1| 2| 3| 4| 5| 6| MapleSpike — AI Agent Setup: ChatGPT, Claude, Gemini, Grok Integration 7| 8| 9| 10| 11| 12| 13| 14| 15| 16| 17| 18| 19| 20| 21| 22| 23| 24| 25| 26| 27| 28| 29| 44| 45| 46| 47| 48| 49| 50| 51| 52| 53| 54| 55| 56| 57| 75| 76| 77|
78| 79|
80| [00] 81| AI Agent Setup 82|
83| 84|

Connect MapleSpike to your AI tools

85|

86| MapleSpike speaks the MCP protocol — the standard that lets AI tools fetch live data. Add it once and your AI gains 27 specialized agents with 79 actions to query Canadian government data in real time. 87|

88| 89| 90|
91|

Things you can ask once it's connected

92|
93|
94| 🗳️ 95|

"What did the ethics committee discuss yesterday?"

96|
97|
98| 💰 99|

"Which companies lobbied the defence department this month?"

100|
101|
102| 🔗 103|

"Show me everything about Catherine Tait across all sources"

104|
105|
106| 🏛️ 107|

"Which meetings this week were in-camera (closed to the public)?"

108|
109|
110| 📄 111|

"Get me the full transcript from last week's finance committee"

112|
113|
114| ⚠️ 115|

"Flag any sole-source contracts over $1M this quarter"

116|
117|
118| 🔒 119|

"Which committee meetings went in-camera (closed to the public) this week?"

120|
121|
122|
123| 124| 125|
126|

Available Agents

127|

MapleSpike exposes 27 agents with 79 actions over MCP. Each agent specializes in a specific area of Canadian government data. Here are the key agents:

128| 129| 130| 131| 132| 133| 134| 135| 136| 137| 138| 139| 140| 141| 142| 143| 144| 145| 146| 147| 148| 149| 150| 151| 152| 153| 154| 155| 156| 157| 158| 159| 160| 161| 162| 163| 164| 165|
AgentDescriptionKey Actions
CommitteesParliamentary committee meetings, evidence, Hansard transcripts, witness lists, in-camera flagssearch_transcripts, get_committee, get_in_camera
LobbyingLobbying registry, monthly communications reports, registrations, spendingsearch_lobbying, get_lobbyist, track_organization
ProcurementGovernment contracts, CanadaBuys RSS, proactive disclosure, sole-source flaggingget_contract, search_procurement, flag_sole_source
ElectionsElections Canada contributions, candidate profiles, electoral district datasearch_contributions, get_candidate, get_district
InfluenceThink tanks, NGOs, IOs, consultancies — funding sources, committee appearances, lobbying connectionstrack_entity, get_signals, cross_reference
LegalFederal court judgments, tribunal decisions, regulatory registriessearch_judgments, get_regulation, search_decisions
Cross-Reference Pro+Entity resolution engine — links people and organizations across committees, lobbying, procurement, and all other sourcesresolve_entity, track_entity, cross_reference
166|

All agents are available on every plan via MCP. The Cross-Reference agent's entity resolution engine requires a Pro+ subscription.

167|
168| 169| 170|

Setup by platform

171| 172| 173|
174|
175| 💬 176|

ChatGPT

177|
178|

OpenAI's ChatGPT supports MCP through the ChatGPT desktop app and GPT Actions. Add MapleSpike as an MCP server to let ChatGPT answer questions about Canadian government data.

179| 180|

Desktop App (macOS/Windows)

181|

Configure in ~/.chatgpt/mcp.json. You can connect via the SSE endpoint (recommended) or run the MCP server from source:

182|
183|{ 184| "mcpServers": { 185| "maplespike-canada": { 186| "url": "https://maplespike.lan/mcp" 187| } 188| } 189|} 190|
191|

Note: The MCP server runs as a service on your K3s cluster. Ask your cluster admin for the endpoint URL and API key. To run locally, build from the repo: pnpm --filter @maplespike/mcp-server start

192| 193|

Or use the ChatGPT Actions API

194|

Point a GPT Action to MapleSpike's MCP endpoint at https://maplespike.lan/mcp (local) or the hosted MCP server endpoint.

195|
196| 197| 198|
199|
200| 🤖 201|

Claude (Anthropic)

202|
203|

Claude was the first major AI to adopt MCP natively. Both Claude Desktop and Claude Code support MapleSpike out of the box.

204| 205|

Claude Desktop

206|

Add to claude_desktop_config.json. Connect via the SSE endpoint (recommended):

207|
208|{ 209| "mcpServers": { 210| "maplespike-canada": { 211| "url": "https://maplespike.lan/mcp" 212| } 213| } 214|} 215|
216|

The MCP server runs as a cluster service. Ask your cluster admin for the endpoint URL. The API key is managed by the operator — set MAPLESPIKE_API_KEY in the server environment.

217| 218|

Claude Code (CLI)

219|
220|claude mcp add maplespike-canada -- https://maplespike.lan/mcp 221|
222| 223|

Try asking: "Using the MapleSpike MCP tools, find me all lobbyist meetings with Innovation department this quarter and check if any of those companies also have sole-source contracts."

224|
225| 226| 227|
228|
229| 🔮 230|

Google Gemini

231|
232|

Gemini supports MCP through Google AI Studio and the Gemini API. Use the MCP client library to connect MapleSpike as a tool.

233| 234|

Via Google AI Studio

235|

Use the built-in MCP client in AI Studio to add MapleSpike as a tool. Configure with:

236|
237|MCP Server URL: https://maplespike.lan/mcp 238|Authentication: Bearer token (your MapleSpike API key) 239|
240| 241|

Via Gemini API (Vertex AI / custom apps)

242|
243|# Python example 244|from google import genai 245|from mcp import Client 246| 247|client = Client("https://maplespike.lan/mcp") 248|tools = await client.list_tools() 249| 250|response = client.generate( 251| model="gemini-2.5-pro", 252| tools=tools, 253| prompt="What happened in committees this week?" 254|) 255|
256| 257|

Try asking: "Using MapleSpike, list all meetings of the Standing Committee on Public Accounts from the last month and summarize the main topics."

258|
259| 260| 261|
262|
263| 𝕏 264|

Grok (xAI)

265|
266|

Grok supports MCP tools through xAI's API. Connect MapleSpike to enable Grok to answer questions about Canadian government data with verified sources.

267| 268|

Via xAI API

269|
270|# Python example 271|from openai import OpenAI 272|import json 273| 274|client = OpenAI( 275| api_key="", 276| base_url="https://api.x.ai/v1" 277|) 278| 279|# MapleSpike MCP tools are called as function calls 280|response = client.chat.completions.create( 281| model="grok-3", 282| messages=[{"role": "user", "content": "Who lobbied the most this month?"}], 283| tools=[{ 284| "type": "function", 285| "function": { 286| "name": "query_gov_data", 287| "description": "Search Canadian government data" 288| } 289| }] 290|) 291|
292| 293|

Try asking: "Using MapleSpike, check which telecom CEOs have appeared before parliamentary committees this year and cross-reference with their company's lobbying activity."

294|
295| 296| 297|
298|
299| ⚙️ 300|

Your Own Pipelines

301|
302|

MapleSpike isn't just for AI chat — it's a full data pipeline that runs on your own infrastructure. Every ingestion module is also available as a local tool or script.

303| 304|
305|
306|

🏛️ Committees Pipeline

307|

Ingests all 30 House of Commons committees daily. Captures evidence XML, Hansard transcripts, speaker attribution, and in-camera flags.

308|
pnpm --filter @maplespike/pipeline-core committees --ingest-all
309|
310|
311|

💰 Lobbying Pipeline

312|

Monthly CSV dumps from the Office of the Commissioner of Lobbying. Tracks who is lobbying which department, on what topic, and how much they're spending.

313|
pnpm --filter @maplespike/pipeline-core lobbying --ingest
314|
315|
316|

📄 Procurement Pipeline

317|

CanadaBuys RSS feed + proactive disclosure contracts. Flags sole-source awards and cross-references with lobbying data.

318|
pnpm --filter @maplespike/pipeline-core procurement --check-sole-source
319|
320|
321|

🌐 Influence Pipeline

322|

32 tracked IOs, NGOs, think tanks, and consultancies. Cross-references committee appearances, lobbying, funding, and CRA charity data.

323|
pnpm --filter @maplespike/pipeline-core influence --ingest-all
324|
325|
326| 327|
328|

🔗 Entity Resolution Engine

329|

The engine package links people and organizations across all data sources. When the same person appears on a committee, in a lobbying report, and as a contract signatory — MapleSpike connects them. Runs 7 signal detectors for revolving doors, lobbying overlaps, and sole-source flagging.

330|
pnpm --filter @maplespike/engine sweep --full
331|
332| 333|

334| 💡 All pipelines run on your own K3s cluster via cron jobs. No external dependencies, no cloud, no data leaves your infrastructure. See How it Works for deployment. 335|

336|
337| 338| 339|

All MapleSpike MCP Tools

340|

These tools become available in your AI once connected. Each returns verified data with source citations.

341| 342| 343| 344| 345| 346| 347| 348| 349| 350| 351| 352| 353| 354| 355|
ToolWhat it doesBest for
query_gov_dataSearch all government data sources by keyword, jurisdiction, dateGeneral research LIVE
fact_checkVerify a claim against official dataJournalism LIVE
get_citationVerify a SHA-256 citation hashSource verification LIVE
latest_releasesLatest government data releasesMonitoring LIVE
search_transcriptsSearch committee evidence and Hansard transcriptsCommittee monitoring PLANNED
get_committeeGet details of a specific committee meetingDeep dives PLANNED
search_lobbyingSearch lobbying registry by company, department, dateInfluence tracking PLANNED
get_contractLook up procurement contracts by vendor or departmentSpending analysis PLANNED
track_entityFind every mention of a person or org across all sourcesCross-referencing PLANNED
get_in_cameraList all in-camera (closed) committee meetingsTransparency tracking PLANNED
get_signalsGet recent detected signals (revolving doors, overlaps, etc.)Alerting PLANNED
356| 357|
358| 359| 401| 402| 403| 404| 405| 406|