Tsz

SuspiciousAudited by ClawScan on May 10, 2026.

Overview

The declared ClawdHub CLI skill is bundled with unrelated FreeRide/OpenRouter code that can read credentials, rewrite OpenClaw configuration, and run a background model-rotation watcher.

Treat this package as a review-needed install. The advertised ClawdHub CLI functionality is plausible, but the supplied artifact set is incoherent and includes unrelated FreeRide code with credential access, persistent state, and global OpenClaw configuration changes. Install only a cleaned, clearly scoped version from a trusted source, and approve any skill install/update/publish commands manually.

Findings (5)

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

A user installing what appears to be a ClawdHub helper may receive unrelated code and hooks, making it hard to know what behavior they are actually approving.

Why it was flagged

The manifest contains multiple SKILL.md/_meta.json entries and files from apparently different projects, including FreeRide and proposal-generation assets, while the reviewed SKILL.md presents only a ClawdHub CLI helper.

Skill content
SKILL.md ... main.py ... watcher.py ... skill.json ... SKILL.md ... _meta.json ... assets/proposal-template.html ... scripts/activator.sh ... hooks/openclaw/handler.js
Recommendation

Do not install until the package is reduced to the intended ClawdHub files, duplicate/unrelated metadata is removed, and provenance is clarified.

What this means

The skill bundle includes code that can access an unrelated provider credential from local agent configuration.

Why it was flagged

The included FreeRide code reads an OpenRouter API key from the environment or from ~/.openclaw/openclaw.json, but the primary ClawdHub SKILL.md and registry requirements do not declare this credential use.

Skill content
api_key = os.environ.get("OPENROUTER_API_KEY") ... config = json.loads(OPENCLAW_CONFIG_PATH.read_text()) ... api_key = config.get("env", {}).get("OPENROUTER_API_KEY")
Recommendation

Only use a version that clearly declares OpenRouter credential access, scopes exactly how it is used, and matches the advertised skill purpose.

ConcernHigh Confidence
ASI08: Cascading Failures
What this means

A configuration change could alter which model the agent uses across sessions or tasks, not just during skill installation.

Why it was flagged

The included code can rewrite OpenClaw's default model configuration, which affects future agent behavior beyond a single ClawdHub CLI task.

Skill content
OPENCLAW_CONFIG_PATH = Path.home() / ".openclaw" / "openclaw.json" ... config["agents"]["defaults"]["model"]["primary"] = formatted_primary ... OPENCLAW_CONFIG_PATH.write_text(json.dumps(config, indent=2))
Recommendation

Require explicit user approval and backup/reversal instructions before any skill modifies global OpenClaw configuration.

ConcernHigh Confidence
ASI10: Rogue Agents
What this means

If enabled, this code could keep acting after the original task by making periodic API calls and changing model settings.

Why it was flagged

The bundle includes a long-running watcher that can operate repeatedly in the background and rotate models, which is not disclosed by the ClawdHub CLI SKILL.md.

Skill content
FreeRide Watcher\nMonitors for rate limits and automatically rotates models.\nCan run as a daemon ... CHECK_INTERVAL_SECONDS = 60 ... while running:
Recommendation

Avoid installing unless the background watcher is removed or clearly documented as opt-in with stop, uninstall, and state-cleanup instructions.

What this means

These commands can change installed agent skills or publish content to a registry.

Why it was flagged

The stated ClawdHub purpose includes installing, updating, and publishing skills, including a force/no-input update example. This is disclosed and purpose-aligned, but it is high-impact behavior that should remain user-approved.

Skill content
clawdhub install my-skill ... clawdhub update --all --no-input --force ... clawdhub publish ./my-skill --slug my-skill
Recommendation

Review each install, update, force-update, and publish action before allowing the agent to run it.