skills.rest
Prismaibrowser
PrismaibrowserOFFICIAL
Explore all skills by this author →
GitHub

recall

Search agentmemory for past observations, sessions, and learnings using hybrid BM25, vector, and graph search.

3|Updated Oct 25, 2025
Search Tags:#context retrieval#recall#vector search#session history#BM25#memory search#agentmemory
One-click install
npx skills add https://github.com/Prismaibrowser/prismspace-web --skill recall-prismaibrowser
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: recall
Source: https://github.com/Prismaibrowser/prismspace-web/tree/main/.agents/skills/recall
Command: npx skills add https://github.com/Prismaibrowser/prismspace-web --skill recall-prismaibrowser

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When working across many sessions, decisions and learnings get lost. This Skill retrieves past context from agentmemory so you can answer questions like "did we ever decide how to cache sessions?" without digging through old logs. ## Core Features & Use Cases - Hybrid Memory Search: Queries agentmemory via memory_smart_search combining BM25, vector, and graph search to surface relevant observations. - Provenance-Aware Results: Groups results by session, respects provenance channels (user, agent, tool, import, shared), and prioritizes high-importance records (importance >= 7). - Hallucination Guardrails: Only surfaces what the tool returned; on empty results it suggests alternative search terms instead of inventing memories. - Use Case: A user asks "what did we do about the rate limiter?" and receives grouped results showing a per-IP counting code note from one session and a deferred sliding-window discussion from another. ## Quick Start Ask the assistant to recall what was decided about jwt refresh token rotation and it will search agentmemory and summarize the matching observations.

Frequently Asked Questions about recall

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

FAQPage Schema
How do I search past agent sessions for a decision?▼

Invoke memory_smart_search with your topic as the query and a limit of 10. Results return observations with type, title, narrative, session id, and importance score, which are then grouped by session with high-importance items first.

What search methods does agent memory recall use?▼

The recall search uses a hybrid approach combining BM25 keyword matching, vector similarity, and graph search through the memory_smart_search tool. This surfaces relevant observations even when exact wording differs from the original record.

What happens when a memory search returns no results?▼

When memory_smart_search returns zero results, the skill says so explicitly and suggests 2-3 alternative search terms, such as narrower synonyms or a session-scoped recap. It never invents observations, session ids, or importance scores.

How are conflicting memories from different sources handled?▼

Each record carries a provenance channel: user, agent, tool, import, or shared. When results conflict, user records are preferred over agent inference, and shared records are flagged as written by another teammate.

Can I scope a memory search to a specific project or repo?▼

Yes. Pass the project parameter to memory_smart_search when the user scopes the query to a specific repository. This filters results to observations recorded within that project context.

Related Skills

Explore More Software Engineering Skills→
joshka0joshka0
3

Remember

Save key learnings, decisions, and gotchas into foxctl memory with typed classification.

Community
Basic
TheycallmehollaTheycallmeholla

handoff

Compacts active conversation context into a structured handoff document.

Community
Intermediate
towrytowry

recent-history

Extract recent chat sessions from .claude/sessions/history.xml to restore user context.

Community
Advanced
psw7205psw7205
1

trace-change-why

Traces the rationale behind code changes from session transcripts and git history.

Community
Intermediate