Every NPV, IRR, ore tonnage, and grade is extracted from an NI 43-101 technical report on file with SEDAR, checked against the source text, and rejected if it isn't there. Nothing from model memory.
One protocol — bring it to the assistant you already use
The outcomes desks connect SEDAR for — before the feature list.
Every NPV, IRR, tonnage, and grade is checked against the NI 43-101 text and dropped if it isn't there — so you cite figures you can stand behind, not model recall.
Bilingual sentence search and a material-change timeline the US side has no equivalent for — the coverage a resource desk actually needs.
Screen 758 projects by commodity, jurisdiction, and economics, then pull a single project's full record with its source accession — no PDF hunting.
One question. Your AI composes the tools and returns a cited answer.
Show me the top 5 lithium projects on SEDAR by NPV.
Give me the full profile of McEwen Mining's Los Azules copper project.
Which Canadian issuers disclosed cyber incidents in 2025?
Show me every material change Ballard Power has filed since 2025.
Technical-report coverage 2020–2026, refreshed on a rolling cadence as new 43-101s are filed. Sentence-level bilingual search spans the broader SEDAR corpus, 2021–2026. Projects are those of Canadian-listed issuers filing on SEDAR. Access is by invitation.
Filter NI 43-101 projects by commodity, jurisdiction, stage, and NPV/IRR thresholds. An economics filter restricts to verified-economics rows; skip it for the full 758-project landscape.
Full structured record for a single project — every extracted field plus the source accession number so you can pull the raw filing for verification.
The Canadian 8-K-equivalent timeline — every Material Change Report classified into 10 event types with a condensed summary from the filer.
Bilingual semantic search across ~46M sentences of SEDAR filings (2021–2026). Ask in plain language, in English or French.
Pull specific sentences from a filing by ticker, company name, or accession — for citation, context, and verification.
Fuzzy company-name lookup — handles typos, aliases, former names, and delisted issuers → the canonical record.
A rich identity card — legal name, ticker, exchange, incorporation, EIN/LEI, former names, and address history.
Semantic search across the tool catalog by intent — call it first when you are not sure which tool to use.
Every economic figure passes three checks before it's shown. Anything that fails comes back blank — “unknown,” never invented — so you only ever see numbers you can defend.
The number was found in the source filing text — verbatim or a clear paraphrase. No model recall, no inference.
The value falls within the expected range for that metric, commodity, and jurisdiction — extraction artifacts rejected.
Values repeated across investor-deck templates — not project-specific — are excluded, never surfaced as unique.
Screen the resource landscape and pull verified NPV/IRR straight from the 43-101.
Compare projects by commodity and jurisdiction, with numbers you can defend.
Bilingual search and the material-change timeline the US side has no equivalent for.
Paste the endpoint into your client's connector settings. Sign in with your allowlisted email — access is by invitation.
https://mcp-sedar.kscope.io/mcp/
Works with Claude (desktop & web), Claude Code, Cursor, and any MCP client over HTTP. On first use your client opens a sign-in tab — enter your email for a one-time link.
Request a demo and we'll show you cited, point-in-time canadian mining answers in your own AI client.
Request a Demo