memory-systems

Guides implementation of agent memory systems, compares production frameworks (Mem0, Zep/Graphiti, Letta, LangMem, Cognee), and designs persistence architectures for cross-session knowledge retention. Use when the user asks to "implement agent memory", "persist state across sessions", "build knowledge graph for agents", "track entities over time", "add long-term memory", "choose a memory framework", or mentions temporal knowledge graphs, vector stores, entity memory, adaptive memory, dynamic memory or memory benchmarks (LoCoMo, LongMemEval).

Install with Tessl CLI

npx tessl i github:muratcankoylan/Agent-Skills-for-Context-Engineering --skill memory-systems

What are skills?

1.33x

Quality

92%

Does it follow best practices?

Impact

83%

1.33x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./skills/memory-systems/SKILL.md

SKILL.md

Review

Evals

Evaluation results

77%

31%

Memory System Architecture Proposal

Memory architecture selection and progression

Criteria

Without context

With context

Start simple first

50%

30%

Progression stages

100%

Trigger for scaling

62%

75%

Graphiti vs Cognee distinction

90%

Reliability over complexity

50%

40%

Benchmark reference

100%

Mem0 user scoping

100%

Complex reasoning trigger

100%

Structured JSON filesystem

25%

12%

No premature complexity

100%

Framework trade-offs

90%

100%

Without context: $0.2508 · 1m 40s · 10 turns · 59 in / 4,568 out tokens

With context: $0.3076 · 1m 42s · 11 turns · 11 in / 4,431 out tokens

87%

21%

Employee Knowledge Graph for an HR Assistant Agent

Temporal validity and error recovery

Criteria

Without context

With context

valid_from tracking

100%

valid_until tracking

100%

Invalidate not delete

100%

Conflict resolution rule

100%

Conflict surfacing

37%

Empty retrieval fallback

25%

100%

Consolidation trigger

25%

100%

Stale result handling

37%

Non-blocking storage write

25%

100%

Storage failure queueing

75%

100%

History preservation rationale

100%

Consolidation scope

83%

50%

Without context: $0.4099 · 2m 19s · 11 turns · 12 in / 8,181 out tokens

With context: $0.7468 · 2m 59s · 22 turns · 80 in / 10,227 out tokens

85%

Context-Aware Memory Integration for a Customer Support Agent

Memory-context integration and anti-patterns

Criteria

Without context

With context

JIT retrieval

100%

Attention-favored placement

70%

100%

Hybrid retrieval

60%

No context stuffing

100%

Context size limit

100%

87%

Retention policy

100%

Privacy by design

100%

Memory growth monitoring

62%

100%

Retrieval latency monitoring

100%

Entity extraction for retrieval

25%

50%

No blocking on write

80%

40%

Without context: $0.4536 · 2m 52s · 13 turns · 13 in / 9,326 out tokens

With context: $0.9329 · 3m 53s · 26 turns · 26 in / 12,839 out tokens

Evaluated: 5 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Memory System Architecture Proposal Employee Knowledge Graph for an HR Assistant Agent Context-Aware Memory Integration for a Customer Support Agent

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.