U Ministry Utility
25“No ministry fit: README describes 'OpenShell is the safe, private runtime for fleets of autonomous AI agents' with no ministry lane present…”
source unavailable
NVIDIA/OpenShell
OpenShell is the safe, private runtime for autonomous AI agents.
Trust Score
44.2
Last full review 2026-09-29 · Trust Score decays −2 points per month without a fresh review (never below 20)
U Ministry Utility
25“No ministry fit: README describes 'OpenShell is the safe, private runtime for fleets of autonomous AI agents' with no ministry lane present…”
source unavailable
M Maturity
50“api.releases: 'latest tag v0.1.2 published 2026-09-28T03:58:00Z'; latest_commit dated 2026-10-02 ('fix(sandbox): restrict provider file…”
source unavailable
S Security / License
46.2 → 46.2 (+0.0 over window)
ranked #4 of 27 in G2
NVIDIA/OpenShell looks like a candidate for ai, agents & mcp servers; No ministry lane; scored as a general AI-agent runtime tool.
Use it with
Selected: Gloo AI Studio
Tool names and logos belong to their owners; shown so you can pick your own client. Not affiliated.
Step 1 — Connect Refinry once (prompt only)
Gloo AI Studio does not document external MCP server setup. Use this prompt with the copied Refinry research brief in Studio.
https://studio.ai.gloo.com/I am choosing open-source software for my ministry using Refinry, a review site. Read Refinry’s plain-language agent feed — https://refinry.dev/docs.md — then help me evaluate NVIDIA/OpenShell: summarize the Trust Score, the strongest and weakest dimensions, and any caution flags it cites. Give me one small first step I can complete this week, and tell me honestly if this looks like more than I can handle.
Some clients need manual setup. Follow the plain-language steps on the Docs page.
Step 2 — Plan your first step
Tell the AI what you want to do; edit this to match your situation.
I am choosing open-source software for my ministry using Refinry, a review site. Read Refinry’s public review of NVIDIA/OpenShell (https://refinry.dev) and help me assess it for using NVIDIA/OpenShell for ai, agents & mcp servers in my ministry. Summarize the Trust Score, the strongest and weakest dimensions, and any caution flags it cites, separating reviewed evidence from what still needs verification. Give me one small first step I can complete this week, and tell me honestly if this looks like more than I can handle.
Prefer a hand-off that works without a connection? Copy the research brief: every reviewed fact, all caution flags, and a feasibility-only first assignment.
license 75 · security 50
“api.repos.license.spdx_id: 'Apache-2.0' (permissive). README badge: 'Security-Report a Vulnerability' (SECURITY.md exists). But security…”
source unavailable
A Ease of Adoption
50dev 50 · volunteer 50
“README Quickstart: 'curl -LsSf | sh' then 'openshell sandbox create --name demo' — one-command install; hosted docs at docs.nvidia.com…”
source
C Community Trust
50“api.repos: 'stargazers_count': 14210, 'forks_count': 1642 — stars are at most one weak input; no external contributor, issue-reporter, or…”
source unavailable
I AI-Agent Integrability
75“README has install/use sections ('Quickstart', 'Run Your First Agent' link), License badge (SPDX 'Apache-2.0'), PyPI badge…”
source unavailable