Unified Web Search (Iyeque)

ReviewAudited by ClawScan on May 10, 2026.

Overview

The skill mostly matches its web/local search purpose, but it runs an undeclared neighboring Tavily script and gives it the process environment, so it needs review before use.

Review the separate tavily-search helper before using Tavily mode, and consider running the skill with a minimal environment that contains only TAVILY_API_KEY. Be aware that local searches can show filenames and paths from your OpenClaw workspace memory and skills directories.

Findings (3)

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.

What this means

Using the Tavily source may execute code from another local skill or file that was not part of this review.

Why it was flagged

The skill runs a sibling Tavily script that is not included in the provided manifest and is not declared by an install spec, creating an unreviewed dependency/provenance gap.

Skill content
const scriptPath = path.join(__dirname, '../tavily-search/scripts/search.mjs'); ... const output = execSync(cmd, {
Recommendation

Install only if you trust and have reviewed the separate tavily-search helper at that exact path, or update the skill to call Tavily directly or declare a reviewed dependency.

What this means

Other API keys or credentials present in the agent environment could become accessible to the helper process.

Why it was flagged

The child process receives the full environment, not only the required Tavily API key. Combined with the undeclared external helper, this can expose unrelated environment secrets to code outside this package.

Skill content
env: { ...process.env, TAVILY_API_KEY: apiKey }
Recommendation

Run this skill in a minimal environment containing only the variables it needs, and prefer passing only TAVILY_API_KEY to any child process.

What this means

Search results may reveal local workspace filenames or paths from memory/skill areas to the agent context.

Why it was flagged

The local search feature reads filenames from persistent OpenClaw workspace memory and skills directories and returns matching absolute paths.

Skill content
const allowedSubdirs = ['memory', 'skills']; ... fs.readdirSync(searchPath, { withFileTypes: true });
Recommendation

Use the local source only when you are comfortable exposing matching workspace filenames and paths in search output.