Archive and Consolidate

Archives completed workflow changes and consolidates extracted knowledge into persistent stores.

2|Updated Jul 6, 2026
One-click install
npx skills add https://github.com/NAMEWTA/learning-open-code --skill archive-and-consolidate-namewta
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Archive and Consolidate
Source: https://github.com/NAMEWTA/learning-open-code/tree/main/speculo/skills/archive-and-consolidate
Command: npx skills add https://github.com/NAMEWTA/learning-open-code --skill archive-and-consolidate-namewta

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) and assets (resource) components.

What problem does it solve? Completed changes pile up in a workflow's active directory while the lessons, decisions, and terminology they contain remain scattered and eventually lost. This Skill archives finished changes into monthly folders, graduates durable knowledge into ADR and context stores, and audits those stores for stale or duplicated content. ## Core Features & Use Cases - Atomic batch archiving: Moves completed changes from changes/ to archive/YYYY-MM/ with pre-flight checks, status.json updates, and post-move re-read verification. - Knowledge graduation and consolidation: Applies graduation criteria (stable mechanism, repeated lesson, must-know) to extract architecture decisions into adr/ and domain terms into context/, with conflict detection and supersede handling. - Cleanup auditing with dry-run safety: Classifies store entries as delete, merge, rewrite, keep, or needs-confirmation, and defaults to a dry-run plan requiring explicit user confirmation before any destructive action. - Use Case: After finishing several spec-driven changes in a project, run this Skill to archive them, promote the JWT authentication decision into adr/0003-jwt-auth.md, merge new domain terms into the glossary, and receive a reviewed list of outdated ADRs safe to delete. ## Quick Start Ask the agent to archive all completed changes in the current workflow and consolidate their knowledge into the declared stores, starting with a dry-run plan for confirmation.

Frequently Asked Questions about Archive and Consolidate

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

FAQPage Schema
How do I archive completed changes in a spec-driven workflow?▼

Run the Skill in dry-run mode first to get a full plan mapping each completed change from changes/ to archive/YYYY-MM/. After you explicitly confirm, it performs atomic moves, updates status.json, and re-reads everything to verify consistency.

How does knowledge consolidation decide what to extract from a change?▼

It applies three graduation criteria: stable mechanisms, lessons repeated across more than one change, and must-know context for future developers. Qualifying architecture decisions go to adr/ with auto-assigned numbers, and domain terms merge into context/ files.

Does the archive process modify files without confirmation?▼

No. The default mode is dry-run, which only produces a plan covering archive moves, knowledge merges, and cleanup candidates. Destructive actions like moves, deletions, and rewrites execute only after explicit user approval in confirmed mode.

What happens if an archive target directory already exists?▼

The batch is blocked entirely because archiving uses all-or-nothing atomicity. The plan reports the specific conflict, and existing archives are never overwritten; you must resolve the collision manually before rerunning.

What is the mechanical-only mode and when should I use it?▼

Mechanical-only mode handles only directory moves and status updates while the caller owns all knowledge policy. Use it when another command or workflow already decided what knowledge to write, so this Skill never reads, merges, or cleans knowledge stores.

Why did the cleanup scan mark an ADR as needs-confirmation instead of delete?▼

Rule modifications, conflicting term definitions, substantive ADR rewrites, and changes to non-standard stores always require human judgment. The Skill presents both versions with a recommendation rather than auto-resolving contradictions.