osborn@0.9.228
Voice AI coding assistant - local agent that connects to Osborn frontend
AI Security Review
scanned 26d ago · by lpm-firewall-aiThe package has a concrete, user-invoked capability to mutate Claude Code configuration and skills. This warrants a warning, but the inspected source does not support a malicious install-hook or credential-theft conclusion.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
A user starts an Osborn command or service that loads the Claude LLM module; skill creation requires a skill_add service message.
Impact
This is an explicit-use agent-control capability that can alter Claude behavior and install skills, but no unconsented npm lifecycle trigger or confirmed exfiltration was found.
Mechanism
It overwrites selected values in ~/.claude/settings.json and writes supplied skill text under ~/.claude/skills.
AI rationale
The package has a concrete, user-invoked capability to mutate Claude Code configuration and skills. This warrants a warning, but the inspected source does not support a malicious install-hook or credential-theft conclusion.
Evidence
dist/claude-llm.jsdist/index.js~/.claude/settings.json~/.claude/skills/<skillName>/SKILL.md
Decision evidence
public snapshotAI called this Suspicious at 86.0% confidence as Dangerous Capability with low false-positive risk.
Evidence for warning
- package.json defines command-line binaries and has no install lifecycle hook.
- The Claude LLM module changes ~/.claude/settings.json values and invokes that routine at module load.
- The running service accepts a skill_add message and writes its supplied content into ~/.claude/skills.
- The package writes an OAuth token only into the local Claude configuration directory with restrictive file permissions.
Evidence against
- No source evidence showed install-time execution, remote payload loading, or credential transmission to an unrelated endpoint.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsUrlStrings
Source & flagged code
9 flagged · loading sourcebin/recall.jsView file
5L6: import { spawn } from 'child_process'
L7: import { fileURLToPath } from 'url'
High
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 · L34dist/index.jsView file
2531const which = await new Promise((res) => {
L2532: const p = spawn('which', ['code-server']);
L2533: let out = '';
...
L2540: await new Promise((res, rej) => {
L2541: const install = spawn('sh', ['-c', 'curl -fsSL https://code-server.dev/install.sh | sh'], {
L2542: env: { ...process.env },
L2543: 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/session-store.js
HighChild Processbin/recall.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@0.9.228.
See all versions of osborn for other recorded verdicts.
- Review the evidence and your use of osborn@0.9.228 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: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.