Public leaderboard

Public assessment

vicseeai/vicsee-mcp-server (@vicsee/mcp-server)

vicsee-mcp-server · v0.4.0 · scanned

What changed in the harness

Selection accuracy 94→89, token cost down 0%, unconfirmed writes 0%→0%.

Category breakdown

Where the score comes from.

Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.

01Safety

0.0 / 30

0.0 out of 30
02Legibility

27.9 / 30

27.9 out of 30
03Economics

15.7 / 20

15.7 out of 20
04Discoverability

12.8 / 20

12.8 out of 20

Highest-impact fix

Estimated gain +30 points

Add explicit identity and permission preflight tools

Expose machine-readable principal/tenant confirmation and a non-mutating permission check so agents can verify both before destructive actions.

Description evidence

Defects and rewrites.

2 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.

Tool Defect types Suggested rewrite
vicsee_list_models
no_return_description
List available VicSee models, optionally filtered by media type (image/video/music). Call this first to find a model id to pass to vicsee_generate. Returns a list of model objects, each with its id, supported media type, capabilities, and credit cost.
vicsee_get_credits
no_return_description
Get the current VicSee credit balance for the configured API key. Returns the remaining credit balance as a numeric value.

Selection evidence

Confusable tool pairs.

3 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.

Tool A Tool B Confidence Why they collide
vicsee_upscale_image vicsee_upscale_video medium Both take a media URL and upscale_factor with nearly identical descriptions; if the task doesn't clearly state image vs video (e.g. 'upscale this media file'), the agent could pick the wrong one, though the media type is usually inferable from the URL/extension.
vicsee_generate vicsee_upload low A task like 'turn my local image into a video' requires uploading first then generating; an agent might conflate the two steps, but vicsee_generate's own image_urls description clarifies local paths are auto-handled, reducing real confusion.
vicsee_get_task vicsee_get_credits low Both are simple 'get' status calls but serve clearly distinct purposes (task polling vs balance check) with no overlapping semantics beyond the verb, so confusion is unlikely for any realistic phrasing.

Compare the field

One score is useful.
The evidence makes it actionable.

Back to the leaderboard