Back to skill
Skillv0.1.2
ClawScan security
๐ซง Seedance 2.0 Pro โ Pro Pack on RunComfy ยท ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 29, 2026, 1:51 PM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's requirements and runtime instructions consistently match its stated purpose of invoking the RunComfy CLI to run the Seedance 2.0 Pro model; nothing requested looks disproportionate or unrelated.
- Guidance
- This skill appears to do what it says: it wraps the RunComfy CLI to call Seedance 2.0 Pro. Before installing/using: (1) verify you install the official @runcomfy/cli package from the expected publisher (npm org and runcomfy.com) rather than a third-party fork; (2) prefer using runcomfy's interactive login (runcomfy login) rather than long-lived environment tokens where possible; if you must use RUNCOMFY_TOKEN, treat it like a secret โ use least-privilege tokens and rotate/revoke when done; (3) be careful about uploading private images/audio โ media referenced may be uploaded to RunComfy and stored/processed by their service; (4) check RunComfy's privacy/terms about generated content and likeness use; (5) you can inspect ~/.config/runcomfy to see stored credentials and revoke them if needed. Overall the skill is internally consistent, but these operational cautions about credentials and media uploads apply.
Review Dimensions
- Purpose & Capability
- okThe skill declares and requires the RunComfy CLI (runcomfy), a RUNCOMFY_TOKEN, and the RunComfy config directory (~/.config/runcomfy) โ all of which are exactly what a CLI-based client for RunComfy would need to call the bytedance/seedance-v2/pro endpoint. No unrelated credentials, binaries, or config paths are requested.
- Instruction Scope
- okSKILL.md contains explicit CLI commands (runcomfy run ...) and input schemas for image/video/audio references and output directories. The instructions focus on invoking the RunComfy model endpoint and do not direct the agent to read unrelated system files or environment variables. Note: the CLI may accept local file paths for media references (or remote URLs), so using local private media will cause uploads to RunComfy โ this is expected for the feature but worth user attention.
- Install Mechanism
- okThere is no install spec in the skill bundle (instruction-only). The README notes installing the official @runcomfy/cli via npm as a prerequisite, which is a normal, proportional step and not performed automatically by the skill. No arbitrary downloads, archives, or unusual install locations are present.
- Credentials
- okOnly RUNCOMFY_TOKEN and the standard RunComfy config path are required. That is proportional: the token is needed to authenticate the CLI to the RunComfy service. Users should understand that this token grants the CLI access to their RunComfy account and any resources the account can access (models, uploads, outputs).
- Persistence & Privilege
- okThe skill is not always-enabled (always: false) and is user-invocable. It does not request system-wide persistent privileges or modifications to other skills. The default model-invocation setting (agent may call the skill autonomously) is standard and not a red flag by itself.
