{"id":"wm-action-discovery-2026-09-03","title":"Working Machines action discovery snapshot","publishedAt":"2026-09-03","reviewedAt":"2026-09-03","protocol":"Working Machines MCP","client":"Codex","methodUrl":"https://www.workingmachines.dev/research/agent-action-discovery-methodology","resultUrl":"https://www.workingmachines.dev/research/action-discovery-benchmark-2026-09","method":{"passes":1,"rankingDepth":5,"providerFilterApplied":true,"goldSetReview":"Manual review against the published action names and descriptions","timing":"Client-observed elapsed time for each MCP search_actions call"},"cases":[{"caseId":"stripe-products","query":"list products and prices","providerFilter":"stripe","limit":5,"elapsedMs":500,"expectedMinimalActions":["stripe.list_prices","stripe.list_products"],"rankedActionIds":["stripe.list_prices","stripe.list_products","stripe.search_prices","stripe.create_price","stripe.search_products"],"notes":"Both actions needed for the requested inventory appeared in the first two ranks. Broader search and mutation actions also appeared below them."},{"caseId":"npm-package","query":"get package metadata and versions","providerFilter":"npm","limit":5,"elapsedMs":125,"expectedMinimalActions":["npm.get_package"],"rankedActionIds":["npm.audit_package_versions","npm.get_package_version","npm.get_package","npm.get_version_download_counts","npm.get_package_download_trend"],"notes":"The smallest sufficient action appeared at rank three, so this case failed top-one selection even though the action was present within five results."},{"caseId":"hackernews-top","query":"get top stories","providerFilter":"hackernews","limit":5,"elapsedMs":125,"expectedMinimalActions":["hackernews.get_top_stories"],"rankedActionIds":["hackernews.get_top_stories","hackernews.get_new_stories","hackernews.get_job_stories","hackernews.get_ask_stories","hackernews.get_show_stories"],"notes":"The expected action ranked first. Other story-list variants occupied the remaining results."},{"caseId":"crossref-papers","query":"search scholarly works by title","providerFilter":"crossref","limit":5,"elapsedMs":109,"expectedMinimalActions":["crossref.list_works"],"rankedActionIds":["crossref.list_works","crossref.list_scoped_works","crossref.list_changed_works","crossref.list_resources","crossref.get_work_agency"],"notes":"The general work search action ranked first. More specialized Crossref actions followed it."}],"limitations":["This is a four-case discovery snapshot, not a statistically powered benchmark.","Each case used a provider filter, so the result does not measure cross-catalog provider selection.","Only one pass was recorded; warm and cold conditions were not separated.","Gold sets were manually reviewed and can be challenged or revised with a dated correction.","Elapsed time includes client, network, and runtime overhead and should not be treated as a service-level objective.","No provider action was executed as part of this discovery dataset."],"aggregate":{"caseCount":4,"meanLatencyMs":214.75,"medianLatencyMs":125,"meanReciprocalRank":0.8333333333333333,"recallAtFive":1,"topOneAccuracy":0.75}}