@oked/claude-code@0.6.0
OKed for Claude Code - zero-code human approval for sensitive actions
AI Security Review
scanned 28d ago · by lpm-firewall-aiThe package explicitly configures a Claude Code tool hook when its `init` command is run, which is a warning-level agent capability under the review policy. Its manifest has no automatic npm install hook, and the inspected source does not show a concrete malicious chain.
Static reason
One or more suspicious static signals were detected.
Trigger
A user runs `oked init`.
Impact
Claude Code tool actions can be routed through the configured approval hook. No confirmed credential theft or other attack was established.
Mechanism
The command adds or updates the OKed hook in the project's Claude settings; the hook can consult the configured OKed service while handling tool events.
AI rationale
The package explicitly configures a Claude Code tool hook when its `init` command is run, which is a warning-level agent capability under the review policy. Its manifest has no automatic npm install hook, and the inspected source does not show a concrete malicious chain.
Evidence
.claude/settings.json
Network endpoints1
api.oked.ai
Decision evidence
public snapshotAI called this Suspicious at 87.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The manifest declares only a prepublish hook, so npm installation does not automatically configure Claude Code.
- The CLI resolves project settings to `.claude/settings.json` and writes a command hook into `PreToolUse` when `init` runs.
- The installed hook is `oked-hook`, matched to Bash, Write, Edit, Agent, and MCP tools.
- The `init` command pairs the device with `api.oked.ai` and saves the returned API key to the OKed user config.
Evidence against
- Hook setup is triggered by the explicit `init` command rather than an npm lifecycle hook.
- Pairing uses the package-aligned `api.oked.ai` service; the inspected source did not establish credential forwarding to an unrelated recipient.
Behavioral surface
ChildProcessEnvironmentVarsFilesystemNetwork
UrlStrings
Source & flagged code
6 flaggeddist/cli.jsView file
4import { join, dirname } from "path";
L5: import { spawn } from "child_process";
L6: import { createInterface } from "readline";
High
4import { join, dirname } from "path";
L5: import { spawn } from "child_process";
L6: import { createInterface } from "readline";
...
L10: import { OKED_RECEIPTS_LOG, verifyReceiptsFile } from "./receipts.js";
L11: const DEFAULT_BACKEND_URL = process.env.OKED_BACKEND_URL || "https://api.oked.ai";
L12: function getSettingsPath() {
High
Same File Env Network Execution
A single source file combines environment access, network access, and code or shell execution; review context before blocking.
dist/cli.jsView on unpkg · L44import { join, dirname } from "path";
L5: import { spawn } from "child_process";
L6: import { createInterface } from "readline";
...
L10: import { OKED_RECEIPTS_LOG, verifyReceiptsFile } from "./receipts.js";
L11: const DEFAULT_BACKEND_URL = process.env.OKED_BACKEND_URL || "https://api.oked.ai";
L12: function getSettingsPath() {
L13: return join(process.cwd(), ".claude", "settings.json");
L14: }
...
L18: try {
L19: return JSON.parse(readFileSync(path, "utf-8"));
L20: }
...
L45: function openBrowser(url) {
Findings
3 High5 Medium4 Low
HighChild Processdist/cli.js
HighSame File Env Network Executiondist/cli.js
HighSandbox Evasion Gated Capabilitydist/cli.js
MediumNetwork
MediumEnvironment Vars
MediumAi Review Evidencepackage.json
MediumAi Review Evidencedist/cli.js
MediumAi Review Evidencedist/cli.js
LowNon Install Lifecycle Scripts
LowScripts Present
LowFilesystem
LowUrl Strings
Affected versions and remediation
This report applies to @oked/claude-code@0.6.0.
See all versions of @oked/claude-code for other recorded verdicts.
- Review the evidence and your use of @oked/claude-code@0.6.0 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.