What it does
5 features
ZeroClaw · Self-hosted personal agent
Product and access
ZeroClaw is listed as a separate product profile.
Core features
5 features from provider sources · checked on 16 October 2026. Availability and limits are listed below.
Usable AI models
Open choice
You can configure your own keys from many providers, OpenRouter or an OpenAI-compatible endpoint, including fallback chains. Locally, ZeroClaw runs with Ollama or a self-hosted inference server, for example.
Model families named
- Anthropic (Claude)
- OpenAI (GPT)
- Google (Gemini)
- xAI (Grok)
- Moonshot (Kimi)
- Alibaba (Qwen)
- MiniMax
- 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
The rating covers the generally available main access. Documented · Sources accessed on
ZeroClaw: Provider Catalog · ZeroClaw: Providers overview · ZeroClaw: README
Does it fit my task?
2 practical tasks · 3 requirements and limits
A good fit if …
- you want to run the runtime yourself on your own machine
- you want to reach one agent through channels such as Discord, Telegram, email or webhooks
- you want to set the model and permissions yourself in a configuration file
Less suitable if …
- you want a hosted solution without installation and maintenance
- you need organisation-level administrator approvals and roles; the reviewed pages mention none
Editorial assessment based on provider documentation. Not a product test.
Practical tasks
Editorial examples based on documented features. No product tests performed.
Try event-driven work
- Your input
- A test webhook and an approval-gated workflow.
- Possible result
- A traceable workflow with a controlled action.
- What to check
- Check content, permissions and actions taken.
Try local model access
- Your input
- A running Ollama service and a test task.
- Possible result
- A result through the selected model connection.
- What to check
- Record time spent, corrections and costs.
Try event-driven work.
Your possible prompt
Organize these shared documents by topic. Create an overview with deadlines and next steps. Prepare draft replies for my approval.
Adjust the task, data and permissions.
Requirements and limits
Can I use it?
With requirements
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:
With requirements
Self-hosting with own model access
ZeroClaw documents a self-operated runtime with Unix and Windows installation. Setup selects a provider, including Ollama and external APIs with their own access requirements.
Region, reason and source
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.
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 ZeroClaw runtime with your own model keys, channels and gateway
Sources reviewed:
Show all 3 points
Open in reviewed sources
Admin approval
The reviewed pages describe no approval process by organisation administrators. You set permissions yourself in the configuration.
EvidenceDocumented with conditions
Data access & team sharing
Permission profiles are deny-by-default, so anything not granted is refused. Browser and HTTP have domain allowlists that by default permit all public hosts, so you must narrow them yourself. According to the provider everything runs on your machine and you own the data.
Evidence · Evidence
Memory and conversation continuity
4 of 6 criteria documented
Explicit memory writes and configurable autosave are separate from prompts and files.
Current agent memory and separate conversation history
Sources accessed:
Documentation by criterion
StorageDocumented
Storage and scope
Persistent memory uses a configured backend store per agent; files and conversation logs form separate storage areas.
SelectionDocumented
Selection and timing
Durable entries arise through memory tools or explicitly configured runtime autosave; prompt content is not automatically stored thereby.
Edit & deletePartly documented
Inspect, edit and delete
Tools support saving and retrieval; complete correction and deletion workflows are not explained in these sources.
Access boundariesDocumented
Agent and channel boundaries
Memory is agent-scoped; additional peer access is explicit. Conversation memories also have session filters depending on access path.
RetrievalDocumented
Retrieval
At turn start, relevant entries can appear as bounded memory context; filters differ between access paths.
New conversationsPartly documented
Session reset and compaction
Long-term memory is separate from session history. History trimming changes context lossily but does not delete memory; reset details remain unspecified.
Cost and effort
Prices as of 16 October 2026
Prices and included usage
Pricing checked
Open source; model and operating costs are separate.
- Total cost depends on access, usage and configuration.
Setup
Install a release and configure the model and agent with Quickstart.
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: ZeroClaw as a self-hosted runtime on your own machine or server
Sources reviewed:
Editorial assessment
You run ZeroClaw yourself, so updates, backups, key management and monitoring are your job. The default approvals help, provided you deliberately avoid YOLO mode.
Show all 6 points
Documented
Files and working state
Gateway chat sessions are persisted in SQLite by default, and memory uses SQLite by default with retention settings. Configuration is a TOML file, and backups go to state/backups by default.
EvidenceDocumented
Sign-ins
Pairing codes are valid for 10 minutes by default, a dashboard session lasts at most 8 hours, and ACP sessions end after an hour without activity. These are default settings.
EvidenceDocumented
Human intervention
The default level, supervised, requires approval for medium-risk operations and blocks high-risk ones. YOLO mode skips these gates. SOPs have approval gates.
EvidenceDocumented with conditions
Detecting errors
Runtime traces go to a JSONL file with daily rotation. Observation via logs, Prometheus or OpenTelemetry depends on the configured backend. Alert channels for high escalations are possible but empty by default.
Evidence
Evidence and open questions
1 source · 2 open points
Open points
The reviewed sources contain no reliable information on these points. A gap says nothing about the product.
Requirements and limits: not yet confirmed
Sources and evidence
Provider sources reviewed; examples are editorial. No hands-on tests.
Editorial team: Michael Meierhoff · sources reviewed on . Provider information is not independent evidence of performance.
Provider
Publishes ZeroClaw and the linked product information.
- Operator
- ZeroClaw Labs
- Organization
- Open-source project
- Website
- github.com
Changes to this profile 1
Added ZeroClaw with capabilities, sources, pricing and open European availability questions.
These notes describe editorial changes. Source and pricing review dates appear in their respective sections.