules
3. core/brain-rules.md -- universal base rules

Step 2: Scan Vault Conventions

Analyze actual vault usage patterns to detect conventions:

1. Frontmatter patterns: Sample 10-20 recent files. What fields are consistently used? Any new fields not in the standard schema?
2. Tag patterns: What tags are actively used? Any new tag namespaces (e.g., #project/name)?
3. File naming patterns: Are files using kebab-case, camelCase, spaces? Is it consistent?
4. Wikilink patterns: How are wikilinks used? Short names vs full paths?
5. Directory usage: Are files landing in the expected directories per brain.yaml?

Step 3: Detect Drift

Compare active-rules.md against actual vault conventions:

Rules without matching reality

Rules in active-rules.md that do not match actual vault behavior:

• A rule says "use kebab-case" but recent files use spaces
• A rule references a directory that does not exist
• A rule describes a process that session logs show is not followed

Reality without matching rules

Conventions observed in the vault that are not captured in active-rules.md:

• Consistent patterns in recent files not documented as rules
• New directories or structures that emerged organically
• Behavioral corrections in Machine/Memory/corrections.md not yet promoted to rules

Configuration drift

Differences between brain.yaml settings and actual practice:

• preferences.wikilinks: true but files do not use wikilinks
• memory.session_logs: true but no session logs exist
• Agents marked enabled: true but config files are missing or outdated

Step 4: Propose Rule Updates

Present a summary of drift findings and propose specific changes:

## Drift Report

### Rules to Update
1. Rule "{name}" -- actual practice differs: {description}
   Proposed: {updated rule text}

### Rules to Add
1. New rule: "{name}" -- observed pattern: {description}
   Proposed: {rule text}

### Rules to Remove
1. Rule "{name}" -- no longer applicable: {reason}

### No Changes Needed
{list rules that are aligned}

Ask for user approval before making any changes to active-rules.md.

Step 5: Update Active Rules (with approval)

If the user approves changes:

1. Update Machine/Rules/active-rules.md with approved changes.
2. Append entries to Machine/Rules/rule-changelog.md:

   ### {YYYY-MM-DD} -- Vault Alignment
   - Updated: {rule name} -- {reason}
   - Added: {rule name} -- {reason}
   - Removed: {rule name} -- {reason}
   

Step 6: Regenerate Agent Configs

For each agent marked enabled: true in brain.yaml:

1. Read the corresponding template from generators/templates/:

   • Claude Code: CLAUDE.md.tmpl
   • Cursor: cursorrules.tmpl
   • Gemini CLI: gemini.md.tmpl
   • Codex: codex.md.tmpl

2. Regenerate the config file by combining:

   • Base rules from core/brain-rules.md
   • Active rules from Machine/Rules/active-rules.md
   • Owner profile from brain.yaml
   • Vault structure from brain.yaml
   • Agent-specific adaptations from the template

3. Write the regenerated config to the location specified in brain.yaml (agents.{name}.config_file).

4. If skills are enabled for an agent (agents.{name}.skills: true), ensure skill references are included in the config.

Step 7: Report Results

## Vault Alignment Complete

### Rules Updated
- {N} rules modified, {N} added, {N} removed
- Changelog updated: Machine/Rules/rule-changelog.md

### Configs Regenerated
- CLAUDE.md -- regenerated (Claude Code)
- .cursorrules -- regenerated (Cursor)
- {or "skipped -- agent not enabled"}

### Remaining Drift
- {any drift the user chose not to address}
- {or "Vault is fully aligned."}