GoKawiilNew open-source tool SIwithin sandboxes Claude Code and Codex CLIs with spend caps
A developer released SIwithin, a single-file Python tool that runs AI coding CLIs like Claude Code and OpenAI Codex unattended, restricting them to specified folders and enforcing a dollar budget per run. It uses a loopback proxy to meter Claude API requests with a per-run token instead of the user's real key, and logs one record per run showing cost and completion status. The tool currently works only on macOS and refuses to run on Linux or Windows.
GoKawiil's interpretation of the reporting above, not reported fact.
As developers increasingly let AI agents operate autonomously on their machines, tools like this address real concerns about runaway costs and unrestricted file access. By refusing to operate without its safety fence, the project signals an emphasis on verifiable guarantees over trust-based assurances, though its single-file, macOS-only design may limit adoption until broader platform support or community review establishes confidence.
- SIwithin restricts Claude Code and Codex CLI runs to specified folders and a set dollar budget
- A loopback proxy meters API usage with per-run tokens rather than exposing the user's actual API key
- The tool is macOS-only, open source, has no telemetry, and is not affiliated with Anthropic or OpenAI
Source: github.com, 2026-10-20
Published there as: “Show HN: Run Claude Code unattended inside a fence, under a dollar budget”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.