# AI Forgets Earlier Constraints

## Ontology Metadata

Code: ISS-0029
Version: ISS-0029@0.1.0
Ontology release: 0.1.0
Updated: 2026-05-10T00:00:00Z
Canonical URL: /workbench/issues/ai-forgets-earlier-constraints/

## Summary

A constraint, instruction, preference, or decision that should persist through the task stops affecting later output.

## Category

- CAT-0020 — Memory & Context
  - URL: /workbench/categories/memory-and-context/
  - Summary: Problems where task state, prior instructions, saved memory, restored context, stale information, or cross-task context affects behavior incorrectly.

## Primary Pattern

- PAT-0350 — Persistence Instability
  - URL: /workbench/patterns/persistence-instability/
  - Summary: A structural condition where persisted state cannot be stored and restored into an equivalent structure without alteration.

## Declared Patterns

- PAT-0350 — Persistence Instability
  - URL: /workbench/patterns/persistence-instability/
  - Summary: A structural condition where persisted state cannot be stored and restored into an equivalent structure without alteration.
- PAT-0170 — Constraints Underspecified
  - URL: /workbench/patterns/constraints-underspecified/
  - Summary: A structural condition where declared constraints are insufficient to eliminate ambiguity or multiple admissible states.
- PAT-0190 — Contract Drift
  - URL: /workbench/patterns/contract-drift/
  - Summary: A structural condition where a declared contract changes but connected structures, implementations, consumers, or expectations do not update in sync.

## Derived Primary Lenses

- LEN-0160 — Constraint Sufficiency Lens
  - URL: /workbench/lenses/constraint-sufficiency-lens/
  - Summary: Evaluates whether declared constraints are sufficient to eliminate structural degrees of freedom.
- LEN-0170 — Convergence Lens
  - URL: /workbench/lenses/convergence-lens/
  - Summary: Compares parallel structural systems to determine whether they align under shared authority.
- LEN-0270 — Reconciliation Lens
  - URL: /workbench/lenses/reconciliation-lens/
  - Summary: Evaluates whether structural changes align with declared authority updates, version changes, or reconciliation rules.
- LEN-0280 — Reference Stability Lens
  - URL: /workbench/lenses/reference-stability-lens/
  - Summary: Evaluates whether structural references, identifiers, nodes, and edges remain consistent across execution cycles or comparable states.

## Derived Secondary Lenses

- LEN-0180 — Determinism Lens
  - URL: /workbench/lenses/determinism-lens/
  - Summary: Evaluates whether identical structural inputs produce equivalent structural outputs across repeated executions.
- LEN-0290 — Variance / Entropy Lens
  - URL: /workbench/lenses/variance-entropy-lens/
  - Summary: Measures structural variability across repeated or comparable evaluations and identifies divergence beyond expected bounds.

## Search Intents

- AI forgets earlier constraints
- ChatGPT forgot instructions
- AI ignored earlier requirement
- model forgot previous constraint
- instruction stopped applying later
- AI lost context rule
