mem-search

Search claude-mem's persistent cross-session memory database using a three-layer search, timeline, and fetch workflow.

Updated Mar 26, 2026
One-click install
npx skills add https://github.com/LGJ-Jonathan/Jonathan_Global_Claude_Skills --skill mem-search-lgj-jonathan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mem-search
Source: https://github.com/LGJ-Jonathan/Jonathan_Global_Claude_Skills/tree/main/claude-mem/plugin/skills/mem-search
Command: npx skills add https://github.com/LGJ-Jonathan/Jonathan_Global_Claude_Skills --skill mem-search-lgj-jonathan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When working across multiple sessions, developers lose track of how past problems were solved. This Skill retrieves observations, sessions, and prompts from claude-mem's persistent memory database so you can answer questions like "did we already fix this?" without re-reading entire conversation histories. ## Core Features & Use Cases - Indexed Search: Query the memory database with filters for project, observation type (bugfix, feature, decision, discovery, change), and date ranges, returning compact results with IDs. - Timeline Context: Retrieve chronological context around any observation to understand what happened before and after a specific event. - Batch Fetching: Pull full details for multiple filtered observation IDs in a single request, achieving roughly 10x token savings over fetching everything. - Use Case: A user asks "how did we handle JWT authentication last month?" The Skill searches for matching observations, reviews the timeline around relevant results, then fetches full details only for the IDs that matter. ## Quick Start Search my claude-mem memory for how we fixed the authentication token expiration bug in previous sessions.

Frequently Asked Questions about mem-search

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search past Claude sessions for previous work?

Use the search MCP tool with a query string and optional filters like project, type, obs_type, and date range. It returns a compact index of matching observations with IDs, timestamps, and titles that you can then fetch in full.

How do I find what happened around a specific observation?

Use the timeline MCP tool with an anchor observation ID and depth_before/depth_after parameters. It returns surrounding observations, sessions, and prompts in chronological order so you can understand the context of an event.

Can I filter memory search results by date or type?

Yes, the search tool accepts dateStart and dateEnd parameters in YYYY-MM-DD or epoch milliseconds, plus obs_type filters for bugfix, feature, decision, discovery, or change. You can also filter by project name and observation type.

Why should I filter results before fetching full observation details?

Search index results cost roughly 50-100 tokens each while full observations cost 500-1000 tokens. Filtering first and batch-fetching only relevant IDs with get_observations yields about 10x token savings and fewer requests.

Does memory search cover the current conversation?

No, this Skill targets previous sessions stored in the persistent memory database, not the active conversation. Use it when users ask about past work such as "did we already solve this" or "what happened last week."