reqall gives your AI tools a shared knowledge graph — issues, specs, architecture decisions, and relationships — so every agent picks up right where the last one left off.
Issues, specs, architecture decisions, and test plans connected by typed links — blocks, implements, tests, parent, related.
Every record is embedded as a 1536-d vector. Agents find relevant context instantly via cosine similarity, with automatic deduplication.
A single Model Context Protocol endpoint lets Claude Code, Cursor, Copilot, and other agents read and write to the same knowledge base.
Organize records into projects. Agents auto-detect which project they're working in and scope searches accordingly.
Walk the link graph from any node to discover everything it blocks, implements, or depends on — recursively.
Create named API keys for each agent or environment. Keys are SHA-256 hashed at rest — the plaintext is shown only once.
Authenticate with your Google account in one click.
Generate keys for each of your AI coding tools.
Point your agents at the MCP endpoint with their key.
Agents read and write issues, specs, and decisions as they work.
First-party plugins for the agents you already use.
Active project namespaces.
Issues, specs, and architecture decisions.
Active agent credentials.
| Name | Created | Actions |
|---|
Install a reqall plugin for your AI coding agent to give it persistent memory. Each plugin connects your agent to the reqall knowledge graph via MCP.
All plugins require an API key. Create one from the API Keys tab, then set it as an environment variable:
export REQALL_API_KEY="your-api-key-here"
You can also add this to your shell profile (~/.bashrc, ~/.zshrc, etc.) so it persists across sessions.
Full-featured plugin with lifecycle hooks, skills, and automatic context injection. Works with Claude Code's native plugin system.
/plugin install https://github.com/ReqallSystem/claude-plugin
reqall:issues, reqall:specs, reqall:projects, reqall:review-issues, reqall:review-specsMCP server integration with a Cursor rules file for automatic context injection.
.cursor/mcp.json:
{
"mcpServers": {
"reqall": {
"url": "https://reqall.net/mcp",
"headers": {
"Authorization": "Bearer ${REQALL_API_KEY}"
}
}
}
}
.cursor/rules/reqall.mdc:
---
description: Reqall persistent knowledgebase for AI agent memory
globs: **/*
alwaysApply: true
---
# Reqall — AI Agent Memory
You have access to the Reqall knowledgebase via MCP tools
prefixed with `reqall:`. Use reqall:search to find context
and reqall:upsert_record to save work.
MCP server integration for VS Code with GitHub Copilot agent mode.
.vscode/mcp.json:
{
"servers": {
"reqall": {
"type": "http",
"url": "https://reqall.net/mcp",
"headers": {
"Authorization": "Bearer ${REQALL_API_KEY}"
}
}
}
}
Declarative adapter with AGENTS.md instructions and MCP server config for the Codex CLI.
config.toml):
[mcp.reqall]
url = "https://reqall.net/mcp"
[mcp.reqall.headers]
Authorization = "Bearer ${REQALL_API_KEY}"
AGENTS.md so Codex knows how to use the reqall tools.
See the codex-plugin package for the full file.
MCP server integration for the Gemini CLI with agent instruction support.
gemini-extension.json:
{
"name": "reqall",
"description": "Persistent semantic knowledgebase for AI agents",
"mcpServers": {
"reqall": {
"url": "https://reqall.net/mcp",
"headers": {
"Authorization": "Bearer ${REQALL_API_KEY}"
}
}
}
}
GEMINI.md.
See the gemini-plugin package for the full file.