Search hmem memory entries using keywords and time hints.
npx skills add https://github.com/YuryZZZ/opencode-global --skill hmem-search
Please help me install this Agent Skill. Skill: hmem-search Source: https://github.com/YuryZZZ/opencode-global/tree/main/skills/hmem-search Command: npx skills add https://github.com/YuryZZZ/opencode-global --skill hmem-search
This Skill helps retrieve relevant information from long-term memory when a user references past conversations without giving an ID, reducing the need to recall exact details manually.
Invoke the memory search to retrieve a relevant past-context entry based on your prompt.
High-intent search queries and answers about installing and using this skill.
Search past conversation history by providing a prompt referencing past context. The Skill extracts keywords and time hints, queries memory, and returns the most relevant summarized entries without needing an ID.
You can retrieve older context from dialogue history. The search applies time-window-aware retrieval that favors recent references but still allows surfacing older memory entries when the query requires them.
Drill down into a specific memory entry by using read_memory(id) with the identifier provided in the initial search results. This fetches the full details of that specific past context hit.
The best way to recall past context on demand is invoking a full-text memory search. It ranks results by relevance, summarizes the top hits, and offers drill-down options for specific dialogue history.
Yes, memory search works with vague keywords. It uses a two-step extraction process to identify keywords and time hints from your prompt, querying long-term memory to surface relevant past discussions automatically.
The main limitation of searching without exact details is reliance on keyword and time extraction accuracy. If the prompt lacks specific context references, the ranking may surface less relevant memory entries.
zeenie-aiSave, retrieve, and clear timestamped markdown conversation history with semantic search.
auroralabs-lociRender per-function timing, energy, stack, and memory trends from LOCI session context.
MburdoRetrieve distilled context and rules from past sessions via cm and cass commands.
AcosmiSearch long-term memory by keyword with semantic retrieval and relevance scores.
