FAQ & Troubleshooting

Yes. CogmemUI is a free hosted service at HiFriendbot.com. You bring your own Ai provider API keys, so you only pay for the Ai usage you consume directly with each provider.

No. You only need one API key to get started. CogmemUI supports 6 providers, but you can use just one. Add more whenever you want to explore other models.

Yes. All API keys are protected with quantum-safe encryption before storage. Your conversations are stored securely on our server and are never shared with third parties. All endpoints require authentication and verify resource ownership to prevent unauthorized access.

When you send a message, it goes directly to the Ai provider you selected — we don't log or store the content beyond what's needed for your conversation history.

Yes. CogmemUI is a Progressive Web App (PWA) that works on any device. You can install it to your home screen for a native app experience. The interface is fully responsive and optimized for touch.

CogmemAi is our persistent memory service for Ai assistants — and it’s far more than storage. The Intelligence Engine makes your memory smarter over time: semantic search finds memories by meaning, related knowledge auto-links into a knowledge graph, contradictions are flagged automatically, and you can ask a question and get one synthesized answer from all your memories. Auto-Skills takes it further — your corrections become behavioral rules that teach your Ai how to work with you. Correct a mistake once, and it never happens again. It runs in three modes: cloud (recommended — full intelligence), local (zero-config, no account needed), or hybrid (both, with offline fallback). Free to start, gets smarter every time you use it.

CogmemAi offers three storage modes: Cloud (recommended) gives you the full Intelligence Engine — semantic search, auto-linking knowledge graph, contradiction detection, self-improving recall, auto-skills, query synthesis, and team collaboration. Local is a zero-config option that requires no API key or account — great for a quick start or offline environments, with basic keyword search. Hybrid saves to both local and cloud, reading from cloud when available and falling back to local when offline. CogmemUI uses cloud mode. Developers using the MCP server in coding editors can choose any mode via npx cogmemai-mcp setup.

When you select "Auto" as your model, CogmemUI automatically picks the best model for each message. It analyzes the task type (code, creative writing, quick questions, reasoning, math, translation) and maps it to the optimal model from your available API keys. It also respects your budget settings.

Yes. CogmemUI supports teams with shared conversations, team memories, presence indicators (see who's online), typing awareness, and role-based access (owner, admin, member). Each team member manages their own API keys and preferences independently. Teams require a CogmemAi Teams subscription ($49.99/seat/month).

Go to Settings > Teams and click Create Team. Enter a team name and you become the owner. Then invite members by username. Each member needs a CogmemAi Teams tier subscription to participate.

Teams have three roles: Owner (full control, can delete team), Admin (can invite/remove members and manage conversations), and Member (can view shared conversations and team memories). The person who creates the team is automatically the owner.

Team memories are CogmemAi memories scoped to your team instead of just you. When team mode is active, the Ai can save and recall memories shared across all team members — decisions, protocols, best practices, and institutional knowledge. Every team memory includes author attribution so you always know who contributed it. Switch between personal and team memory scope in the Memory Panel.

When you ask the Ai a question with team mode active, CogmemAi automatically searches your team’s shared memory and injects the most relevant knowledge into the conversation. You don’t have to search for it or even know it exists — if a teammate documented it, the Ai surfaces it automatically. This is the core of CogmemAi Teams: the knowledge base that seeks you, not the other way around.

Yes. You can belong to as many teams as you need and switch between them with one click in the header dropdown. Each team has its own shared memory, shared conversations, and member list — completely isolated from each other. Common setups include a team per project, per department, per client, or per research group.

Absolutely. Personal memories and team memories are completely isolated. Your personal memories are never visible to teammates. Team memories belong to the team. When team mode is off, the Ai only accesses your personal memories. You can toggle between scopes in the Memory Panel. There is zero bleed between personal and team data.

Nobody. Conversations are private by default, even with team mode active. Only conversations you explicitly share using the “Share with Team” option become visible to team members. Team owners and admins cannot see your private conversations.

Images (JPEG, PNG, GIF, WebP), PDFs, Word documents (DOCX), spreadsheets (XLSX), and text files (TXT, CSV, Markdown, HTML, JSON). Images are automatically optimized before upload to reduce API costs.

Extensions let you add custom slash commands, tools, UI panels, and system prompt fragments. Create extensions in Settings > Extensions or install from a URL. All URL installs require HTTPS and show a code preview before execution for security review.

Yes. Go to Settings > Budget to set daily and monthly cost limits. You can also enable auto-downgrade to automatically switch to a cheaper model when approaching your limit. A color-coded budget indicator in the header shows your current status at a glance.

Yes! Install the CogmemUI Local Companion to route chat through your Claude Code subscription. Run npm install -g cogmemui-local, then cogmemui-local, paste the token into Settings > Local Server, and enable “Use Local Ai”. On Windows, cogmemui-local setup creates a desktop shortcut and auto-start on login. Your CogmemAi memory works automatically.

Model Context Protocol (MCP) is an open standard for connecting Ai models to external tools and data sources. It lets your Ai interact with databases, APIs, file systems, and custom tools through a standardized interface. CogmemUI supports MCP servers that use the Streamable HTTP transport.

Go to Settings > MCP Servers and click Add Server. Enter a server ID (slug), display name, the server’s Streamable HTTP endpoint URL, and choose an auth type if needed (none, basic, or bearer token). Then click Discover Tools to auto-detect the server’s tools.

Any MCP server that supports Streamable HTTP transport works with CogmemUI. This includes popular servers for file systems, databases, web search, code execution, and more. Check the MCP specification for a directory of available servers.

Try these steps: 1) Click Discover Tools on the server to refresh the tool list. 2) Verify the endpoint URL is correct and the server is running. 3) Check that auth credentials are correct if the server requires authentication. 4) Make sure the server is enabled (toggle it on in Settings > MCP Servers).

Check that you have a valid API key for the currently selected model. Open Settings > API Keys and verify the key for your provider is entered correctly. Also check that you haven't exceeded your budget limits in Settings > Budget.

Voice input uses the Web Speech API, which is only available in Chrome and Edge. Make sure your browser has microphone permissions enabled for this site. If the mic button is not visible, your browser doesn't support the API.

Screen capture uses the getDisplayMedia API, supported in Chrome, Edge, and Firefox. Safari does not support this feature. If the screenshot button is not visible, your browser doesn't support the API.

Make sure you have a valid API key for that model's provider. Some models require specific API plans or have usage limits. Try switching to a different model to confirm the issue is provider-specific.

If you see a 429 rate limit error, wait a moment and try again — the provider is temporarily limiting your requests.

Type /theme in the chat input and choose from Light, Dark, Auto, or Terminal. You can also change it in Settings > Appearance.

The file browser uses the File System Access API, which requires Chrome or Edge. On other browsers, you can still attach files via the paperclip button, or configure an MCP filesystem server in Settings for full file access.

Try refreshing the page. If the issue persists, check your browser's developer console (F12) for error messages. Make sure you're logged in — if your session expired, you'll need to log in again.