Tilde.run Releases AI Agent Sandbox: Rollbackable Transactional Version File System
Tilde.run launches an AI Agent sandbox environment that unifies GitHub, S3 and Google Drive into a single versioned file system and supports transaction rollback. Each Agent run is treated as an auditable transaction record, solving the security and recoverability issues of autonomous AI code running on real data.
Core conclusion
Tilde.run is a sandbox running environment for AI Agents. The core innovation lies in the "transactional version file system" - unified mounting of GitHub code, AWS S3 data and Google Drive documents into a single rollback file system view. Each Agent runs to form an auditable "transaction", and after an error occurs, it can be rolled back to any historical state with one click.
Key Points
- Release time: 2026-05-07 (Hacker News received 158 likes on the day of release and topped the homepage)
- Core capabilities: versioned file system + network call audit + manual approval gate control
- Target users: developers and teams who use AI Agent to process real data
- Open model: CLI and SDK are open source, and the hosting service provides free entry
Background: Industry pain points for safe operation of AI Agents
As AI Agent tools such as Claude Code, OpenAI Codex, and LangGraph enter the production environment, a core question has become increasingly prominent: How to ensure security and recoverability when autonomous code runs on real data? **
The current solutions common in the industry have obvious flaws:
- Pure Docker container isolation is not fine-grained enough to audit every network call
- Manual backup and recovery costs are high, and the Agent runs much more frequently than human jobs.
- Lack of unified version view for cross-data source operations (GitHub + S3 + Drive)
Tilde.run is precisely to solve this "Agent sandbox" problem - not simple container isolation, but turning every run into an auditable and rollable "transaction".
Core functions at a glance
| Features | Description | Value to Developers |
|---|---|---|
| Versioned file system | GitHub/S3/Drive is mounted as ~/sandbox | No need to manually synchronize multiple data sources |
| One-click rollback | Any running state can be restored to historical commit | Error recovery reduced from hours to seconds |
| Network auditing | Every outbound call is logged and inspected | Prevents agents from accidentally operating external systems |
| Manual gate control | Supports operation-by-operation approval strategy | Key operations can be set to "requires manual confirmation" |
| Isolated runs | Each run is independent of each other | Parallel experiments do not interfere with each other |
Technical Highlights: Transactional Version File System
The most unique innovation of Tilde.run is its file system layer. Unlike traditional Docker layers (which are built from scratch every time), Tilde's versioned file system has the following features:
1. Unified mounting of multiple sources
~/sandbox/
├── github/ → acme/ml-pipeline (847 objects, 12 GB)
├── s3/ → production-data
└── drive/ → team-wiki/output**2. **
Agent 。 Agent ， commit—— git revert 。
**3. ** API 、、。、 Agent 。
| Tilde.run | ✅ | ✅ | ✅ | ✅ | ✅ |
| Docker | ❌ | ❌ | ❌ | ❌ | ❌ |
| git + backup | ✅ | ❌ | ❌ | ❌ | ❌ |
| lakeFS | ✅ | ✅ | ❌ | ❌ | ❌ |
AI
waytoclawearn n8n / OpenClaw / Claude Code ，Tilde.run
- **** n8n ，
- **** GitHub、S3、Drive AI Agent
- ****，
- ** Agent ** AI Agent
AI Agent、Claude Code、LangGraph、n8n、OpenAI、GitHub and other tool names are automatically matched by the platform side to the maintained tool library without additional configuration.
Internal link guidance
- Want to learn how to use n8n to build AI automation processes? See: AI Agent-Driven Content Automation: n8n MCP Building Guide from Scratch
- Real case: Independent developer uses AI Agent to automate workflow and earn monthly income $5,000: Indie Developer: n8n + OpenClaw Automation Workflow Earning $5,000/mo
Topic hub
AI Agent Tutorials & Workflow Guides
Evergreen how-tos for coding agents, content pipelines, and n8n automation—linked to news context and real earn cases.
Explore AI Agent Tutorials & Workflow Guides →Monetization angle
How can you make money from this trend?
WayToClawEarn focuses on verified earn playbooks—not just news. Start from these cases.
DeepSeek + Claude Code Micro SaaS
Run multiple small products on cheap inference
Claude Code bug bounty
Productize agent skills into security services