01Safety
0.0 / 30
What changed in the harness
Selection accuracy 100→100, 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.
02Legibility
28.6 / 30
03Economics
18.2 / 20
04Discoverability
12.2 / 20
Highest-impact fix
Estimated gain +30 pointsAdd 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.
5 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 |
|---|---|---|
list_models |
no_return_description |
List available AI image generation models and their capabilities, returning each model's ID, provider, active status, and supported options (e.g. resolutions, tiers). For up-to-date pricing, see https://www.meigen.ai/model-comparison. |
get_inspiration |
no_return_description |
Get the full prompt and all image URLs for a gallery entry, returning the prompt text and an array of image URLs. Show the images to the user as visual examples. The prompt can be used directly with generate_image(), and image URLs can be passed as referenceImages for style transfer. |
manage_preferences |
no_return_description |
Read or update user preferences: default style, aspect ratio, model, style notes, and favorite prompts. Returns the current preferences object after the requested action completes (full preferences for "get"/"set", updated favorites list for "add_favorite"/"remove_favorite"). Call with action "get" at conversation start to load preferences. |
comfyui_workflow |
no_return_description |
Manage ComfyUI workflow templates: list, view parameters, import from file, modify settings, or delete. Returns action-dependent data — an array of workflow names for "list", node/parameter details for "view", or a confirmation of the change for "import"/"modify"/"delete". |
generate_video |
no_return_description |
Generate a video using AI via MeiGen platform (text-to-video, image-to-video, or Seedance 2.0 reference-video continuation), returning the resulting video URL once generation completes (typically 1-5 minutes, up to ~8 min for Veo at 4K). Supports models Seedance 2.0 (mini/fast/pro tiers, 4-15s), Happyhorse 1.0 (3-15s), Veo 3.1 (fast/pro, 4/6/8s, native audio), and Grok Video 1.5 (image-to-video only, native audio, 4-15s, 480p/720p). Pricing varies by model and tier — see https://www.meigen.ai/model-comparison. |
Selection evidence
Confusable tool pairs.
1 pair where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
generate_image |
generate_video |
low | Both create AI media from a prompt, but the definitions clearly diverge on output type (image vs video) and distinct parameters (aspectRatio/resolution/referenceImages vs duration/firstFrame/referenceVideo); a task explicitly asking for an 'image' or 'video' leaves little ambiguity, though a vague 'create visual content of X' request could momentarily cause hesitation. |
Compare the field