# oddly > The world, on the record. AI infrastructure built on the observed web. - name: com.myoddly/oddly-corpus - version: 1.1.0 - connection class: R0 (autonomous) - trust: 82/100 flags: no-repository - quality: 80/100 (good) - owner: not claimed - descriptor: https://api.protogrid.dev/v1/servers/com.myoddly%2Foddly-corpus - tools: https://api.protogrid.dev/v1/servers/com.myoddly%2Foddly-corpus/tools **An agent can connect right now, no human step.** Remote endpoint, no authentication, reachable on the last probe. ## Connection (mcpServers) ```json { "mcpServers": { "oddly-corpus": { "type": "http", "url": "https://myoddly.com/api/public/mcp" } } } ``` ## Trust - hygiene: 80 - liveness: 98 - freshness: 100 - provenance: 55 - drivers: +dns-namespace +website +reachable +uptime-94% +updated-11d-ago -no-repository - Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed. ## Quality 80/100 (good) ### protocol: 75 (weight 25) - warn `protocol.modern`: newest supported version is 2025-06-18; 2026-07-28 not supported, older versions are deprecated until 2027-07-28 - n/a `protocol.stateless`: only applies to 2026-07-28 servers - pass `protocol.transport`: streamable HTTP - n/a `protocol.list_ttl`: only applies to 2026-07-28 servers ### authorization: n/a (weight 15) - n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth - n/a `auth.secret_in_url`: no templated URL ### tool hygiene: 86 (weight 30) - pass `tools.descriptions`: every tool has a description - pass `tools.description_length`: descriptions are concise - pass `tools.schemas`: every tool has an object input schema - pass `tools.annotations`: 6 of 6 tools declare readOnlyHint or destructiveHint - fail `tools.directory_hints`: no tool declares all four of readOnlyHint, destructiveHint, idempotentHint and openWorldHint; missing: record_answer (destructiveHint, idempotentHint); record_subject_history (destructiveHint, idempotentHint); record_verify_receipt (destructiveHint, idempotentHint); … - pass `tools.token_cost`: about 1,165 tokens to load every tool - pass `tools.api_dump`: tools are not a one-to-one API dump ### stability: 75 (weight 15) - warn `stability.changes`: 9 tool changes in 30 days - pass `stability.rug_pull`: no tool changed its meaning under the same name ### dependencies: n/a (weight 15) - n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package - tools: 6, about 1,165 tokens to load them all - tool set hash: 085902c36e873641005e83138952540e - tools last changed: 2026-10-09T00:04:57.359Z - badge: [![protogrid quality](https://protogrid.dev/badge/com.myoddly/oddly-corpus.svg)](https://protogrid.dev/servers/com.myoddly/oddly-corpus) - Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed. ### Tool changes - 2026-10-09 `store_audit` changed (annotations) - 2026-10-09 `benchmark_lookup` changed (annotations) - 2026-10-09 `benchmark_categories` changed (annotations) - 2026-10-09 `record_answer` changed (annotations) - 2026-10-09 `record_verify_receipt` changed (annotations) - 2026-10-09 `record_subject_history` changed (annotations) - 2026-10-08 `record_answer` added - Answer one question from oddly's record: the newest observed value of one attribute of one subject, with its figure, both clocks, its source, its receipt and its scope. The figure is what the source published when oddly looked, not an estimate, a forecast or advice. Rate: meter data.answer, US$0.001 per answer, first 100 per month free. An unanswered question (a subject or attribute the record has not observed) is never metered. - 2026-10-08 `record_verify_receipt` added - Recompute and check one receipt end to end: the row's receipt hash recomputed from its own sealed document, the link to the row before it, the day root its day folds into, and that root's anchor state (confirmed with a block height, or pending). Every check is named in the verdict, and a check that could not run says so rather than passing. Rate: never metered (verification is on the price book's never_metered list). - 2026-10-08 `record_subject_history` added - One subject's change history and current state from oddly's record, with both clocks on every row: observed_at (when the source says it was true) and fetched_at (when oddly looked), and the receipt hash of each row so any value can be checked with record_verify_receipt. Rate: meter data.history, US$0, attribution required (cite oddly and the subject). Returns an honest empty answer for a subject the record has never observed, which is never metered. Full history: https://api.protogrid.dev/v1/servers/com.myoddly%2Foddly-corpus/changes ## Remotes - https://myoddly.com/api/public/mcp (streamable-http, auth none, reachable true, uptime30d 0.9411765) ## Packages _none_ ## Tools (6) - `benchmark_categories`: List the product categories the oddly corpus currently covers, with how many observations and metrics back each one. Call this FIRST to discover valid category paths before calling benchmark_lookup, rather than guessing a path. - `benchmark_lookup`: Observed price and assortment benchmarks for a product category: the median and the p25-p75 band across real public storefront listings, with the sample size, the currency, and the exact scope of each figure. Figures are observed market reference, not targets or advice. Returns an honest empty answer when a category has too little data to serve; it never estimates or fills a gap. This corpus does NOT hold conversion rates, revenue, or AOV, so do not ask it for those. - `record_answer`: Answer one question from oddly's record: the newest observed value of one attribute of one subject, with its figure, both clocks, its source, its receipt and its scope. The figure is what the source published when oddly looked, not an estimate, a forecast or advice. Rate: meter data.answer, US$0.001 per answer, first 100 per month free. An unanswered question (a subject or attribute the record has not observed) is never metered. - `record_subject_history`: One subject's change history and current state from oddly's record, with both clocks on every row: observed_at (when the source says it was true) and fetched_at (when oddly looked), and the receipt hash of each row so any value can be checked with record_verify_receipt. Rate: meter data.history, US$0, attribution required (cite oddly and the subject). Returns an honest empty answer for a subject the record has never observed, which is never metered. - `record_verify_receipt`: Recompute and check one receipt end to end: the row's receipt hash recomputed from its own sealed document, the link to the row before it, the day root its day folds into, and that root's anchor state (confirmed with a block height, or pending). Every check is named in the verdict, and a check that could not run says so rather than passing. Rate: never metered (verification is on the price book's never_metered list). - `store_audit`: Run oddly's free storefront audit against a public ecommerce domain and return the findings: detected platform, the checks that failed, and how the store's catalogue compares to observed category norms where the corpus can say. Reads the store's own public pages only, obeys robots.txt, and is rate limited. Use the bare domain (example.com). Never pass an internal host or an IP address.