skills.rest
xtrm-dev
xtrm-devOFFICIAL
Explore all skills by this author →
GitHub

memory-audit-transaction

Audits large bd memory stores via chunked, file-backed JSONL decision ledger.

4|1|Updated Nov 1, 2025
Search Tags:#bd#chunking#pruning#jsonl#context-window#ledger#memory-audit
One-click install
npx skills add https://github.com/xtrm-dev/specialists --skill memory-audit-transaction
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: memory-audit-transaction
Source: https://github.com/xtrm-dev/specialists/tree/main/config/skills/memory-audit-transaction
Command: npx skills add https://github.com/xtrm-dev/specialists --skill memory-audit-transaction

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires bd, jq, and includes scripts (resource) components.

What problem does it solve?

This Skill prevents large bd memory audits from overwhelming the agent context window by moving classification, evidence tracking, and pruning into a resumable file-backed workflow.

Core Features & Use Cases

  • Chunked memory review: Process hundreds or thousands of memories in bounded batches instead of accumulating decisions in chat.
  • Evidence-backed decisions: Record each memory classification in a JSONL ledger with hashes, references, and explicit status values.
  • Safe pruning workflow: Verify content hashes before forgetting stale or redundant memories and keep backups for recovery.
  • Use case: Run it when consolidating .xtrm/memory.md, reviewing recent session reports, or auditing a memory-processor specialist that already hit context limits.

Quick Start

Ask the agent to perform a full memory audit transaction for the current project and synthesize the kept memories into .xtrm/memory.md with an evidence-backed decision ledger.

Frequently Asked Questions about memory-audit-transaction

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit large bd memory stores without exceeding the agent context window?▼

To prevent context blowups during a memory audit, process hundreds of project memories in bounded chunked batches instead of accumulating decisions in chat. This file-backed workflow records classifications in a JSONL ledger.

What is hash-guarded pruning and how does it protect memory consolidation?▼

Hash-guarded pruning protects memory consolidation by verifying content hashes before removing stale or redundant memories. It maintains backups for recovery, ensuring every pruning decision remains evidence-backed and resumable.

Do I need jq and bd installed to run a chunked memory review?▼

Yes, the chunked memory review requires both bd and jq dependencies to parse bulk-exported memory artifacts and manage the JSONL decision ledger. These tools handle the underlying data extraction and validation.

When should I use a file-backed JSONL ledger for memory-processor runs?▼

Use a file-backed JSONL ledger when consolidating `.xtrm/memory.md` or reviewing recent session reports involving hundreds of memories. It is required when a memory-processor specialist already hit context limits and needs resumable tracking.

What's the best way to classify hundreds of project memories in bounded batches?▼

The best way to classify hundreds of project memories is bulk-exporting memory artifacts and processing them in chunked batches with completeness validation. This records explicit status values, hashes, and references into a JSONL ledger.

Related Skills

Explore More Software Engineering Skills→
devopstalesdevopstales
1

beads-sync

Synchronize markdown artifacts with a Beads graph database.

Community
Advanced
peteknowsaipeteknowsai

dream

Consolidate recent learnings into topic-based memory files under .claude/memory/.

Community
Advanced
shihwesleyshihwesley

self-learn

Analyze recent session learnings and update MEMORY.md with verified insights.

Community
Intermediate
eron1703eron1703
1

remember

Persist user-provided information to MEMORY.md and GitLab CI/CD variables.

Community
Advanced