Recall Context
npx machina-cli add skill davegoldblatt/total-recall/recall-context --openclawShow what memory context is currently available in this session.
What To Do
1. Check What's Auto-Loaded
These files load automatically via Claude Code's native mechanisms:
Via .claude/rules/ (deterministic):
- .claude/rules/total-recall.md — memory protocol
Via CLAUDE.local.md (deterministic):
- CLAUDE.local.md — working memory
2. Check What Exists On Disk
Read and report on what memory files are present:
Schema:
- memory/SCHEMA.md — exists?
Daily Logs:
- Today's log (memory/daily/[today].md) — exists? Entry count?
- Yesterday's log — exists?
Registers:
- List all .md files in memory/registers/ with entry counts
Archive:
- List files in memory/archive/
3. Display
Memory Context — Current Session
─────────────────────────────────
Auto-loaded (deterministic):
.claude/rules/total-recall.md ✓ Protocol rules
CLAUDE.local.md ✓ Working memory ([N] words)
On disk:
memory/SCHEMA.md ✓ Protocol docs
memory/daily/[today].md ✓ [N] entries
memory/daily/[yesterday].md ✓ [N] entries
memory/registers/open-loops.md ✓ [N] active items
Available registers:
people.md [N] entries
projects.md [N] entries
decisions.md [N] entries
preferences.md [N] entries
tech-stack.md [N] entries
Archive: [N] files
Hooks:
SessionStart [configured/not configured]
PreCompact [configured/not configured]
4. Highlight Gaps
If expected files are missing:
Missing:
✗ CLAUDE.local.md — run /recall-init to create
✗ No daily log for today — will be created on first /recall-write
5. Quick Actions
Use /recall-search <query> to pull context from any tier.
Use /recall-status for memory health metrics.
Source
git clone https://github.com/davegoldblatt/total-recall/blob/main/skills/recall-context/SKILL.mdView on GitHub Overview
Recall Context reveals what memory files are loaded and available in the current session. It distinguishes auto-loaded sources (via Claude Code's deterministic rules and CLAUDE.local.md) from on-disk assets like memory schemas, daily logs, registers, and archives, and highlights gaps to help you audit your working memory.
How This Skill Works
The skill checks both auto-loaded sources (e.g., .claude/rules/total-recall.md and CLAUDE.local.md) and on-disk assets (memory/SCHEMA.md, memory/daily/[today].md, memory/daily/[yesterday].md, memory/registers/*.md, memory/archive/*). It then presents a structured view showing Auto-loaded, On disk, Available registers, and Archive status, and marks any missing items for quick remediation.
When to Use It
- Audit the current session's memory context before starting a task
- Verify which memory files are auto-loaded versus on disk
- Confirm availability of memory schema, daily logs, registers, and archive
- Identify missing files or gaps to address before recall
- Use quick actions to pull context from other tiers or check memory health
Quick Start
- Step 1: Trigger the memory context display for the current session
- Step 2: Review Auto-loaded vs On disk sections and note any missing items
- Step 3: Run /recall-search <query> or /recall-status to enrich or check memory health
Best Practices
- Start with Auto-loaded sources (.claude/rules and CLAUDE.local.md) to understand deterministic memory
- Validate on-disk assets (memory/SCHEMA.md, daily logs, registers, archive) are present
- Check today’s and yesterday’s daily logs for recency and entry counts
- List and audit registers (e.g., people.md, projects.md, decisions.md) to know available context
- Use /recall-search and /recall-status to enrich context and monitor memory health
Example Use Cases
- A developer reviews auto-loaded and on-disk memory before prompting Claude Code
- A data scientist confirms CLAUDE.local.md is loaded as working memory
- An engineer checks that memory/SCHEMA.md and daily logs exist for today and yesterday
- A product manager inventories registers like people.md, projects.md, and decisions.md
- An operator notes Hook configurations and archive contents before a recall task
