Json Repair Kit

PassAudited by VirusTotal on May 12, 2026.

Overview

Type: OpenClaw Skill Name: json-repair-kit Version: 1.0.0 The skill is classified as suspicious due to a critical Remote Code Execution (RCE) vulnerability in `index.js`. The `repairJson` function uses `vm.createContext` to evaluate user-provided JSON content, which is susceptible to known sandbox escape techniques. A malicious JSON file can be crafted to execute arbitrary Node.js code on the agent's system, despite the `SKILL.md` claiming the `vm` usage ensures 'no access to global scope or process.' This flaw allows an attacker to compromise the agent by providing a specially crafted input file.

Findings (0)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

ConcernHigh Confidence
ASI05: Unexpected Code Execution
What this means

A crafted or untrusted .json file could run JavaScript expressions inside the VM, hang the repair process, consume resources, or potentially exploit VM sandbox weaknesses.

Why it was flagged

The tool turns the target file contents into JavaScript code and executes it in a VM context. This is broader than repairing JSON syntax and lacks a timeout or other execution limits.

Skill content
const script = new vm.Script('result = (' + content + ')'); ... script.runInContext(context);
Recommendation

Use this only on files you trust. Prefer a non-executing JSON5/loose-JSON parser for repair, or add strict input limits, a timeout, and clear warnings against untrusted files.

What this means

Users may incorrectly believe it is safe to run this on arbitrary downloaded or repository files.

Why it was flagged

The safety wording is stronger than the implementation supports. The code still executes raw file contents as JavaScript, and Node VM execution should not be treated as a full security boundary for untrusted code.

Skill content
**Eval Sandbox**: Uses `vm.runInNewContext` to parse, ensuring no access to global scope or process. It is safer than `eval()`.
Recommendation

Revise the documentation to say the VM approach is not safe for untrusted input, and clearly recommend trusted files only.

What this means

Running recursive repair could rewrite many invalid JSON files and create backup files throughout the selected directory.

Why it was flagged

The skill documents recursive repair of all .json files under a directory. This is purpose-aligned and user-directed, but it can cause broad local file mutations if used on a large project tree.

Skill content
node skills/json-repair-kit/index.js --dir config/ --recursive
Recommendation

Run on a small, reviewed directory first, keep backups enabled, and review diffs before committing or relying on repaired files.

What this means

It may be harder to verify maintainership, update history, or report issues.

Why it was flagged

The skill has no declared source repository or homepage. The provided code is small and has no external dependencies, so this is a provenance note rather than a direct unsafe behavior.

Skill content
Source: unknown; Homepage: none
Recommendation

Review the included code before use and prefer a package with a clear source repository if using it in sensitive environments.