Loading ToolBench
MCP server that gives AI assistants access to LinkedIn Sales Navigator contact and account search.
Server has 10 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-50 chars, below the 50-200 char baseline for LLM optimization) and input schemas are present but minimal. Parameter descriptions are sparse or missing entirely. Output schemas are not documented. Error handling is generic. The server masks error details (mask_error_details=True) which prevents actionable recovery guidance. Tool naming follows verb_noun convention (search_*, get_*, send_*, list_*) which is good, but descriptions lack WHEN/WHY context and dependency hints that LLMs need for tool selection.
Check if the browser has an active LinkedIn Sales Navigator session
Enrich lead records with additional profile information
Export search results to JSON or CSV file
Retrieve detailed profile information for a lead
Retrieve saved search results from the local store
List all saved search queries and their progress
Run an unattended batch of outreach messages using MCP sampling
Search for accounts/companies in LinkedIn Sales Navigator
Descriptions are too short (10-50 chars vs. 50-200 char baseline). 'Search for contacts in LinkedIn Sales Navigator' lacks WHEN to use, prerequisites, or distinction from search_accounts. LLMs cannot reliably select between similar tools.
Parameter descriptions are missing or minimal. 'url' param in search_contacts has description 'Full Sales Navigator search URL' but does not explain format, constraints, or how to construct it. 'pages' lacks guidance on valid range or performance impact.
Output schemas are not documented. Tools return data but LLMs cannot plan downstream calls without knowing response structure. E.g., search_contacts returns contacts but fields (profile_id, name, email, etc.) are not specified.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | <=2025-11-25 | v2 |
Search for contacts in LinkedIn Sales Navigator
Send an outreach message to a lead
mask_error_details=True prevents actionable error messages. When a tool fails, LLMs receive generic errors with no guidance on retry, user input, or next steps. Violates recovery-guide pattern.
run_batch tool uses 'MCP sampling' (per description) but sampling is deprecated as of 2025-03-26. Tool should integrate directly with LLM provider API or use Multi Round-Trip Requests (result 'input_required') for user confirmation instead.