registry / AI-agent capability abuse / osborn / 0.9.173

osborn@0.9.173

Voice AI coding assistant - local agent that connects to Osborn frontend

AI Security Review

scanned 2mo ago · by lpm-firewall-ai

This is a risky voice-agent runtime with global Claude mutations and a remote shell installer, but inspection found no install-time execution or confirmed credential exfiltration. The capabilities merit a warning rather than a malware block.

Static reason

High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source

Trigger

The user runs the CLI, completes a Claude compaction, or sends the start_ide runtime request.

Impact

Changes can persist in the user's Claude control surface; the installer executes remote script content.

Mechanism

Global Claude configuration writes and a request-triggered curl-to-shell installer.

AI rationale

Evidence

dist/index.jsdist/claude-llm.js~/.claude/settings.json~/.claude/skills/*/SKILL.md

Network endpoints1

code-server.dev/install.sh

Decision evidence

public snapshot

AI called this Suspicious at 90.0% confidence as Dangerous Capability with low false-positive risk.

Evidence for warning

The manifest exposes an executable CLI, while registry installation has only a prepublish build hook.
When the IDE feature is requested and code-server is absent, runtime pipes downloaded code into sh.
Importing the Claude LLM module changes the user's ~/.claude/settings.json.
A post-compaction hook writes model-extracted content into ~/.claude/skills/*.
The code-server download is reachable through the start_ide runtime request.

Evidence against

No preinstall, install, or postinstall hook is present.
No source evidence showed credential harvesting or secret exfiltration.
The remote installer and skill operations are runtime features, not automatic install-time behavior.

Behavioral surface

Source

ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetworkShell

Supply chain

HighEntropyStringsUrlStrings

Source & flagged code

9 flagged

bin/cli.jsView file

Lines 1-22javascript

1#!/usr/bin/env node

3import { spawn } from 'child_process'

4import { fileURLToPath } from 'url'

High

Child Process

Package source references child process execution.

bin/cli.jsView on unpkg · L2

5import { dirname, join } from 'path'

6import { readFileSync, existsSync, lstatSync, symlinkSync } from 'fs'

7import os from 'os'

9// On Fly machines, home is /root (ephemeral) but sessions live on /workspace/.claude.

10// Recreate the symlink on every startup so sessions survive image updates.

11try {

12 const home = os.homedir()

13 const target = '/workspace/.claude'

14 const link = join(home, '.claude')

15 if (existsSync(target) && home !== '/workspace') {

16 let needsLink = true

17 try {

18 const stat = lstatSync(link)

19 needsLink = !stat.isSymbolicLink()

20 } catch {}

21 if (needsLink) symlinkSync(target, link)

22 }

dist/jsonl-search.jsView file

Lines 14-54javascript

14import { join } from 'path';

15import { readSessionHistory } from './session-access.js';

16// ============================================================

17// BM25 CACHE

18// ============================================================

19let bm25Cache = null

dist/index.jsView file

Lines 1-31javascript

1// Load environment variables FIRST before any other imports

2import 'dotenv/config';

3import { voice, initializeLogger } from '@livekit/agents';

4import { Room, RoomEvent, } from '@livekit/rtc-node';

5import { AccessToken } from 'livekit-server-sdk'

deploy.shView file

•path = deploy.sh kind = build_helper sizeBytes = 2309 magicHex = [redacted]

Medium

Ships Build Helper

Package ships non-JavaScript build or shell helper files.

deploy.shView on unpkg

dist/session-access.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 1

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/session-access.js#virtual:normalized:round1View on unpkg

dist/claude-llm.jsView file

•matchType = previous_version_dangerous_delta matchedPackage = osborn@0.9.172 matchedIdentity = npm:b3Nib3Ju:0.9.172 similarity = 0.964 summary = stored previous version shares package body but lacks this dangerous source file

Critical

Previous Version Dangerous Delta

This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.

dist/claude-llm.jsView on unpkg

Source & flagged code

9 flagged · loading source

bin/cli.jsView file

2L3: import { spawn } from 'child_process' L4: import { fileURLToPath } from 'url'

High

Child Process

Package source references child process execution.

bin/cli.jsView on unpkg · L2

dist/jsonl-search.jsView file

34// @vscode/ripgrep bundles the rg binary — this is our own dependency L35: const { rgPath } = require('@vscode/ripgrep'); L36: if (rgPath && existsSync(rgPath))

Medium

Dynamic Require

Package source references dynamic require/import behavior.

dist/jsonl-search.jsView on unpkg · L34

dist/index.jsView file

2459const which = await new Promise((res) => { L2460: const p = spawn('which', ['code-server']); L2461: let out = ''; ... L2468: await new Promise((res, rej) => { L2469: const install = spawn('sh', ['-c', 'curl -fsSL https://code-server.dev/install.sh | sh'], { L2470: env: { ...process.env }, L2471: stdio: ['ignore', 'pipe', 'pipe'],

Findings

5 Critical3 High5 Medium5 Low

CriticalSame File Env Network Executiondist/index.js

CriticalCredential Exfiltrationdist/index.js

CriticalCommand Output Exfiltrationdist/index.js

CriticalDownload Executedist/index.js

CriticalPrevious Version Dangerous Deltadist/claude-llm.js

HighChild Processbin/cli.js

HighShell

HighSemantic Analysis Limiteddist/session-access.js#virtual:normalized:round1

MediumDynamic Requiredist/jsonl-search.js

MediumNetwork

MediumEnvironment Vars

MediumShips Build Helperdeploy.sh

MediumStructural Risk Force Deep Review

Affected versions and remediation

This report applies to osborn@0.9.173.

See all versions of osborn for other recorded verdicts.

Review the evidence and your use of osborn@0.9.173 before allowing it. Restrict the permissions described in this report.
Choose an independently verified alternative or release. This report does not establish that other versions are safe.

Evidence last updated: 2026-09-03 23:33:56.275Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Runtime surface

Behavioral surface

Source & flagged code