{"name":"taskexposure","description":"Task-level AI exposure for 923 US occupations, scored against a dated capability reference and published with its own inter-rater reliability.","endpoint":"https://taskexposure.org/mcp","transport":"streamable-http","protocolVersions":["2025-06-18","2025-03-26","2024-11-05"],"authentication":"none","release":"v2026.Q3","tools":[{"name":"search_occupations","description":"Find US occupations by job title and get their AI task exposure scores. Use this first when the user names a job in words rather than a slug. Returns exposed, assisted and untouched shares of weighted task load, the band, the rank out of 923, and median annual pay where published."},{"name":"get_occupation","description":"Get the full published record for one occupation: the three shares, capability and friction, rank and percentile, the mean of each of the six rubric dimensions, BLS median pay and employment, and the most and least exposed tasks. Accepts a slug or a job title."},{"name":"compare_occupations","description":"Compare two occupations and get the reason they differ, separating whether the gap comes from what AI can do (capability) or from what stops it being used (friction). Use this instead of two separate lookups when the question is comparative."},{"name":"rank_occupations","description":"List occupations ranked by exposed share, optionally filtered to one band or to a minimum median wage. Use for questions like 'which jobs are most exposed' or 'what well paid work is least exposed'."},{"name":"get_methodology","description":"Get how this index is built and what it does not claim: the two-factor rubric, the banding rule, inter-rater reliability, and the release identifier the figures belong to. Call this before presenting any figure from this index as a fact."}],"licence":"CC BY 4.0","documentation":"https://taskexposure.org/agents"}