Loading ToolBench

Reading the current corpus…

jobspy-mcp-server MCP server · C · ToolBench

jobspy-mcp-server

github.com/chinpeerapat/jobspy-mcp-server

MCP server exposing the python-jobspy job-board scraper (LinkedIn, Indeed, Glassdoor, ZipRecruiter, Google, Bayt, Naukri, BDJobs)

2026-07-28+STDIO1 tools

65/100

Definition Quality89

Protocol Readiness50

Supportability35

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence · evidence: stateless requests

Current-spec patterns detected

Summary

Single tool 'search_jobs' has excellent naming (verb_noun), comprehensive parameter schema with proper enums and constraints, and well-structured output schema (SearchJobsResult). Description is detailed (400+ chars) and explains use case clearly. Tool annotations present (readOnlyHint, idempotentHint, destructiveHint). However, some parameter descriptions could be more concise, and error handling guidance is minimal. Output schema properly documented with Pydantic models. No critical gaps, but room for improvement in error messaging and recovery guidance.

Tools (1)

search_jobsread onlysource verified89/100

Search live job postings across job boards. Returns compact, point-in-time job records (title, company, location, salary, posted date, URL) scraped at call time. Use offset to page through more results (offset applies per site). Does NOT apply to jobs, fetch company reviews, or save searches/alerts - it only reads current public listings.

Issues (3)

MEDIUMrecovery-guide ↗

Error handling lacks recovery guidance. ToolError exceptions (e.g., invalid country_indeed) are raised but do not suggest next steps or available alternatives to the LLM.

search_jobs

LOWtool-description ↗

Parameter descriptions are verbose (some 200+ chars). Descriptions like 'google_search_term' could be condensed to 50-100 chars while retaining clarity, reducing token overhead.

search_jobs

LOWtool ↗

No batch or multi-query variant. If an agent needs to search multiple job titles or locations, it must call search_jobs repeatedly, wasting tokens and latency.

search_jobs

Recommendations

Enhance error messages with recovery hints: e.g., 'Unknown country_indeed: "xyz". Supported countries: usa, uk, canada, australia. See jobspy://countries resource.'
Condense parameter descriptions to 50-100 chars where possible. E.g., 'google_search_term' → 'Natural-language query for Google Jobs (required if google in sites).'
Add a dry-run or validation mode to preview results before committing to a full scrape, reducing wasted API calls.
Document timeout behavior and retry guidance in the tool description (SCRAPE_TIMEOUT_SECONDS = 120.0 is set but not exposed to LLM).
Consider adding a batch_search_jobs variant accepting an array of search configs to reduce multi-query overhead.

Score history

Overall score trend

First recorded score · v2 rubric

65/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	C	65	2026-07-28+	v2