documentation-budget

To prevent bloated AI documentation from wasting context tokens, classify documentation requests into risk-aligned tiers to eliminate redundant persistent artifacts and enforce pointer-over-duplication rules. This reduces future context load and prevents stale conflicting copies.

Documentation tier classification for AI workflows categorizes AI-consumed documentation requests into 5 risk-aligned tiers (0-4) based on change risk, future AI value, and necessity. This ensures proportional documentation cost and prevents unnecessary persistent artifacts from being created merely because templates exist.

To eliminate redundant content across specs, decision logs, and handoff documents, enforce pointer-over-duplication rules that cite canonical artifact owners instead of restating content. Clear ownership rules assign canonical ownership for every documentation artifact type to prevent duplication.

You need a full spec versus a change-note for AI-consumed documentation based on change risk and reversibility. A small, reversible bug fix is classified as a Tier 1 change-note instead of a full spec, saving hours of unnecessary writing and hundreds of tokens on future AI session loads.

Yes, documentation budget classification works for business logic indexes and project-internal skills. It applies to all project AI-consumed documentation including specs, decision logs, task lists, verification gates, handoff documents, change notes, and product docs.

A limitation of using pointer-over-duplication for AI context optimization is that canonical artifact owners must be maintained to prevent stale conflicting copies. If the original owned file is deleted or moved without updating pointers, the AI loses access to the necessary rationale and context.