Back to skill
Skillv1.0.0
ClawScan security
Ai Image To Video Joyfun · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
SuspiciousApr 16, 2026, 5:11 AM
- Verdict
- suspicious
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill's actions mostly match an image->video cloud service (it needs a NEMO_TOKEN and calls nemovideo.ai), but there are small incoherences around declared config paths and filesystem detection that warrant caution.
- Guidance
- This skill appears to be a thin integration with the nemovideo.ai rendering service and needs a NEMO_TOKEN to operate. Things to consider before installing: - Confirm you trust the domain mega-api-prod.nemovideo.ai and the skill author (source unknown, no homepage). - If you don't want the skill to obtain credentials for you, set NEMO_TOKEN yourself (do not paste private tokens into chat). Otherwise the skill will call the anonymous-token endpoint and persist the returned token for 7 days. - The skill asks the agent to detect install paths (~/.clawhub, ~/.cursor/skills) and references ~/.config/nemovideo/ — decide whether you are comfortable with the agent reading those filesystem locations. - Only upload images you are willing to send to an external service; uploads and exports will go to the service's API endpoints. - If you need higher assurance, request the skill source or a trusted homepage, or ask the author to remove filesystem-detection behavior. Revoke any auto-generated tokens you don't recognize. Why I marked this suspicious: minor inconsistencies (declared configPaths in SKILL.md vs registry metadata) and the instruction to probe local install/config paths expand the agent's read scope beyond the minimal needs for uploading images. More provenance (source/homepage) or removal of unnecessary filesystem access would raise confidence.
Review Dimensions
- Purpose & Capability
- noteThe declared primaryEnv (NEMO_TOKEN) and the API endpoints in SKILL.md align with an image-to-video cloud service. However, the SKILL.md frontmatter lists a config path (~/.config/nemovideo/) and references detecting install paths for header population, which is inconsistent with the registry metadata that said no required config paths and is not strictly necessary for the stated purpose.
- Instruction Scope
- concernRuntime instructions direct the agent to obtain or generate a bearer token, create sessions, upload local files, and poll SSE endpoints on mega-api-prod.nemovideo.ai — all expected. But the instructions also tell the agent to detect an install path (~/.clawhub/ or ~/.cursor/skills/) and derive X-Skill-Platform from it, and reference a local config dir (~/.config/nemovideo/). This implies filesystem reads beyond just user-supplied images and may persist tokens/session IDs; that scope creep is not fully justified in the description.
- Install Mechanism
- okInstruction-only skill with no install spec and no code files — lowest install risk. All runtime behavior is network/API oriented.
- Credentials
- noteOnly one credential is required: NEMO_TOKEN (declared as primaryEnv). That is appropriate for a cloud rendering service. The skill also describes generating an anonymous token via an API call if no env var is provided — acceptable but means the skill will autonomously obtain and store credentials if run without a pre-set token.
- Persistence & Privilege
- okSkill is not always-enabled and does not request elevated platform privileges. It does instruct saving session_id and token values for operation, which is normal for service sessions but means credentials may be persisted to agent state or local config if implemented.