One-click install
npx skills add https://github.com/davidlee/spec-driver --skill reviewing-memory
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reviewing-memory
Source: https://github.com/davidlee/spec-driver/tree/main/.spec-driver/skills/reviewing-memory
Command: npx skills add https://github.com/davidlee/spec-driver --skill reviewing-memory

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prevents systemic drift by turning vague or stale agent memories into a verified, source-backed, correctly scoped set of guidance before releases, migrations, or complex refactors.

Core Features & Use Cases

  • Staleness-first review workflow: Surfaces the highest-impact memories using stale ranking and tiering (scoped/attested/high commit count) plus contextual filtering for active subsystems.
  • Strict checklist for quality: Validates provenance sources, freshness/verification recency, scope correctness against your queries, visibility discipline, actionability, and duplication/conflicts.
  • Thread hygiene enforcement: Ensures context-dependent threads do not linger by moving anything not recently verified to archived/superseded or converting it into a durable, properly scoped record.
  • Outcome-based results: Produces verified, corrected, superseded, archived/obsolete, or promoted-from-draft outcomes so future agents default to accurate guidance.
  • Link integrity verification: Checks [[...]] references via a dry-run resolution to avoid persisting broken or renamed targets.

Quick Start

Use the reviewing-memory skill when stability matters by running a staleness-ranked review of memories and then correcting, superseding, or archiving each surfaced item until the top results match your common path/command/tag queries and are recently verified.

Frequently Asked Questions about reviewing-memory

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

FAQPage Schema
How do I prevent stale agent memory from causing drift during large refactors?

To prevent stale agent memory drift, run a staleness-ranked review workflow that surfaces high-impact memories, then correct, supersede, or archive each item based on strict provenance and freshness checks.

What is spec-driven development memory review and when do I need it?

Spec-driven development memory review is the process of auditing outdated guidance records. You need it during releases, migrations, onboarding unfamiliar subsystems, or when repeated agent confusion indicates accumulated thread-type working sets.

How do I validate provenance and scope correctness for agent memory records?

Validate provenance and scope correctness by applying a strict checklist that verifies source links, freshness recency, and actionability against your current queries, ensuring each memory record matches its intended operational boundary.

What is the best way to clean up conflicting context-dependent threads in agent memory?

The best way to clean up conflicting threads is through thread hygiene enforcement, which moves anything not recently verified to archived or superseded status, or converts it into a durable, properly scoped record.

How do I verify link integrity for references in agent memory before a release?

Verify link integrity by performing a dry-run resolution of [[...]] references to detect and avoid persisting broken or renamed targets within your agent memory records before finalizing updates.

Does agent memory staleness review handle draft promotion gating for new guidance?

Yes, agent memory staleness review includes optional draft promotion gating, allowing you to evaluate and promote verified draft records into durable guidance only after they pass strict quality and actionability checks.