Guides

Practical docs for builders.

MCP, SDKs, auth, projects, troubleshooting, and self-hosting. This is the “how it actually works” layer.

Popular

ChatGPT + ctxvault

Create a Custom GPT with persistent memory. The easiest way to get started --- no coding, no terminal needed.

Read guide →

MCP (Codex, Claude, Cursor)

Interactive one-command setup, what it writes, how it chooses a project, and how to verify tools are live.

Read guide →

OpenClaw Agent

Run tasks, queue async jobs, and optionally write execution summaries back to memory.

Read guide →

Secrets Block

How outbound DLP works, configuration modes (redact/block/warn/off), and the ctxvault_security_status tool.

Read guide →

Scaffolds, Templates & Skills

Generate CLAUDE.md, AGENTS.md, .cursor/rules, and more from curated templates and skills.

Read guide →

Pulse --- Developer Comprehension

Passive concept tracking, LLM self-reporting, session digests, and your personal skill graph.

Read guide →

Claude Chat Connector (Remote MCP)

Use ctxvault inside claude.ai via a remote MCP connector (OAuth).

Read guide →

Projects + Default Project

How scoping works, slug rules, default project behavior, and why it prevents 'project not found' errors.

Read guide →

Authentication (Magic Links + API Keys)

Industry-standard passwordless sign-in for the web app, plus API key auth for SDK/MCP and service-to-service use.

Read guide →

Memory Model

What memory is (and isn't), lifecycle states, and how to store high-signal items that improve future sessions.

Read guide →

Knowledge Sources (GitHub)

Index a repository, sync it, and search it (hybrid semantic + keyword). Built for long-lived engineering context.

Read guide →

Troubleshooting

Fix common issues fast: wrong slug, missing auth, DNS/SSL mismatches, mail delivery problems.

Read guide →

Self-hosting

What it takes to run ctxvault in your infra (VM/Kubernetes), without open-sourcing your code.

Read guide →