A Next.js-based AI chat application with artifact generation (code, text, images, sheets), document management, and blockchain integration for Sei network transactions
SeiTrage has 6 tools with severe definition gaps. Only 2 tools (createDocument, updateDocument) have descriptions; 4 tools (getWeather, getBalance, transferSei, requestSuggestions) lack descriptions entirely. No input schemas are visible in the provided source code for any tool. Tool names are verb-based (good), but without schemas and descriptions, LLMs cannot reliably select or invoke these tools. The codebase shows tool definitions in app/api/chat/route.ts but the actual schema definitions are not included in the provided excerpt, making it impossible to verify parameter types, constraints, or output structures. This is a critical blocker for production use.
Creates a new artifact document (code, text, image, or sheet) with AI-generated content
Requests AI-generated suggestions for a document
Updates an existing artifact document with new AI-generated content
4 of 6 tools lack descriptions entirely (getWeather, getBalance, transferSei, requestSuggestions). LLMs cannot determine when or why to select these tools.
No input schemas visible in source code for any tool. Cannot verify parameter types, constraints, or whether parameters have descriptions. Hard cap on schema scores.
transferSei is a DESTRUCTIVE tool (transfers cryptocurrency) with no description, no visible schema, and no error handling guidance. Agents cannot safely invoke this without explicit recovery instructions.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 32 | <=2025-11-25 | v2 |
createDocument and updateDocument have minimal descriptions (65 chars each). Descriptions lack WHEN to use, prerequisites, or dependency hints. Parameter descriptions for 'kind' enum are present but others may be missing.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., document IDs for chaining).