The supply-chain guardrail for the agentic coding era
Check every package before your AI agent installs it.
Audito plugs into Claude Code, Cursor, and any MCP client — flagging typosquats, deprecations, and known CVEs before they touch your lockfile, then fixing what's already there without leaving your workflow.
Works with
Claude Code·
Cursor·
Claude Desktop·
any MCP client
Monitors
·OSV.dev
·npm
·PyPI
·crates.io
·Go
Capabilities
Built for the way you actually ship now.
The agent picks the dependency. Audito makes sure the pick isn't a footgun — at install time, in CI, and forever after.
Headline
Pre-install guard
Your agent asks Audito before it writes a new dep. Typosquats, deprecations, known CVEs, and license mismatches get an allow / warn / block verdict with a plain-English summary — across npm, PyPI, crates.io, and Go.
Claude: I'll add colorss for ANSI colors.
AuditoBlocked
Likely typosquat of colors. Registered 6 days ago, 0 weekly downloads, no repo. Try chalk instead.
Whole-lockfile audit
Audit an entire lockfile in one call — npm, pnpm, yarn (classic + berry), requirements.txt, poetry, Pipfile, Cargo, and go.sum. Eight formats, one tool.
Headline
Agent-native remediation
Audito computes the nearest safe upgrade per CVE — a minimal, operator-preserving manifest edit and the exact per-ecosystem install command, with breaking changes flagged. Your agent applies it; you review and merge.
Bumplodash
^4.17.20 → ^4.17.21
Bumpnext
15.2.1 → 16.2.4
Bumpaxios
^1.6.0 → ^1.7.4
CI gate on every PR
A GitHub Action and an installable App post an Audito Check Run with inline per-package annotations. Malware and CRITICAL/HIGH block; the App can open remediation PRs.
Daily scan + real-time alerts
A daily OSV.dev scan keeps your inventory current; CRITICAL and HIGH findings email you immediately — not at the end of the week.
23open findings
−8 · 7d
1 Critical3 High7 Medium12 Low
Policy, posture & digest
License allow / warn / block on SPDX IDs, 30-day posture trends, a findings inbox with accept / dismiss / snooze triage, and a Monday digest of what changed.
Three steps. The dashboard is the report; your agent is the workspace.
01
Connect your agent
One line in Claude Code. Cursor and Claude Desktop work too over mcp-remote. Authenticate with an audt_… token — no SDK, no CI step required.
claude mcp add audito --remote …
02
It checks every dep
Pre-install, the agent asks Audito before writing the package. Post-install, a daily scan keeps watching what's already there and alerts you when something new lands.
03
It fixes them too
Audito returns the nearest safe upgrade; your agent applies it, or the GitHub App opens a draft remediation PR. You review and merge.
Works with Claude Code & Cursor
npm · PyPI · crates.io · Go
Powered by OSV.dev
No CI plumbing required
Free to start. $5 / seat when you grow.
Free covers 3 libraries on npm + PyPI with real-time alerts. Pro unlocks lockfile audits, the GitHub CI gate, all four ecosystems, and license policies — 14-day free trial, no card, starting at signup.