Model Router

ReviewAudited by ClawScan on May 1, 2026.

Overview

The skill is coherent and purpose-aligned, but it stores AI provider API keys locally and routes tasks to spawned model sessions, so users should review those behaviors before use.

Before installing or running setup, decide which providers you trust, use limited or budget-controlled API keys when available, verify that ~/.model-router is private, and avoid routing sensitive prompts to models/providers you have not approved.

Findings (4)

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

Anyone using the setup wizard will place AI provider credentials on disk, which could affect billing or account access if the local account or file is compromised.

Why it was flagged

The setup wizard persistently stores provider API keys in a local file. This is expected for a model-router setup, and the code applies restrictive permissions, but API keys are high-value credentials.

Skill content
API_KEYS_FILE = CONFIG_DIR / ".api-keys" ... keys[env_key] = key ... os.chmod(API_KEYS_FILE, 0o600)
Recommendation

Only enter provider keys you intend this skill to use, prefer limited-scope or budget-limited keys where possible, verify ~/.model-router permissions, and rotate keys if you remove or stop trusting the skill.

What this means

Tasks may be processed by different model providers or spawned sessions, which can have different cost, privacy, and retention implications.

Why it was flagged

The intended workflow sends task text to selected model sessions/providers. This is core to the skill, but it means sensitive prompts may leave the current session boundary.

Skill content
Support for 6 major AI providers ... sessions_spawn --task "Debug this memory leak" --model claude-sonnet
Recommendation

Use explicit model choices for sensitive work, avoid routing secrets or regulated data unless the selected provider is approved, and review provider privacy/billing settings.

What this means

Running the wizard executes local code that creates files and stores configuration/API keys.

Why it was flagged

The quick start asks the user to run a local Python setup script. This is disclosed and central to configuration, not hidden execution.

Skill content
python3 scripts/setup-wizard.py
Recommendation

Run the script only from the installed skill directory you intended to use, and inspect the included script first if you are concerned about local credential handling.

What this means

A user could overestimate the protection of the local .api-keys file if their disk or home directory is not actually encrypted.

Why it was flagged

The security section frames stored API keys as encrypted via the operating system. Users should understand this is not application-level encryption by the skill and depends on the user's OS/storage configuration.

Skill content
Encrypted at rest (via OS filesystem encryption)
Recommendation

Treat the key file as sensitive plaintext protected mainly by local account permissions unless you have verified full-disk or filesystem encryption.