Sandbox to create a local isolation for agentic run. "Eye of Sauron" for agents, observe everything from file access to network calls with rayTrace sandbox
How fast Godcreatebugs/raytrace-cli is gaining stars, and where it lands on Trendshift and GitHub Trending
Activity and trending
Stars, forks and merged pull requests each day or month, lined up with the repository's rank on Trendshift Trending
Nov 9 – Nov 22, 2026
Stars +4in 14 days
012
2
Forks +10in 14 days
024
4
Merged PRs 39in 14 days
0510
10
Trendshift Trendingwhere the repository ranked each day it made the top 25
Not in the Trendshift top 25 in this period
Nov 9Nov 11Nov 13Nov 15Nov 17Nov 19Nov 21
The left and right arrow keys step from day to day; Home and End go to the first and the last
Stars, forks, merged pull requests and Trendshift Trending rank, by day
Day
Stars gained
Forks gained
Merged PRs
Mon, Nov 9, 2026
+0
+0
1
Tue, Nov 10, 2026
+0
+0
0
Wed, Nov 11, 2026
+0
+0
0
Thu, Nov 12, 2026
+0
+1
5
Fri, Nov 13, 2026
+0
+0
0
Sat, Nov 14, 2026
+0
+0
0
Sun, Nov 15, 2026
+1
+1
1
Mon, Nov 16, 2026
+0
+0
7
Tue, Nov 17, 2026
+2
+0
10
Wed, Nov 18, 2026
+0
+3
7
Thu, Nov 19, 2026
+0
+4
6
Fri, Nov 20, 2026
+0
+1
2
Sat, Nov 21, 2026
+0
+0
0
Sun, Nov 22, 2026
+1
+0
0
Total
+4
+10
39
1#1 on the list
7Rank that day
Not trending
Social mentions
Recent discussions about this repository across the web
3 posts
When an agent run goes wrong at step 14, you shouldn't have to rerun steps 1-13 to test a fix. RayTrace lets you fork a Claude Code / Codex session at any step and re-run from there (I help the…
A search worth running after every agent session: status:failed reported:success Runs where the agent said it worked and the sandbox saw a failure. RayTrace, free + MIT (I help the founder):
Claude Code's new Code Mode shows "Read 8 files, ran 16 shell commands" by default. Great for tokens. Less great when you need to know which 16. RayTrace logs each command, exit code and file from…
Debug agents like code: fork a Claude Code session at any step, change one instruction ("never read .env"), re-run in a fresh sandbox, compare. Open source:
My favourite search in RayTrace: status:failed reported:success Every agent run where it claimed success and the evidence says otherwise. Free, open source:
Anthropic's Claude Code team wrote about making agents verify their work with evidence. Agreed. And the evidence shouldn't come from the agent itself. RayTrace records it from outside the sandbox:
Run Claude Code in an isolated sandbox with one command: npm i -g @raytrace-cli/cli raytrace sandbox create --local raytrace connect <name> --claude Then open a dashboard of every file, command and…
Your coding agent: "All tests passed." The sandbox: npm test exited 1. RayTrace runs Claude Code / Codex in a gVisor sandbox and shows what it said next to what it did. MIT, local-first:
Until Claude Code 2.1.325, a Read(./.env) deny rule silently stopped matching after the agent ran cd. Rules are a promise. I want receipts. RayTrace logs every file your agent actually opened,…
fuck it, we're open sourcing how we debug coding agents what the agent said vs what it actually ran, now on Github for anyone to: utilize it damage it enhance it show us some love (or roast it) …
Leverage full potential of coding agents without the risks. Run them in a sandbox with restricted networking, file access and secrets hidden. By Protectorate.
Run coding agents unattended. Agents write the code. Axocoatl writes the receipt. Claude Code, Codex or a local model in a sandbox; the host runs your checks, a different model reviews, and every run ends with a CI exit code and a signed receipt.
Run Claude Code, Codex and other coding agents in a local VM on a clone of your project, then review their changes like a pull request. Mac app, CLI for Linux and Windows.