remember:status

/remember:status - Remember Status

Displays brain statistics: file counts, recent activity, and brain health.

Usage

/remember:status

⚠️ MANDATORY: Use Built-in Tools Only (NO Bash!)

NEVER use Bash commands for brain operations. Use Claude Code's built-in tools which are auto-approved and require zero permission prompts:

Operation	✅ Use This	❌ NOT This
List files	LS tool	bash ls
Find files	Glob tool	bash find
Search content	Grep tool	bash grep
Read files	Read tool	bash cat
Write/create files	Write tool	bash echo > / bash tee
Count files	Glob tool + count results	bash wc
Check env vars	Already available as $REMEMBER_BRAIN_PATH	bash echo $VAR

For complex operations (stats, counting, multi-step), use a subagent (Task tool) that uses the same built-in tools.

Why: additionalDirectories auto-approves LS/Glob/Grep/Read/Write. Bash always prompts.

Steps

1. Resolve Brain Path

Read $REMEMBER_BRAIN_PATH env var, fallback ~/remember. Use this as {brain_path}. If brain path doesn't exist → tell user to run /remember:init.

2. Show Brain Statistics

See "What to show" below for the full output template.

3. Show Recent Activity

List recently modified files (last 7 days) across People/, Projects/, Journal/, Notes/.

4. Show Persona Summary

Read {brain_path}/Persona.md and show a brief summary of captured patterns.

5. Backup health check

Check if {brain_path}/.git/ exists (use LS tool on {brain_path}/.git):

• Exists → show Backup: ✅ git-tracked. If you want to be helpful, also run git -C {brain_path} status --porcelain | wc -l and surface uncommitted-file count if > 0 (e.g. Backup: ✅ git-tracked · 12 uncommitted files — consider committing).

• Missing → show a warning block at the end of the output:

  ⚠️  Backup: NOT git-tracked
  
  Your brain lives at {brain_path} but has no git history. A single bad
  capture or accidental delete could lose months of work.
  
  Recommended:
    cd {brain_path}
    git init
    git add .
    git commit -m "initial brain snapshot"
  
  (Optional) push to a private GitHub repo for off-machine backup.
  

This is non-blocking — never refuse to show stats just because backup is missing.

What to show

Brain path: {brain}
Counts:
  - {N} Notes
  - {N} People
  - {N} Projects (active: {N})
  - {N} Areas
  - {N} Journal entries (latest: {date})
  - {N} Tasks (Focus: {n}, Next Up: {n})

Schema breakdown (by type):
  - world-fact: {N}
  - belief:    {N}
  - observation: {N}
  - experience: {N}
  - untyped (legacy): {N}

Freshness (L2 files only):
  - stable:        {N}
  - strengthening: {N}
  - weakening:     {N}
  - stale:         {N}
  - contradicted:  {N}

Top 5 beliefs by confidence:
  1. [[Notes/x]]  conf={c}  sources={n}  freshness={f}
  2. ...

Recent evolution.log (last 5 entries):
  {timestamp}  {TYPE}  {message}
  ...

Backup: ✅ git-tracked    (or)    ⚠️  NOT git-tracked — see warning below

To produce this:

1. Walk the brain — read frontmatter for each L2 markdown file (use scripts/config.js parseFrontmatter).
2. Tally by type and freshness.
3. Filter type: belief files; sort by confidence desc; take top 5.
4. Read tail of ~/.local/state/remember/evolution.log (use tail -n 5).

Existing files without type: count under "untyped (legacy)" — the upgrade is lazy.

If ~/.local/state/remember/evolution.log does not exist yet (no evolve runs have occurred), show (none yet) for the "Recent evolution.log" section or skip it.

Implementation

Use LS and Glob tools to read directories and count files. For frontmatter parsing, use parseFrontmatter from scripts/config.js. For complex stats, use a subagent.

Notes

• Fast command (uses built-in LS/Glob/Read tools — no Bash needed)
• Can run anytime
• No JSON metadata files required — counts come from the filesystem