MCP server for searching IT job offers on pracuj.pl with filters, JSON output, and market stats. Provides API endpoints as tools and local SQLite search/SQL capabilities.
This server exposes 9 tools with basic MCP registration via mark3labs/mcp-go. All tools have descriptions and input schemas are visible in the code. However, critical gaps limit production readiness: (1) Parameter descriptions are minimal or generic, most lack actionable constraints, ranges, or format guidance. (2) Output schemas are NOT documented anywhere; LLMs cannot predict response structure or plan downstream calls. (3) No error handling guidance; tools return raw API responses without recovery hints. (4) Tool naming is inconsistent: 'companies_list_metadata' vs 'job_offers_list_offers' mixes verb_noun with list_resource patterns. (5) Several parameters lack type clarity (e.g., 'ids' is marked as number but described as 'can be repeated', should be array). (6) No pagination guidance in descriptions despite tools accepting page/per-page. (7) The 'sql' and 'search' tools expose raw database access without clear constraints or examples. Average tool score: 52/100.
Get profile metadata for companies by IDs. Required: ids. Returns the companyMetadata.
Get API domain context: resource taxonomy, auth requirements, query tips, and unique capabilities. Call this first.
Count job offers matching search criteria. Optional: keyword, page, per-page (plus 2 more). Returns the jobOffersCount.
Get grouped counts of job offers by filter categories. Optional: keyword, page, per-page (plus 2 more). Returns array of groupedCount.
List job offers matching search criteria. Optional: keyword, page, per-page (plus 4 more). Returns the jobOffersListing.
Get recommended job offers grouped by category. Optional: language, offset, limit (plus 1 more). Returns the recommendedGrouped.
Output schemas completely undocumented. No tool describes what fields it returns, data types, or structure. LLMs cannot plan downstream calls or extract required IDs for chaining.
Parameter descriptions lack actionable constraints. 'page' has no min/max; 'per-page' has no range; 'keyword' has no length limit. Descriptions should state format, range, and allowed values explicitly.
No error handling guidance. Tools return raw API responses without recovery hints. E.g., if 'keyword' is invalid, what error does the LLM see? How should it retry?
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
Full-text search across all synced data. Faster than paginating list endpoints. Requires sync first.
Run read-only SQL against local database. Use for ad-hoc analysis, aggregations, and joins across synced resources. Requires sync first.
Get total job offer counts by category (IT, main, blue-collar, engineering, sales). Returns the subserviceCounts.
Type mismatch: 'ids' parameter marked as 'number' but description says 'can be repeated'. Should be array type or clarify if multiple calls are needed.
Raw database access via 'sql' tool lacks safeguards. No mention of query limits, timeout, or what tables are available. LLMs could craft expensive queries.