Proposed specificationSource-located exemplarH3

Agent-maintained memory pipelines

Update reusable task memory.

JOIN POINTreusable_memory_updateRECORD STATUSCandidate · v0.1.0

UML class view

The proposed contract drawn as participant roles, received calls, and numbered dependencies.

Object interactions

The same roles and calls as a class collaboration and a numbered message sequence.

Object interaction diagram for Agent-maintained memory pipelines, showing role collaboration and a numbered message sequence — FIG. P28 / OBJECT INTERACTIONS
Open full SVG Download PDF

Pointcut

Where the mechanism applies. All four dimensions belong to the proposed match, not just its vocabulary.

Context: The harness selects and stores knowledge across tasks or sessions.
Trigger: An observation, reflection, or memory write is processed.
Structure: Selection, consolidation, provenance, access control, retrieval, and revision.
Effect: Stored memory is traceable, retrievable within scope, and updateable.

Protected contract

The obligation that clients, operators, and auditors should be able to rely on.

Requires: Candidate memory has a source and intended scope.
Ensures: Retrieval identity and revision history survive consolidation.
Invariant: Untrusted memory cannot silently redefine instructions or permissions.
Violation: Stale, poisoned, or conflicting memory is quarantined or rejected.

Negative control

A chat transcript retained in memory is not an update pipeline.

Variation & lifecycle

The hot spot is admitted only where the protected semantics survive.

Permitted variation

Observational memory, vector store, summary file, scoped project memory.

Test obligation

Test source provenance, poisoning, stale update, deletion, and cross-scope read.

Evolution question

Did persistent memory add provenance and poisoning defenses?

Prompt + code design

An illustrative prompt and a small design sketch for the proposed contract. Names and APIs are invented; this is not a repository excerpt.

PROMPT SKETCH

Write candidate memories with a source and intended scope. Consolidate without losing retrieval identity or revision history, and make memory retrievable only within its declared scope.

CODE DESIGN / PYTHON

def consolidate(candidate, store):
    existing = store.find(candidate.key, scope=candidate.scope)
    if existing and existing.source == candidate.source:
        return store.revise(existing.id, candidate.content)
    record = MemoryRecord(
        key=candidate.key,
        content=candidate.content,
        source=candidate.source,
        scope=candidate.scope,
        revision=1,
    )
    return store.insert(record)

Illustrative design sketch. A memory without a source or scope cannot be audited or safely revoked. This example is a teaching aid, not implementation evidence or a conformance claim.

Detection & matching

The frozen rule that retrieves this record, the canonical source exemplar, and the control that should stay unmatched.

Join point: reusable_memory_update
Source locator: Barbaste et al., §9.6, p. 33 (“Persistent Memory Pipelines”); Table 12 “Agent-Maintained Memory”, p. 53; source label
Lexical cue: (?i)\b(?:memory[_-]?(?:pipeline|store|update|manager)|remember|memories)\bFrozen expression, v0.1.0. A match is a retrieval cue, not a confirmed implementation.
Aliases: None
Canonical exemplar: mastra-ai/mastrapackages/memory/src/processors/observational-memory/observational-memory.ts:423Support: packages/memory/src/processors/observational-memory/__tests__/observational-memory.test.ts:1
Negative control: A chat transcript retained in memory is not an update pipeline.

Scope-anchored review. pohe-agent analyze <repo> --pattern P28 anchors the cue to its enclosing declaration and checks a separate operational signature for all four pointcut dimensions. Missing roles and negative-control risks remain visible; a structural match is not E2/E3. Calibration report ↗

Repository files

Up to five public files retrieved by the frozen cue and pinned to the audited commit. A match is a place to inspect, not a confirmed implementation.

01mastra-ai/mastrapackages/memory/src/processors/observational-memory/observational-memory.ts:423CANONICAL SOURCE 02aaif-goose/goosecrates/goose-cli/src/commands/configure.rs:1217RETRIEVAL CUE 03affaan-m/ECCskills/skill-comply/scripts/spec_generator.py:39RETRIEVAL CUE 04agno-agi/agnolibs/agno/agno/memory/manager.py:41RETRIEVAL CUE 05anomalyco/opencodepackages/cli/src/index.ts:105RETRIEVAL CUE

Evidence & provenance

Each layer answers a different question. Do not substitute retrieval for review.

Source-located exemplar

A concrete source anchor is available for this proposed record. It is one exemplar, not evidence that all repositories—or every contract field—conform.

Review boundary: this candidate and its specification come from an AI-assisted study. Independent human validation remains open.

ANY-EVIDENCE FAMILY COVERAGE34/36

94.4% of repository families in the frozen panel have at least one current cue, historical cue, or canonical source layer. This is not confirmed prevalence.

Current lexical-cue repositories: 38 / 41

Canonical source anchormastra-ai/mastra 899383db98e0packages/memory/src/processors/observational-memory/observational-memory.ts:423

Supporting surfacepackages/memory/src/processors/observational-memory/__tests__/observational-memory.test.ts:1

Observation, reflection, and persistence pipeline.

Exploratory retrieval cue Regex / v0.1.0

This expression is a discovery instrument only. A match is a retrieval cue—not a confirmed implementation, pointcut match, or contract proof.