npx agentspec-cli install e9855e06-6790-41c0-a688-953187cb54c4 --file-type windsurf-rules→ .windsurf/rules/
<!-- remb-instructions:v2 --> # Remb — AI Context Management You have access to Remb tools for persistent memory and context across coding sessions. Current project: **worldstreet-dashboard-revamp** ## MANDATORY Session Protocol Follow this protocol in EVERY session. Skipping causes knowledge loss. ### Session Start (do FIRST, before any other work): 1. Call `remb_loadProjectContext` — loads the full project context bundle (memories, features, tech stack). Without this, you have NO context about this project. 2. Call `remb_conversationHistory` — loads what was discussed and accomplished in prior sessions so you can pick up where the user left off. ### During Work: 3. Call `remb_conversationLog` after completing any significant task, bug fix, or feature — records what was done for future sessions. 4. Call `remb_createMemory` when you discover important patterns, architectural decisions, or gotchas worth preserving. ### Session End (do LAST, before the conversation ends): 5. Call `remb_conversationLog` with a summary of: what was asked, what was done, key decisions made. ## Available Tools | Tool | Purpose | When to Call | |------|---------|--------------| | `remb_loadProjectContext` | Full project context bundle — memories, features, tech stack | **Session start** (mandatory) | | `remb_conversationHistory` | Prior session history — what was done before | **Session start** (mandatory) | | `remb_conversationLog` | Record work done in this session | After completing tasks, and at session end | | `remb_saveContext` | Save feature-specific context or decisions | When you learn something about a specific feature | | `remb_getContext` | Retrieve context for a specific feature | When you need details about a feature | | `remb_listMemories` | Browse persistent memories | When searching for past decisions or patterns | | `remb_createMemory` | Save a new persistent memory | When discovering patterns, decisions, gotchas | | `remb_triggerScan` | Re-scan the codebase from GitHub or locally | After significant code changes | | `remb_scanStatus` | Check scan progress | After triggering a scan | ## Decision Matrix | Situation | Action | |-----------|--------| | Starting any session | `remb_loadProjectContext` + `remb_conversationHistory` | | Completing a task | `remb_conversationLog` with what was accomplished | | Found a reusable pattern | `remb_createMemory` with category "pattern" | | Made an architectural decision | `remb_createMemory` with category "decision" | | Discovered a gotcha or bug | `remb_createMemory` with category "gotcha" | | Need info about a feature | `remb_getContext` filtered by feature name | | User says "remember this" | `remb_createMemory` with appropriate tier | | Code changed significantly | `remb_triggerScan` to refresh context | | Ending the session | `remb_conversationLog` with session summary |
Remb — AI Context Management
You have access to Remb tools for persistent memory and context across coding sessions. Current project: **context-management**
by AgentHub Crawler
Build a Context Database for AI Agents
Mei leads AI platform at a 25-person company building AI agents for enterprise customers. Agents forget everything between sessions — a support agent asks the same questions it asked yesterday. Stuffing entire conversation history into the prompt burns tokens ($500/day on GPT-4) and hits context limits. Agents can't share knowledge: the sales agent learns a customer's preferences but the support agent doesn't know. They need a context database: store agent memories, retrieve relevant context via
by AgentHub Crawler
Soul
This project uses [Soul Protocol](https://github.com/qbtrix/soul-protocol) for persistent AI memory via MCP. **On session start:** 1. Call `soul_recall` with the current task context to load relevant memories
by AgentHub Crawler
Sign in to leave a comment.
Memorix — Automatic Memory Rules
You have access to Memorix memory tools. You MUST follow these rules to maintain persistent context across sessions. These rules are NOT optional — they are critical for cross-session memory continuity.
by AgentHub Crawler