Loading ToolBench
JavaScript/TypeScript SDK for the SuperMe API — ask experts, search perspectives, manage knowledge, and more.
SuperMe SDK exposes 26 tools with consistent naming (verb_noun pattern) and reasonable descriptions (avg ~100 chars). All tools have input schemas with typed parameters and descriptions. However, output schemas are not documented in the source, only inferred from examples. Error handling is minimal; no recovery guidance or categorization. Several tools lack parameter constraints (enums, ranges). Streaming tools (ask_my_agent_stream, group_converse_stream) are well-designed but lack explicit output format documentation. Security: API key is passed to constructor, not exposed as tool param (good). Overall: solid foundation but missing output documentation and error guidance expected of production tools.
Submit URLs to be crawled and added to your knowledge base.
Save notes or knowledge to your personal library.
Ask a question to a specific user's SuperMe agent.
Talk to your own SuperMe AI agent.
Stream your own agent's response. Yields string chunks. Final item includes `{ _done: true, conversation_id }`.
Check which URLs are not yet in your knowledge base.
Output schemas not documented. Examples show streaming tools yield chunks with _done flag, but no formal schema definition exists in source. LLMs cannot plan downstream operations without knowing return structure.
No error handling guidance. Tools lack recovery hints (e.g., 'User not found. Try find_user_by_name() first.'). Errors are not categorized as retryable, user-fixable, or fatal.
Missing parameter constraints. 'limit' and 'max_tokens' parameters lack min/max bounds. 'platform' in connect_social should be an enum, not free-form string.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2025-06-18+ | v2 |
Connect a custom blog or website.
Connect a social platform account. GitHub and Beehiiv require a token.
Disconnect a custom blog or website.
Disconnect a social platform account.
Search for users by name.
Resolve multiple names to SuperMe users in one call.
List all connected social accounts and blogs.
Fetch a single conversation with all its messages.
Poll interview status and stages.
Get full transcript for a completed interview.
Get a user's public profile. Omit identifier to get your own profile.
Start or continue a multi-turn group conversation between multiple users.
Stream a group conversation. Yields per-perspective chunks. Final item has `{ _done: true }`.
List active job roles across all companies.
List your most recent conversations.
List your interviews.
Get perspectives from multiple experts on a topic.
Start a background agent interview for a role. Returns immediately with status "preparing".
Stream interview events via SSE. Yields event dicts; stops at terminal status.
Update an existing note in your library.
Streaming output format undocumented. ask_my_agent_stream description mentions '_done' and 'conversation_id' but no formal schema. group_converse_stream mentions 'per-perspective chunks' without defining structure.
No pagination documentation. list_conversations and list_active_roles accept 'limit' but no mention of offset, cursor, or total_count in response. Large result sets could exhaust context.