SibylCross-agent memory for AI coding tools

The Problem ​

Every app wants to remember you now. Your assistant imports your history from the last one, builds a profile, and turns the context you generate into someone else's asset. And none of it travels: what you taught Claude Code, Cursor never sees, so you re-explain the same project to every tool you use. The question stopped being whether your tools remember and became who the memory works for.

The Solution ​

Sibyl is cross-agent memory where the answer is you. A knowledge graph, memory loop, and task workflow shared across every coding agent you use, self-hosted on your own hardware so the memory works for you and stays yours:

• Memory: Store patterns, decisions, and solutions that persist across sessions
• The Memory Loop: context → act → remember → reflect, built into every surface
• Task Tracking: Manage work across sessions with full lifecycle support
• Semantic Search: Find knowledge by meaning, not exact keywords
• Synthesis: Generate verified documents grounded in your own memory
• Source Ingestion: Crawl external docs and import sources into one graph

How It Works ​

flowchart TD
    W["Your Actual Workflow<br/>Claude Code · Editors · Scripts · Teammates"]
    S["Skills<br/>Teach the workflow"]
    H["Hooks<br/>Inject context"]
    C["CLI<br/>Express power"]
    SRV{{"Sibyl Server<br/>SurrealDB graph · Semantic search · Memory loop · Tasks"}}
    UI["Web UI<br/>Human collaboration · Memory workspace · Graph explorer"]
    W --> S & H & C
    S & H & C --> SRV
    SRV --> UI

Skills + Hooks ​

Skills teach your tools and teammates structured workflows:

# Load context before implementing
sibyl context "authentication work" --intent build

# Track work with full lifecycle
sibyl task start task_xyz

# Capture learnings when done
sibyl task complete task_xyz --learnings "OAuth tokens need refresh..."

Hooks automatically inject relevant knowledge into every prompt so useful context shows up before you have to go looking for it.

For Humans: Web UI + CLI ​

Web UI for collaboration and oversight:

• Visual knowledge graph exploration
• Project and task management
• The memory workspace: captures, imports, and synthesis
• Document source configuration
• Team-wide dashboards

CLI for power users and scripting:

• The full memory loop from the terminal
• Semantic search and graph navigation
• Task lifecycle management
• Source crawling and ingestion

Quick Start ​

Install the CLI and connect it to your Sibyl server in one line:

# macOS
brew install hyperb1iss/tap/sibyl && sibyl setup https://sibyl.example.com

# Linux, or anywhere with uv
uv tool install --upgrade sibyl-dev && sibyl setup https://sibyl.example.com

sibyl setup signs you in and installs the Sibyl skill and the Claude Code hook. To let your agent do it, give it the sentence on the web app's Connect card, such as Set up Sibyl on this machine by following https://sibyl.example.com/agent (other deployments). No server yet? Installation starts one locally with one command.

Then run the loop:

# Capture a learning
sibyl remember "Redis insight" "Pool size must be >= concurrent requests" --kind rule

# Load it as working context
sibyl context "redis connection pool"

# Manage tasks
sibyl task list --status doing
sibyl task complete <task_id> --learnings "OAuth tokens expire after 1 hour"

The Memory Loop ​

Sibyl is built around a durable cycle that both humans and agents follow:

flowchart LR
    R(["CONTEXT<br/>Pull working context before you act"])
    A(["ACT<br/>Do the work with context in hand"])
    M(["REMEMBER + REFLECT<br/>Capture learnings, distill session notes"])
    R --> A --> M
    M -.->|the graph gets smarter| R

Every completed task makes your knowledge graph smarter. Every pattern discovered helps future sessions move faster. The system learns as you work.

Why Sibyl? ​

Get Started ​

1. Installation Set up Sibyl in five minutes
2. Quick Start Your first knowledge graph session
3. Skills & Hooks Teach your tools and teammates the workflow
4. Capturing Knowledge Use the memory loop well

Built for projects that deserve to remember.