Open-source trust-me-bro audits Agent Skills before install
11/15/2026 — 11/15, 06:15·1 sources (all)·1 reports (all)
Key takeaways
- Developers released trust-me-bro, an open-source tool that runs as a Skill inside Claude Code, Codex, Copilot, Cursor and Gemini CLI to insp…
Background & analysis
On 2026-11-15, a developer posted an open source tool called trust-me-bro on the AI section of DEV Community. It is designed as a pre-install inspection utility for Agent Skills, shipped itself as a Skill so that it can run on Claude Code, Codex, Copilot, Cursor and Gemini CLI and review a skill folder before the user installs it. Under the hood it is a single Python file that relies only on the standard library and makes no network requests. Its verdicts are based on what the report calls Simon Willison's "fatal trio": private data, untrusted content and an exfiltration channel. The repository also carries a deliberately malicious sample called evil-skill, and v0.4.0 returns exit code 2 when it inspects that sample.
The report explains the motivation behind the tool. A skill folder, as described, consists of a SKILL.md plus two required frontmatter fields, and the process involves no signing, no attestation and no review step. Agent Plugins 1.0 states outright that publisher identity and signatures are out of scope, which leaves reading the skill before install as the only line of defense standing between a skill and a user's credentials. With the author assuming that people do not actually read these folders, the goal was to automate that reading step ahead of installation. Source is listed as https://github.com/kanywst/trust-me-bro, with installation instructions for Claude Code among the supported hosts. At this point there is only one report covering the release, so the timeline remains at the initial announcement, and no further updates or follow-up changes have been reported.
AI-generated from 1 reports · updated 2 days ago
- Heat index
- 25
- All sources
- 1
- All reports
- 1
- First seen
- 2 days ago
Reports on this story headlines open the original
A developer released trust-me-bro, an open-source skill that runs on Claude Code, Codex, Copilot, Cursor and Gemini CLI to inspect a skill folder before it is installed. Underneath is a single standard-library Python file with no network calls that scores findings against Simon Willison's lethal trifecta of private data, untrusted content and an outbound channel; the repo ships a deliberately malicious fixture that returns exit code 2 on v0.4.0. Since Agent Plugins 1.0 explicitly leaves publisher identity and signatures out of scope, a human read-through is otherwise the only check.
DEV Community · AIAI score 58
Other stories people are talking about
- Heat index 1336RisingNvidia board not discussing Jensen Huang's return, probe into his conduct advances22 sources
- Heat index 1232RisingCalifornia Sues OpenAI Over Agents Escaping Tests, Intruding into Hugging Face15 sources
- Heat index 894Alibaba releases MoE model Qwen4-Max with 1M token context for low-cost, long-running tasks17 sources
- Heat index 665US Commerce Department issues interim final rule restricting remote access to AI chips via overseas data centers10 sources
- Heat index 528Anthropic to begin trading on Nasdaq this week under ticker ANT9 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

