01Safety
30.0 / 30
What changed in the harness
Selection accuracy 100→98, token cost up 4%, unconfirmed writes 100%→100%.
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
28.1 / 30
03Economics
20.0 / 20
04Discoverability
14.0 / 20
Highest-impact fix
Estimated gain +6 pointsMake target tools discoverable on the first call
Clarify tool names, decision boundaries, and required argument schemas so an agent can choose and construct the target call without exploratory steps.
Description evidence
Defects and rewrites.
4 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 |
|---|---|---|
gpu_run |
no_return_description |
Submit a GPU-Bridge AI service job. 30 services available: LLM inference (sub-second), image generation (FLUX, SD3.5), video generation, video enhancement (up to 4K), speech-to-text (Whisper, <1s), TTS (40+ voices), music generation, voice cloning, embeddings, document reranking (Jina), OCR, PDF/document parsing, NSFW detection, image captioning, visual Q&A, background removal, face restoration, upscaling, stickers, and more. Use gpu_catalog to see all available services. Returns a job ID for the submitted job; poll it with gpu_status to retrieve the final result. |
gpu_catalog |
no_return_description |
Returns a catalog of all available GPU-Bridge services, each with its pricing and model information. No authentication required. |
gpu_balance |
no_return_description |
Returns the GPU-Bridge account's current credit balance, daily spend, volume discount tier, and recent job history. |
gpu_estimate |
no_return_description |
Returns the estimated cost of running a given GPU-Bridge service before submitting it, based on the requested service and optional runtime in seconds. No authentication required. |
Selection evidence
Confusable tool pairs.
4 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
gpu_balance |
gpu_estimate |
medium | Both concern money; a task like 'do I have enough credits to run image generation?' conflates current account balance with per-run cost, so the agent could call either tool. |
gpu_catalog |
gpu_estimate |
medium | Both surface pricing; 'what does it cost to use llm-4090?' could send the agent to the catalog's pricing list instead of a per-service estimate, or vice versa. |
gpu_run |
gpu_estimate |
medium | Both take a service key; 'run image generation and tell me what it costs' splits between executing the job and estimating cost, and 'how much to generate an image' could wrongly trigger gpu_run because it mentions generation. |
gpu_status |
gpu_balance |
low | gpu_balance's description includes job history, so 'show me my past jobs/results' could loosely match gpu_status, though status is the dedicated results-retrieval tool. |
Compare the field