Like I Said - Memory MCP Server

Memory persistence MCP server for AI agents — enabling agents to store, retrieve, and search memories across conversations. Provides local storage for facts, preferences, and context that persists beyond individual conversation sessions. Enables AI assistants to remember previous interactions, user preferences, project context, and learned information without relying on the LLM's ephemeral context.

Evaluated Mar 06, 2026 (0d ago) vcurrent
Homepage ↗ Repo ↗ AI & Machine Learning memory persistence mcp-server context recall ai-memory local
⚙ Agent Friendliness
78
/ 100
Can an agent use this?
🔒 Security
85
/ 100
Is it safe for agents?
⚡ Reliability
68
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
68
Documentation
70
Error Messages
68
Auth Simplicity
100
Rate Limits
100

🔒 Security

TLS Enforcement
85
Auth Strength
90
Scope Granularity
80
Dep. Hygiene
75
Secret Handling
90

Local files. No credentials. Memory may contain sensitive info. Protect storage directory access. No external exposure.

⚡ Reliability

Uptime/SLA
70
Version Stability
68
Breaking Changes
65
Error Recovery
68
AF Security Reliability

Best When

An AI agent or assistant needs to remember information between conversations — providing continuity and personalization that the LLM's context window alone cannot provide.

Avoid When

You need team-shared memory, distributed storage, or large-scale semantic search — use vector databases (ChromaDB, Pinecone) for those requirements.

Use Cases

  • Persisting conversation context and learned facts across sessions from assistant agents
  • Storing user preferences and settings for personalized agent behavior from personalization agents
  • Building up knowledge about projects and codebases over time from development agents
  • Remembering previous decisions and their rationale from decision-support agents
  • Maintaining relationship and interaction history from communication agents

Not For

  • Shared team memory requiring multi-user access (this is personal/single-agent memory)
  • High-volume vector search (use ChromaDB or Weaviate for large-scale semantic memory)
  • Production applications requiring high-availability memory (local file-based storage)

Interface

REST API
No
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
No
Webhooks
No

Authentication

Methods: none
OAuth: No Scopes: No

No authentication — local file-based memory storage. Data stored in a local directory (configurable). No external service required.

Pricing

Model: free
Free tier: Yes
Requires CC: No

Free local memory persistence tool. No external API costs.

Agent Metadata

Pagination
none
Idempotent
Partial
Retry Guidance
Not documented

Known Gotchas

  • Memory quality depends on what the agent chooses to store — implement memory curation strategies
  • Local file storage means memory is machine-specific — no sync across devices
  • Memory can grow unbounded — implement periodic cleanup to remove stale or irrelevant memories
  • Retrieval quality depends on search implementation — semantic vs keyword search affects recall
  • No conflict resolution for contradictory memories — implement logic to handle outdated facts

Alternatives

Full Evaluation Report

Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Like I Said - Memory MCP Server.

$99

Scores are editorial opinions as of 2026-03-06.

5182
Packages Evaluated
26151
Need Evaluation
173
Need Re-evaluation
Community Powered