recall-promote

SkillDocs & knowledge

Review daily logs and promote entries to registers or working memory.

Use recall-promote in Claude, ChatGPT or Ahel Desktop

Free. Sign in, add recall-promote and connect your AI. About a minute.

Also: Claude Code · Cursor · Codex

Then ask your AI: use the recall-promote skill

Details

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

Add Ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

recall-promoteStart free

What this skill tells your AI

The instructions your AI receives, as published by davegoldblatt/total-recall in skills/recall-promote/SKILL.md and read by Ahel’s review.

Review recent daily logs and surface candidates for promotion to registers or CLAUDE.local.md.

What To Do

1. Gather Recent Logs

Read all daily log files from the last 7 days (memory/daily/YYYY-MM-DD.md).

If no daily logs exist, tell the user:

No daily logs found. Use /recall-write or /recall-log to start capturing.

2. Identify Promotion Candidates

Scan each entry and evaluate:

Promote to CLAUDE.local.md if:

  • Active commitment or deadline that affects every session
  • Behavioral preference that changes default output
  • Current project state that's essential context

Promote to a register if:

  • Decision with rationale → decisions.md
  • Person context → people.md
  • Technical choice → tech-stack.md
  • Preference with staying power → preferences.md
  • Project milestone or state change → projects.md

Skip if:

  • Already captured in a register or CLAUDE.local.md
  • One-off observation with no future impact
  • Debugging detail or transient state

3. Present Candidates

For each candidate, show:

Candidates for promotion:

1. [2026-02-03] "Project deadline moved to March 15"
   → Suggest: CLAUDE.local.md (active commitment)

2. [2026-02-02] "Prefers tabs over spaces, 2-space indent"
   → Suggest: registers/preferences.md

3. [2026-02-01] "Decided to use Postgres over SQLite for scale"
   → Suggest: registers/decisions.md

4. Ask User

For each candidate, ask the user what to do:

  • Promote — write to the suggested destination
  • Register — write to a different register (ask which one)
  • Skip — don't promote, leave in daily log
  • Promote all — accept all suggestions at once

5. Execute

For each promoted item:

  1. Write to the destination with proper metadata (claim, confidence, evidence, last_verified)
  2. Do NOT remove from daily log (daily logs are append-only history)
  3. Check for contradictions with existing entries before writing
  4. Assign a durable ID: For single-line list items promoted to CLAUDE.local.md or registers, append a ^tr ID (10 random lowercase hex chars). Create or update memory/.recall/metadata.json with an entry for the new ID (created_at, last_reviewed_at = now, pinned: false, snoozed_until: null, status: "active", tier from destination file). Create memory/.recall/ directory if needed. Write metadata.json with sorted keys.

6. Summary

Promotion complete:
  → 2 entries promoted to registers
  → 1 entry promoted to CLAUDE.local.md
  → 1 entry skipped

CLAUDE.local.md word count: [N]/1500

Warn if CLAUDE.local.md is approaching the 1500 word limit.

Signals

GitHub stars
204
Forks
14
Last commit
Feb 2026
Advanced
Item type
skill
Key
recall-promote
Source
github.com/davegoldblatt/total-recall