Werkbits Explorer · free
How hard is the move from one occupation to another, and what closes the gap?
Compare any two of the 910 O*NET occupations with skill ratings, see where workers in a role tend to move, and read the arithmetic behind every score. It is the free tool behind Werkbits briefs, running on the same engine.
Explorer
Connect an agent
The explorer above is itself an MCP client. It sends the same JSON-RPC calls an AI assistant sends, to the same public endpoint.
Claude
In Claude, open Settings, then Connectors, and add a custom connector named Werkbits with this URL. Leave authentication empty.
https://werkbits.onrender.com/mcpThen ask in plain words: “How hard is it for a quality control analyst to become an industrial engineer?”
Any HTTP client
Streamable HTTP, MCP protocol 2026-07-28, stateless. One POST per call:
curl https://werkbits.onrender.com/mcp \
-H 'Content-Type: application/json' \
-H 'Accept: application/json, text/event-stream' \
-H 'MCP-Protocol-Version: 2026-07-28' \
-H 'Mcp-Method: tools/call' -H 'Mcp-Name: score_transition' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{
"name":"score_transition",
"arguments":{"source":"wb:quality-control-analysts",
"target":"wb:industrial-engineers"},
"_meta":{"io.modelcontextprotocol/protocolVersion":"2026-07-28"}}}'Tools
- search_occupations
- Find occupations by title or alternate job title.
- decompose_occupation
- Weighted tasks, work activities and rated skills for one occupation.
- rate_skills
- O*NET importance and level for skills, by occupation or by skill.
- score_transition
- Similarity, skill gap, difficulty and skills to train; omit the target for ranked moves.
- glossary
- Data sources, license, scoring definitions.
- tool_info
- Input and output schemas for one tool.
Limits: 60 requests per minute per network. The service sleeps when idle; the first call after a quiet period can take up to a minute.
Methodology
What the scores mean
- similarity
- 0–1. 0.4 × shared work activities (Jaccard) + 0.6 × skill-importance cosine. Validated against observed job changes.
- gap_score
- 0–1. Importance-weighted level shortfall over the target's important skills. What must be learned; not a likelihood of moving.
- transition_score
- similarity × (1 − gap_score).
- difficulty
- low < 0.05 ≤ medium < 0.10 ≤ high, on gap_score. Provisional thresholds.
- importance, level
- 0–1. O*NET importance (1–5) and level (0–7) ratings, rescaled.
How it was checked
Similarity was tested against people who actually changed occupation in the Current Population Survey (CPS ASEC, 2021–2025). The ten destinations Werkbits ranks most similar captured about 36% of observed moves; ranking by destination size alone captured about 16%.
This validates similarity: where people go. The gap score measures what must be learned and is not a probability of moving.
Known limits
13 of 923 occupations have no O*NET skill ratings. Werkbits returns their work activities and says that no gap estimate is possible, rather than guessing.
Data and license
This information includes data from the O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. Mpingo Systems has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications.
What was modified
- Importance (IM, 1–5) rescaled to 0–1 as (IM−1)/4; level (LV, 0–7) rescaled to 0–1 as LV/7.
- Task weights derived from task importance ratings and normalized to sum to 1 per occupation.
- Task-to-skill links derived through the DWA → GWA chain; O*NET publishes skill ratings per occupation.
- Similarity, skill-gap and transition scores are Werkbits computations, not O*NET data.
- Ratings O*NET flags “Recommend Suppress” are excluded.