maple-procure
MapleProcure is an MCP server (stdio + streamable HTTP) that answers questions about Canadian federal government contracts from official CanadaBuys open data.
search_tenders— find open tenders still accepting bids by keyword, category, region, or closing-before date; results ordered soonest-closing or by relevance, with citation-readynotice_urlwhen one exists.get_tender— fetch one open notice in full by exact reference number; buyer contact details only with thepiiscope and only when the user actually asks for them.search_awards— look up awarded contracts: who won, for how much, when, for which buyer; filter by keywords, supplier name, or date range.summarize_spend— rank total disclosed contract value by buyer, supplier, or category over a period, for one currency at a time.export_report— write a markdown or CSV brief of specific notices; requires theexportscope and never writes on the first call (needs a content-bound, single-useconfirm_handlewithin five minutes).Grounded answers — every result carries provenance (
as_of,source_file,source_row_count) plus plain-languagenotesstating what the data does not cover (unvalued awards ~30%, missing regions, null notice URLs, snapshots vs. live feed).Scoped access — static bearer tokens with
read,pii, andexportscopes; RFC 6750-style 401 challenges distinguish missing vs. wrong credentials, and scope refusals surface in-protocol with a remedy.Pipeline-friendly REST routes —
GET /v1/tenders/searchandGET /v1/tenders/{reference_number}return the same envelope as the tools but express failures as HTTP statuses (401/403/422/503) so automation can branch and retry.Public endpoints —
/version,/healthz(liveness), and/readyz(verifies ingest provenance per source) need no credential.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@maple-procure Find open IT tenders closing this month."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MapleProcure
This lets an AI assistant answer questions about Canadian government contracts: what is open for bid right now, who won past work, and for how much. Every answer is built from the official CanadaBuys records and carries the date and the source file it came from.
Ask "which IT tenders close in the next 30 days?" or "who has won snow-removal contracts in Ontario, and for how much?" and get an answer traceable to the government notice wherever the source publishes one.
How it works
An MCP server in Python, so any MCP client can call it. Five tools over stdio and over streamable HTTP. FastAPI hosts the HTTP transport, SQLite holds the data, and
make ingestrebuilds that database from the CanadaBuys open-data CSVs at runtime. No data is committed here.Every caller has scopes, and the boundary is enforced. A bearer token carries scopes, and reading a tender, reading a buyer's contact details and writing an export are three different permissions. A missing credential and a wrong one return different errors, per RFC 6750.
The only tool that writes takes two calls.
export_reportrenders a preview and writes nothing. Writing needs a second call carrying a sealed handle bound to a hash of the rendered document, so what a person approved was the document, not an argument list.
Status: deployed and serving. Driven end to end from Claude Desktop on 2026-08-16 and
deployed the same day. Live at maple-procure.onrender.com
(the root path answers 401 by design; /version and /readyz are public). The transcript under
Try it live is real output from the running service, and a 62-second recording of
it driven by curl is in docs/demo.gif. This README does not describe a
capability until it exists in the code. Current state is in PROGRESS.md; accepted
limitations are in KNOWN_LIMITATIONS.md.
Related MCP server: govtenders-mcp
Run it
Requires Python 3.12 or newer and uv. From a clean clone:
make install # dependencies into .venv
make test # 373 tests
make ingest # download the CanadaBuys CSVs (about 17 MB) and build data/maple.sqlite
make serve-stdio # or: make serve-httpmake test needs no network and no database: the suite runs against the fixtures in
tests/fixtures/. make ingest is the step that reaches the network, and how long it takes is
download time, not processing. make lint runs ruff.
A clean clone reproduces the database from the official source, so nothing here has to be trusted
as a snapshot. Row counts move as CanadaBuys publishes; the counts quoted in
docs/DATA.md are what a full load showed on 2026-08-16.
Try it live
The server is deployed at https://maple-procure.onrender.com (Render free tier: the
first request after idle triggers a fresh ingest from CanadaBuys, so give a cold start up to a
minute, measured at 41s from fully idle).
/version and /readyz need no credential:
curl https://maple-procure.onrender.com/versionEverything else needs a bearer token, and there are two ways to get one.
A read-only demo token for the deployed service is available on request. Email
samarpartapp@gmail.com. It holds read and nothing else, so it can search tenders and awards but
cannot export reports or read buyer contact details.
Or run your own and mint your own tokens. One environment variable,
MAPLE_PROCURE_TOKENS, holds them all, each entry subject:secret:scope,scope. The three scopes
are read, pii and export, and an unknown scope name aborts at startup rather than surfacing
later as a permission error. See Over HTTP for the whole setup.
Every command below reads the token from $TOKEN:
export TOKEN=... # the demo token, or one of your own
curl -X POST https://maple-procure.onrender.com/mcp \
-H "Authorization: Bearer $TOKEN" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'Verified in production
The session below was run against https://maple-procure.onrender.com on 2026-08-16.
Commands and responses are copied verbatim, with two edits: tools/list is trimmed to the tool
names, which is what the jq filter does, and the bearer token is shown as $TOKEN.
The five tools are live. Responses come back as SSE, so data: is stripped before jq:
$ curl -s -X POST https://maple-procure.onrender.com/mcp \
-H "Authorization: Bearer $TOKEN" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' \
| sed -n 's/^data: //p' | jq -r '.result.tools[].name'
search_tenders
get_tender
search_awards
summarize_spend
export_reportThe auth boundary is real, not decorative. Both failures are RFC 6750 challenges, and they are distinguishable. A missing credential is not the same error as a bad one:
$ curl -si -X POST .../mcp -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | grep -i 'HTTP/\|www-authenticate'
HTTP/2 401
www-authenticate: Bearer error="invalid_request", error_description="an Authorization header is required"
$ curl -si -X POST .../mcp -H "Authorization: Bearer not-a-real-token" ... | grep -i 'HTTP/\|www-authenticate'
HTTP/2 401
www-authenticate: Bearer error="invalid_token", error_description="the bearer token is not recognised"The demo token really is read-only. It holds read and nothing else, so the two
tools that reach past reading are refused at the scope check. The export never previews, and
the contact fields are never assembled:
$ ... -d '{... "name":"export_report","arguments":{"tender_refs":["MX-2026-00001"]}}}'
{"code":-32002,"message":"this request needs the 'export' scope, which this caller does not
hold. Held scopes: read.","data":{"remedy":"this needs a token carrying the missing scope;
do not retry as-is"}}
$ ... -d '{... "name":"get_tender","arguments":{"reference_number":"...","include_contact":true}}}'
{"code":-32002,"message":"this request needs the 'pii' scope, which this caller does not
hold. Held scopes: read.","data":{"remedy":"this needs a token carrying the missing scope;
do not retry as-is"}}Every result is dated and traceable. A live search_tenders for software, limited to
two rows. Note that the provenance block and notes are part of the payload, not added here:
$ ... -d '{... "name":"search_tenders","arguments":{"keywords":"software","limit":2}}}'
{
"rows": [
{
"reference_number": "SSC-26-00034447:T",
"title": "VIRTANA HARDWARE & SOFTWARE MAINTENANCE AND SUPPORT",
"buyer_name": "Shared Services Canada",
"closing_date": "2026-08-17T14:00:00",
"category": "Services related to goods",
"regions_of_delivery": "Ontario (except NCR)",
"notice_url": null
},
{
"reference_number": "WS3971158100-Doc3971158168",
"title": "Software Licensing Supply Arrangement (SLSA) RFSA",
"buyer_name": "Department of Public Works and Government Services (PSPC)",
"closing_date": "2028-09-29T10:00:00",
"category": "Goods",
"regions_of_delivery": "Canada",
"notice_url": "https://canadabuys.canada.ca/en/tender-opportunities/tender-notice/WS3971158100-Doc3971158168"
}
],
"row_count": 2,
"total_matches": 53,
"withheld": 51,
"truncated": true,
"as_of": "2026-08-16T21:13:41+00:00",
"source_file": "https://canadabuys.canada.ca/opendata/pub/openTenderNotice-ouvertAvisAppelOffres.csv",
"source_last_modified": "Sun, 16 Aug 2026 10:20:15 GMT",
"source_row_count": 937,
"notes": [
"Data as of 2026-08-16T21:13:41+00:00, from the last ingest of ... — a snapshot, not a
live feed. Newer notices may exist.",
"Showing 2 of 53 matches; 51 withheld. Narrow the search or raise `limit` (max 100) —
do not describe these rows as the complete set."
]
}The first row is the honest case for the caveat above: notice_url is null because its
legacy SSC- reference does not resolve on the modern CanadaBuys site, so the row carries no
URL rather than a broken one. That is 5% of open tenders, and 64% of awards, which is why
neither tool promises a link on every row. See
KNOWN_LIMITATIONS.md.
Tools
Tool | Question it answers | |
| What's open right now, by keyword, category, region, or closing date? | ✅ |
| Everything on one notice. | ✅ |
| Who won what, when, and for how much? | ✅ |
| Totals by buyer, supplier, category or period. | ✅ |
| Write a brief. Only after a human confirms. | ✅ |
Every result carries the source file it came from and when that file was last ingested, so an answer can always be dated and traced back to the official notice.
Connect a client
Point a client at the stdio server. For Claude Desktop or Claude Code:
{
"mcpServers": {
"maple-procure": {
"command": "uv",
"args": ["run", "--directory", "/absolute/path/to/maple-procure",
"python", "-m", "maple_procure.server"]
}
}
}Over stdio there is no transport that could carry a credential, so scopes come from the
environment and default to read alone. Buyer contact details and exports each need theirs
turned on explicitly:
MAPLE_PROCURE_SCOPES=read,pii,export make serve-stdioAn unknown scope name aborts at startup rather than surfacing later as a permission error. This variable is the stdio fallback only. Over HTTP, scopes come from the token.
The write needs two calls
export_report is the only tool that writes, and it cannot be made to write on a single call,
on any client, under any argument combination. The first call renders the exact document,
returns it as a preview, and writes nothing. Writing needs a second call carrying the
confirm_handle the preview minted.
A handle is sealed with the SDK's AES-256-GCM request_state codec, so a client cannot read or
forge one. It expires in five minutes, works exactly once, and is bound to a hash of the
rendered content. Change the notices or the format and the old handle stops matching, because
what a human approved was a document, not an argument list.
Clients that declare form elicitation get a real prompt instead (MRTR: the server returns
input_required and the client retries with the answer). Both paths redeem the same handle, so
the optional one is not a weaker gate. Today's Claude Desktop and Claude Code negotiate
2025-11-25, so in practice you will see the handle.
What this does not prove is that a human read the preview. On the handle path, a model that passes the handle straight back has satisfied the mechanism, and in live testing one did exactly that when told to "export it again". So state the guarantee precisely:
What gets written is exactly what was previewed: content-bound, single-use, tamper-evident. Not: a human read the preview.
Only the elicitation path can make the confirmation terminate at a person. The full live-session
account is in KNOWN_LIMITATIONS.md.
Over HTTP
export MAPLE_PROCURE_TOKENS="analyst:$(openssl rand -hex 16):read,pii"
make serve-http # HOST and PORT are overridablePOST /mcp is the MCP endpoint. GET /healthz, GET /readyz and GET /version need no
credential. A platform health check arrives without one, and a health endpoint that 401s marks
a deploy unhealthy forever.
/healthz and /readyz answer different questions, and the difference matters here.
/healthz is liveness and never opens the database, because this deploy rebuilds it at boot and
a check that failed during the rebuild would keep the service from ever coming up. The cost of
that is a blind spot: it still returns ok when an ingest has failed and every tool call is
about to return -32003. /readyz closes it: it reads each source's ingest provenance and
returns 503 with a reason when a source is missing or loaded no rows:
$ curl -s https://maple-procure.onrender.com/readyz | jq -c '{status, tenders: .sources.tenders.source_row_count}'
{"status":"ready","tenders":937}Leave /healthz as the platform health check and point external monitoring at /readyz.
Tokens are configured as space-separated subject:secret:scopes entries. The server refuses to
start if none are set, rather than serving 401 to everyone including its owner. subject is a
label for logs, not a credential.
MAPLE_PROCURE_CONFIRM_KEYS (space-separated, each ≥32 bytes) keeps export confirmations valid
across restarts. Leaving it unset is the safer default and what the deploy below does: a key is
then generated at startup, so a restart invalidates every outstanding handle. Persisting the key
without also sharing the spent-handle set does the opposite. A restart clears that set in
memory while already-issued handles stay cryptographically valid, opening a five-minute window
for a replay. Set a key only when spent handles are shared too.
The auth model, stated plainly
These are static bearer tokens, not OAuth. The spec's authorization model is OAuth 2.1 with this server as a resource server; there is no authorization server here, no rotation and no per-user identity. What is real is the boundary: a token carries scopes and they are enforced. This is a stand-in and should not be described as OAuth-compliant.
Refusals come at two layers, because they are different answers:
Meaning | Response | |
No | We do not know who you are |
|
Unrecognised token | That credential is wrong |
|
Token without | Known, but not for this endpoint |
|
Token without | Known, allowed here, not for that |
|
The last row is deliberate. Once a caller is authenticated, a per-tool refusal is returned inside the protocol so the model reads an error with a remedy and can explain it, rather than a transport failure it cannot interpret.
REST routes for pipelines
/mcp answers a model. GET /v1/tenders/search and GET /v1/tenders/{reference_number} answer a
pipeline: something that polls on a schedule, branches on what comes back, and cannot ask a
follow-up question.
They exist because JSON-RPC hides failure from an automation client. Over /mcp, a tool-level
error arrives inside an HTTP 200. A workflow engine reads that 200 as success, and a -32001 is
something it can neither branch on nor retry correctly. These routes return every failure as a
status, divided on the one question an automated caller asks: can retrying ever work?
Status | Meaning | Retry? |
| No credential, or one we do not know | No, fix the token |
| Known credential, insufficient scope | No, never |
| The arguments cannot be served as written | No, fix the request |
| The server is fine, its data is not. A cold start re-ingesting | Yes |
Both transports share one query layer. The routes call tenders.py, the module the MCP tools
call. They take the same arguments under the same names and return the same envelope, provenance
and notes included. Without a model reading the prose, a triage decision has to be explainable
from the payload alone. The bearer token is checked by the same auth.resolve on both surfaces, so
they cannot answer differently for the same credential.
The token is scoped the same way. read reaches both routes. include_contact=true also needs
pii, and is refused with 403 rather than -32002: the gate is the same scopes.require the
tool calls, and only the shape of the refusal differs, because a pipeline can branch on a status but
cannot read a remedy.
Two further differences are deliberate. region is refused with 422 rather than offered: the
filter drops every notice that states no region, which a pipeline has no reader to notice, and
ignoring the parameter silently would hide that the filter never ran. An unknown reference is
200 with empty rows, as the tool answers it, so the provenance still says which snapshot was
searched. The reasoning is in docs/DECISIONS.md.
Verified in production. Run against https://maple-procure.onrender.com on 2026-09-14 with a
read-only token. Commands and responses are copied verbatim; the token is shown as $TOKEN, and
... stands for the host and the token after the first call. The 503 is not shown: a warm
instance has its data.
$ curl -s "https://maple-procure.onrender.com/v1/tenders/search?keywords=software&limit=2" \
-H "Authorization: Bearer $TOKEN" | jq
{
"rows": [
{
"reference_number": "WS3971158100-Doc3971158168",
"title": "Software Licensing Supply Arrangement (SLSA) RFSA",
"buyer_name": "Department of Public Works and Government Services (PSPC)",
"closing_date": "2028-09-29T10:00:00",
"category": "Goods",
"regions_of_delivery": "Canada",
"notice_url": "https://canadabuys.canada.ca/en/tender-opportunities/tender-notice/WS3971158100-Doc3971158168"
},
{
"reference_number": "cb-901-3543961",
"title": "Spectrum Auction Verification Services for the mmWave Auction",
"buyer_name": "Department of Industry (ISED)",
"closing_date": "2026-09-21T14:00:00",
"category": "Services",
"regions_of_delivery": "National Capital Region (NCR); Canada",
"notice_url": "https://canadabuys.canada.ca/en/tender-opportunities/tender-notice/cb-901-3543961"
}
],
"row_count": 2,
"total_matches": 47,
"withheld": 45,
"truncated": true,
"as_of": "2026-09-14T00:19:57+00:00",
"source_file": "https://canadabuys.canada.ca/opendata/pub/openTenderNotice-ouvertAvisAppelOffres.csv",
"source_last_modified": "Sun, 13 Sep 2026 10:20:16 GMT",
"source_row_count": 901,
"coverage": {},
"notes": [
"Data as of 2026-09-14T00:19:57+00:00, from the last ingest of https://canadabuys.canada.ca/opendata/pub/openTenderNotice-ouvertAvisAppelOffres.csv — a snapshot, not a live feed. Newer notices may exist.",
"Showing 2 of 47 matches; 45 withheld. Narrow the search or raise `limit` (max 100) — do not describe these rows as the complete set."
]
}
$ curl -si ".../v1/tenders/search?keywords=software" | grep -i 'HTTP/\|www-authenticate'
HTTP/2 401
www-authenticate: Bearer error="invalid_request", error_description="an Authorization header is required"
$ curl -si ".../v1/tenders/WS3971158100-Doc3971158168?include_contact=true" -H "Authorization: Bearer ..." | grep -i 'HTTP/\|^{'
HTTP/2 403
{"error":"insufficient_scope","error_description":"this request needs the 'pii' scope, which this caller does not hold. Held scopes: read."}
$ curl -s -o /dev/null -w "%{http_code}\n" ".../v1/tenders/search?keywords=software®ion=Ontario" -H "Authorization: Bearer ..."
422
$ curl -s ".../v1/tenders/cb-000-00000000" -H "Authorization: Bearer ..." | jq -c '{row_count, as_of, notes: .notes[1]}'
{"row_count":0,"as_of":"2026-09-14T00:19:57+00:00","notes":"No such reference number in the open tender notices."}What gets logged
One JSON line per tool call, on stderr. MCP's own Logging feature is deprecated as of
2026-07-28 and the named migration is stderr and OpenTelemetry.
{"event": "tool_call", "tool": "get_tender", "subject": "analyst", "transport": "http",
"arguments": {"include_contact": true, "reference_number": "str(14)"},
"outcome": "ok", "rows": 1, "duration_ms": 0.41, "request_id": "3", "ts": "..."}Argument values are redacted by name, not by type. reference_number becomes str(14);
only names on an allowlist of closed-domain arguments (limit, group_by, currency,
include_contact, the dates) are recorded as themselves. The default is redaction, so a tool
added later is safe before anyone thinks about it, and users type contact details into search
boxes, because this dataset puts them in free-text descriptions.
include_contact is logged deliberately: it is a bool, so it discloses nothing, and it is the
one field a reviewer asking who pulled contact details needs. Rejected credentials are never
logged. Mistyping a hostname sends a valid token somewhere it shouldn't go, and that is exactly
how logs come to hold secrets.
Spans come free: the SDK installs OpenTelemetryMiddleware on every server and it already reads
traceparent from _meta. It is inert until you configure an exporter, and once you do, each
log line carries the trace_id that joins it to its span.
Deploying
render.yaml is a Render blueprint: connect this repo as a Blueprint in the
Render dashboard and it builds the service described there. The one thing it cannot carry is the
token list, which is set in the dashboard as MAPLE_PROCURE_TOKENS and never committed.
Two properties of the deploy are deliberate, and both are argued in the blueprint's own comments:
The database is rebuilt at boot, because Render's disk is ephemeral. A cold start downloads ~16 MB from CanadaBuys and loads it in about six seconds, which serves fresher data than a persisted volume would. The costs are a slow first request after a spin-down and a dependency on CanadaBuys being up at boot. Its WAF returns transient 403s to repeated bulk downloads, so a service that cold-starts constantly may occasionally fail to start.
No confirmation key is configured, so a restart invalidates outstanding export handles rather than leaving replayable ones behind. See the note above.
PORT comes from the platform and the Makefile defers to it; HOST is 0.0.0.0; and
MAPLE_PROCURE_HOST is filled from Render's own RENDER_EXTERNAL_HOSTNAME, because the SDK
only auto-enables DNS-rebinding protection for localhost.
Why the tool descriptions read the way they do
A tool description is the only thing a model sees when it decides which tool to call, so in
this repo they are treated as code and snapshot-tested
(tests/snapshots/tools.json). An accidental edit is a
reviewable diff, not a silent behaviour change.
Three problems shaped the current wording, and each is a limit of the data rather than of the code:
A model will summarise the rows in front of it and never mention the ones that are missing.
About 30% of award notices publish no dollar figure, so a "total spend" from this data is
always a floor. Reporting unvalued_rows: 810 next to the rows was not enough. A number in a
payload gets used for arithmetic, not repeated in prose. So every result also carries a notes
list of plain sentences ("These totals are a floor, not a complete figure… Say this when
reporting a total"), and both the tool description and the server instructions tell the model
to pass them on. Sentences get repeated; fields do not.
Two tools a human can't tell apart, a model can't either. search_tenders and
search_awards both take keywords and both return procurement notices. Each description
therefore names the other and says what it is not for: open bids versus completed
contracts, and summarize_spend and search_awards point at each other across the same
line, totals versus individual contracts.
The search is dumber than it looks. It is SQLite FTS5: all words must match, and there are
no synonyms. Left unsaid, a model responds to an empty result by adding more words, which can
only narrow it further. So the keywords description states that adding words narrows the
search and to try fewer, and that OR and NEAR are ordinary words here, not operators.
A half-finished flow reads like a bug worth working around. export_report returning a
preview instead of a file looks, to a model, like a step it should complete. So its description
opens by saying the first call never writes by design, and both the description and the
preview's own notes say to ask the user before sending the handle back. The mechanism blocks
a single-call write on its own; this wording is what stops the second call being automatic.
What a live session actually showed
First run against Claude Desktop, 2026-08-16. Two of the bets above were finally testable, and they did not both come out the same way.
The notes bet held. Unprompted and in its own prose, the model relayed that spend totals
were a floor rather than a total, that keyword coverage was limited, that the data was a
snapshot rather than a live feed, and that individual notices might be stale. Sentences in the
payload get repeated where numbers in a field do not. The wording stands as written.
The confirmation wording held only under ambiguity. Asked for "the three tenders closing this week", the model got a preview listing four, stopped, and asked which set was wanted before confirming, the instruction working exactly as intended. Told "export it again", it ran a fresh preview and passed the new handle back to itself in the same turn, writing without showing anyone the new preview.
That second behaviour is defensible given an explicit instruction, and it is the honest limit of
this design: on the handle path the checkpoint is a norm the model observes, not a wall it
cannot cross. Recorded rather than papered over in
KNOWN_LIMITATIONS.md, with the log evidence.
Built on
mcp2.0.0, targeting MCP spec revision2026-07-28SQLite with FTS5
FastAPI, hosting the streamable-HTTP transport and its auth boundary
Both stdio and streamable HTTP work. Notes on what the 2026-07-28 revision changed, and how
those changes shaped this design, are in docs/SPEC-NOTES.md.
Data
Tender and award notices published by Public Services and Procurement Canada via CanadaBuys, used under the Open Government Licence – Canada.
Contains information licensed under the Open Government Licence – Canada.
This project is not affiliated with or endorsed by the Government of Canada. Source files,
refresh cadence and the column map are documented in docs/DATA.md.
Licence
MIT. See LICENSE. The licence covers this software, not the government data it indexes; the terms for that data are in NOTICE.
Available Tools
5 toolsexport_reportA
Write a report of specific tender notices to a file on the server, as markdown or CSV.
This tool never writes on the first call, by design. Call it with tender_refs and it
returns a preview of the exact document plus a confirm_handle, having written nothing. To
actually write, call it again with identical arguments plus that handle — or answer the
confirmation prompt, on clients that show one. Tell the user what the preview contains and let
them decide; do not confirm on their behalf because the flow looks unfinished.
A handle lasts five minutes, works once, and is tied to the exact content previewed. If the notices or the format change, the old handle stops matching and a fresh preview is needed.
Requires the export scope. Reports contain notice summaries and links only — never buyer
contact details, whatever scopes the caller holds.
Inputs: tender_refs (required), format, confirm_handle. Returns: object
get_tenderA
Fetch one open tender notice in full by its exact reference number: complete description, dates, category, trade agreements and buyer.
Use after search_tenders has found a notice worth reading. This does not search — the
reference number must match exactly, and an unknown one returns no rows rather than an error.
Buyer contact details are personal data and are withheld by default. Set include_contact to
true only when the user has actually asked how to reach the buyer; the call is refused without
the pii scope.
Inputs: reference_number (required), include_contact. Returns: object
search_awardsA
Search awarded Canadian federal contracts — who won, for how much, when, and for which buyer. Use this for a supplier's track record, for what similar work has gone for, or for "who won X".
Not for opportunities still open: search_tenders covers notices accepting bids.
For ranked totals rather than individual contracts, use summarize_spend.
About 30% of award notices disclose no dollar figure. Those rows come back with award_value
null and are counted in coverage.unvalued_rows — that is a withheld amount, not a free
contract. Read notes and pass on what it says.
Inputs: limit, since, until, keywords, +1 more. Returns: object
search_tendersA
Search open Canadian federal tender notices — solicitations still accepting bids. Use this for "what can we bid on", "which contracts are coming up", or to find work in a field or region.
Not for contracts already awarded: search_awards covers who won what, for how much.
Results are ordered soonest-closing first, or by relevance when keywords is given. Cite the
notice_url when a row has one — about 5% are null, because legacy reference numbers do not
resolve on the current site and a null is preferred to a link that 404s. Quote the reference
number instead for those; do not construct a URL. Read the notes field and pass on what it
says; it states what this result does not cover.
Inputs: limit, region, category, keywords, +1 more. Returns: object
summarize_spendA
Rank total disclosed contract value by buyer, supplier or category. Use for "who spends most on X", "which suppliers earn the most", or comparing buyers over a period.
For the individual contracts behind a total, use search_awards.
Two things make every total a floor rather than a true total, and both are reported on the
result: awards that disclosed no amount are excluded rather than counted as zero, and only one
currency is summed per call because the source mixes CAD, USD and EUR with no conversion. The
notes field states both in words — repeat them when you report a figure. A total presented
as complete is wrong.
Inputs: since, top_n, until, currency, +1 more. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool has a distinct role: searching open tenders, fetching tender details, searching awarded contracts, aggregating spend, and exporting reports. No overlap between any pair.
All tools follow a consistent verb_noun pattern with snake_case: search_tenders, get_tender, search_awards, summarize_spend, export_report. The naming is uniform and predictable.
Five tools is well-scoped for a procurement data access server, covering search, detail retrieval, aggregation, and export without unnecessary redundancy.
The tool set provides complete coverage for the domain: search open tenders, get full tender details, search awards, summarize spend, and export reports. The workflow from discovery to export is fully supported.
Related MCP Connectors
Canada Government Procurement MCP — CanadaBuys open data (keyless).
MCP server for French (BOAMP) + EU (TED) public procurement data via TenderAPI.
全国の官公庁や自治体の入札・公募を、ClaudeやChatGPTから検索できる「入札公募検索」の公式MCPサーバーです。Search Japanese public tenders.
Government tenders, awards and pre-tender pipelines from 21 official sources as MCP tools.
Related MCP Servers
- AlicenseAqualityBmaintenanceThe most comprehensive keyless federal-data MCP server. 36 tools for SAM.gov + USAspending + Federal Register + eCFR + Grants.gov. No API key, no registration, no signup. Works in Claude Desktop, Claude Code, Codex CLI, Cursor, Continue, Gemini CLI, and any MCP-aware host.153235 npm17MIT
- AlicenseAqualityDmaintenanceMCP server for searching government tenders from CanadaBuys and SAM.gov with free stats and paid search, latest, and AI matching tools using x402 micropayments.41MIT
- AlicenseNot gradedqualityCmaintenanceMCP server for analyzing Canadian federal spending data, offering tools for contract search, NLP, semantic search, anomaly detection, and money-flow tracing.MIT
- FlicenseNot gradedqualityBmaintenanceMCP server for Canadian procurement intelligence, enabling unified search of federal and Alberta tender opportunities, deadline tracking, profile-based matching, daily briefs, and AI-assisted bid analysis.-