mem-search
Search claude-mem's persistent cross-session memory database. Use when user asks "did we already solve this?", "how did we do X last time?", or needs work from previous sessions.
Security Assessment
About mem-search
The mem-search skill provides a robust mechanism to query and retrieve information from claude-mem's persistent cross-session memory database. It is designed to help users recall previous interactions, decisions, and work across multiple sessions, addressing the common problem of needing context from past conversations without manually sifting through historical records. By leveraging this skill, users can quickly determine whether certain issues were resolved, review the approach taken in prior work, and maintain continuity across projects or tasks without redundancy.
This skill offers a structured 3-layer workflow: search, timeline, and fetch. The search layer allows users to retrieve a filtered index of relevant records using specific query parameters, such as project, type, and date range, returning concise summaries of past observations. The timeline layer enables contextual exploration around selected observations, showing related items before and after a central anchor to provide temporal context. Finally, the fetch layer allows users to obtain the complete details of selected records efficiently, consolidating multiple observations into a single request to save time and tokens. The workflow emphasizes filtering prior to fetching to optimize performance and reduce unnecessary data retrieval.
Typical use cases include developers and project managers revisiting prior bug fixes, feature implementations, or design decisions; analysts needing historical context for recurring issues; and any user who wants to maintain a coherent understanding of ongoing projects without manually tracking each session. This skill is ideal for scenarios where continuity and efficiency are critical, enabling users to retrieve actionable insights from past work rapidly and accurately, ensuring informed decision-making and consistent progress across sessions.
FAQ
When should I use mem-search?
Use mem-search when you need information from previous sessions, such as checking if an issue was already resolved, reviewing past solutions, or recalling historical work. It is not intended for current session data.
What types of records can mem-search retrieve?
Mem-search can search observations, sessions, and prompts. Observations can be further filtered by types such as bugfix, feature, decision, discovery, or change.
How does mem-search optimize data retrieval?
It uses a 3-step workflow: search to get concise indexes, timeline to view contextual sequences around selected items, and fetch to retrieve full details only for relevant IDs. This filtering-first approach reduces token usage and minimizes unnecessary data fetching.
Are there limits on search results?
Yes, the search tool allows setting a limit up to 100 results per query. Timeline depth and fetch parameters also have maximum limits to ensure efficient retrieval.
Can mem-search be used with knowledge agents?
Yes, for synthesized or queryable answers from past observations, you can integrate mem-search results with knowledge agents to build a corpus from historical data.
Install mem-search
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
thedotmack/claude-mem