# Moltis Personal agent server with voice, memory and browser automation. Canonical profile: https://ki-agentenkompass.de/en/agents/moltis/ Editor: [Michael Meierhoff](https://ki-agentenkompass.de/en/about/) Sources reviewed: 2026-10-16 Basis: provider documentation. Examples are editorial suggestions, not hands-on product tests. ## Overview Moltis is listed as a separate product profile. - Provider: Moltis project - Product family: Moltis - Profile type: Self-hosted personal agent - Environment: Own computer or server - Connections: Tools and channels according to configuration - Setup: Install through Homebrew, Docker or a release; connect a model. - Pricing model: Open source; model and operating costs are separate. ## Key features ### Use voice Set up speech recognition and synthesis through providers. Source: https://github.com/moltis-org/moltis ### Reuse context Use memory and cross-session search. Source: https://github.com/moltis-org/moltis ### Control a browser Perform web tasks with browser tools. Source: https://github.com/moltis-org/moltis ### Schedule work Trigger recurring work through schedules. Source: https://github.com/moltis-org/moltis ### Connect tools Connect MCP and messaging channels. Source: https://github.com/moltis-org/moltis ## Example workflows ### Try a voice briefing - Your input: A short voice message with a task. - Possible result: A response through the configured voice services. - What to check: Check content, permissions and actions taken. ### Prepare web research - Your input: A research question and approved websites. - Possible result: A sourced overview for manual review. - What to check: Record time spent, corrections and costs. ## Requirements and limits ### Setup Requirement: Configure the model, channels and permissions yourself. Source: https://github.com/moltis-org/moltis ### German in everyday use Not yet confirmed: End-to-end German interaction and output quality have not been tested here. Source: https://github.com/moltis-org/moltis ### Reliability Not yet confirmed: Success rates for specific workflows have not been measured here. Source: https://github.com/moltis-org/moltis ## A good fit if … - you want to run an agent server on your own hardware or a cloud instance - you want commands to run in containers controlled through approval_mode - you want to address the agent via Telegram, Slack, Signal or voice ## Less suitable if … - you want a hosted service with support and contract without running your own server - you need a solution without configuring approvals and monitoring yourself Editorial assessment based on provider documentation. Not a product test. ## Usable AI models Open choice You can use many providers with your own key or sign-in, OpenRouter and any OpenAI-compatible endpoint. Models run locally through Ollama, LM Studio or directly as GGUF files on your machine. - Model families named: Anthropic (Claude), OpenAI (GPT), Google (Gemini), DeepSeek, Mistral, xAI (Grok), MiniMax, Moonshot (Kimi), Z.ai (GLM) · Other models can be connected freely - Model choice: You choose - Your own API key: Yes - Router or custom endpoint: Yes - Local models: Yes Documented · Sources accessed on: 2026-11-05 - [Moltis: LLM Providers](https://docs.moltis.org/providers.html) - [Moltis: README](https://github.com/moltis-org/moltis) ## Europe & regional availability Installation instructions and model prerequisites reviewed. The assessment covers the local operating path described; complete country and feature availability of all connected services remains open. Reviewed scope: Local software and documented self-hosting; external services assessed separately Reviewed on: 2026-10-19 ### Self-hosting with own model access With requirements Moltis documents installation as a binary, through Homebrew or Docker on your own hardware. Local and external model providers are supported; check external models, voice services and channels separately. Region covered: Europe: own computer or server; model services are country-specific Documented reason: Technical assessment based on installation and provider selection. These sources do not establish a legal cause for regional restrictions. Official source: https://raw.githubusercontent.com/moltis-org/moltis/main/README.md ## Accounts & permissions Based on provider documentation. Permissions depend on the account, integration and organization policies. Review the permissions actually requested before use. Configuration covered: Self-hosted Moltis server (open source) Sources reviewed: 2026-11-05 ### Account used Documented: Server access uses a password, passkey or API key; API keys need at least one scope (read, write, admin, approvals, pairing). Sensitive data in the database, such as API keys and tokens, is stored encrypted. Evidence: https://docs.moltis.org/security.html ### Admin approval Documented with conditions: The approval_mode setting decides whether commands are always confirmed, confirmed by assessment (smart, the default) or never. A built-in blocklist still requires confirmation under "never". Evidence: https://docs.moltis.org/security.html ### Data access & team sharing Documented with conditions: Sandbox containers have no network access by default, and managed files are mounted read-only by default. Channels such as Telegram or Slack let outside parties talk to the agent, which requires approving senders. According to the README each agent has its own memory workspace. Evidence: https://docs.moltis.org/security.html Evidence: https://github.com/moltis-org/moltis ## Prices and included usage Pricing checked: 2026-10-16 Open source; model and operating costs are separate. Total cost depends on access, usage and configuration. Pricing details: https://github.com/moltis-org/moltis ## Operation & oversight Based on public provider documentation. Assessments are editorial inferences from the stated properties. Actual time requirements and reliability have not been measured. Deployment covered: Self-hosted Moltis server on your own machine or cloud instance Sources reviewed: 2026-11-05 ### Files and working state Documented: Sessions are stored as JSONL files and memory in SQLite with full-text and vector search. Sessions compact automatically, and recall works across sessions. Evidence: https://github.com/moltis-org/moltis ### Sign-ins Documented: The vault unseals automatically at login and reseals when the server restarts. While sealed, requests to protected endpoints get a 423 response. Evidence: https://docs.moltis.org/security.html ### Starting further runs Documented: Further runs start manually via the web UI, API or channels such as Telegram, Slack or Discord. A built-in cron scheduler triggers timed tasks. Evidence: https://github.com/moltis-org/moltis ### Human intervention Documented with conditions: Depending on approval_mode, commands require confirmation. You can list pending requests with /approvals and decide via Telegram or WhatsApp with /approve or /deny. Evidence: https://docs.moltis.org/security.html ### Detecting errors Documented with conditions: The README names Prometheus metrics, OTLP export and Langfuse traces for monitoring. The security page mentions a network audit log that records allowed and denied requests. Evidence: https://github.com/moltis-org/moltis Evidence: https://docs.moltis.org/security.html ### Resuming after failures Documented with conditions: Moltis takes automatic checkpoints before changes to built-in skills and memory, with restore tooling and session branching. Evidence: https://github.com/moltis-org/moltis ### Editorial assessment You run and update the server yourself, but approval mode and monitoring are configurable. Beyond skills and memory, you need to define your own procedure for failures and backups. ## Memory and conversation continuity Scope: Built-in memory; optional alternative backends Sources accessed: 2026-11-04 Markdown and hybrid indexes connect fact maintenance, retrieval and configurable handoffs. ### Storage and scope Documented: Markdown under ~/.moltis/MEMORY.md and memory/ is combined by default with SQLite, full-text and vector indexes; backends are replaceable. - [Moltis memory](https://docs.moltis.org/memory.html) ### Selection and timing Documented: Write modes control automatic and explicit saving; a silent memory flush can occur before compaction. - [Moltis memory](https://docs.moltis.org/memory.html) ### Inspect, edit and delete Documented: Files and settings are accessible; save, forget and delete tools and reindexing after changes are described. - [Moltis memory](https://docs.moltis.org/memory.html) ### Agent and channel boundaries Documented: Agents have their own workspaces; profile and conversation archives are distinguishable storage surfaces. - [Moltis memory](https://docs.moltis.org/memory.html) - [Moltis repository](https://github.com/moltis-org/moltis) ### Retrieval Documented: The default mode combines prompt memory with hybrid retrieval; alternative modes and search backends are configurable. - [Moltis memory](https://docs.moltis.org/memory.html) ### Session reset and compaction Documented: New sessions and resets can export sanitized transcripts; selected knowledge can be saved persistently before context compaction. - [Moltis memory](https://docs.moltis.org/memory.html) ## Context Configure the model, channels and permissions yourself. Provider sources reviewed; examples are editorial. No hands-on tests. ## Open points Requirements and limits, not yet confirmed: - German in everyday use - Reliability ## Sources - https://github.com/moltis-org/moltis ## Provider Publishes Moltis and the linked product information. - Operator: Moltis project - Organization: Open-source project - Website: https://github.com/moltis-org/moltis - Reviewed: 2026-10-16 - Provider source: https://github.com/moltis-org/moltis [Sources and method](https://ki-agentenkompass.de/en/sources/)