I'm not a fan of Deno's security model (2024)
Web page · https://goldstein.lol/posts/deno-security
First seen by Oriel on .
AI summary
The author argues Deno’s permissions system is not a true sandbox and creates a false sense of security when executing untrusted JavaScript, especially through terminal prompts and --allow-run.
- Permission prompts share attacker-controlled terminal channels, allowing malicious code to manipulate display or stdin and potentially conceal or pre-approve requests.
- --allow-run=uname can execute a different binary through a child-controlled PATH, undermining command-name restrictions.
- LD_PRELOAD enabled arbitrary code execution through permitted commands; Deno patched LD_ variables, but the author says configurable programs remain an escape route.
- The author recommends --no-prompt, avoiding --allow-run and --allow-env, handling symlinks carefully, and separately controlling CPU and memory consumption.
- For untrusted code, the author recommends a proper operating-system sandbox such as bwrap instead of relying on Deno’s standard-library call filtering.
#deno #sandboxing #runtime security #javascript
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 102, latest observation ; +92 over observed period since
- Lobsters: 44, latest observation ; +42 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 56, latest observation ; +56 over observed period since
- Lobsters: 17, latest observation ; +17 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hacker News | 50413769 | https://goldstein.lol/posts/deno-security/ Read discussion |
| Lobsters | sh62d1 | https://goldstein.lol/posts/deno-security/ Read discussion |