Airside Labs Aviation Tools MCP¶
Remote MCP server (Streamable HTTP, OAuth 2.1 or API key) - a hosted reference service that gives an AI agent defensible answers to aviation reference questions. Every answer arrives with its evidence attached - confidence, alternates, temporal validity and per-field provenance - and a question the data cannot answer comes back as unanswerable rather than as a plausible guess. Built for the failure mode most aviation data trips on: codes are reused, names collide, registrations move between operators, and most reference data online carries no date and no source.
Server type: Remote (hosted, Streamable HTTP, protocol 2025-06-18)
Auth: OAuth 2.1 (primary), X-API-Key or Bearer (alternative), no credential for 3 free-tier tools
Endpoint: https://mcp.airsidelabs.com/mcp
Registry: com.airsidelabs/aviation-tools (Official MCP Registry, v1.1.0)
Docs: https://airsidelabs.com/tools/docs
Tools - 22 in Three Groups¶
Identity (six tools). resolve_airport, resolve_airline, resolve_aircraft_type, resolve_registration and parse_flight_identifier turn IATA/ICAO codes, names, tail numbers, Mode-S hex and flight designators into canonical entities. validate_identifiers checks a set of them against each other on a given date. Pass as_of for any historical question: codes are reused over time and an undated answer is silently wrong for past data.
Airport facts (five tools). airport_network_integration, airport_operator, airport_operations_status, airport_connectivity and airport_size_band answer what is true of an airport, each field cited to the public document that states it.
Use-case atlas (eight tools). use_case_landscape, search_use_cases, get_use_case, similar_use_cases, data_requirements, trace_data_lineage, trace_workflow and easa_ai_framework cover 6,700+ aviation AI use cases by organisation type and role, the data each needs and how fresh, the lineage from a requirement to the message that evidences it, 33 operational workflows, and a screen against the EASA AI-level and hazard framework.
Feedback (three tools). report_unmet_need, submit_suggestion, feedback_status let an agent tell the vendor what it could not get and see what happened to it.
Three tools (use_case_landscape, get_use_case, submit_suggestion) work without any credential, subject to a per-address daily cap.
Pricing¶
| Plan | Price | Calls / month | Keys |
|---|---|---|---|
| Explorer | $0 | 500 | 1 |
| Pro | $59 | 10,000 | 1 |
| Team | $249 | 50,000 | 5 |
Self-serve, monthly, USD. Over the cap, a call answers in-protocol with the reset time rather than failing silently.
Example Prompts¶
- "Which airline operated flight BA2668 on 14 March 2024, and what type flew it?"
- "Resolve G-XLEA and tell me who operates it as of today."
- "Is LGW integrated with the EUROCONTROL network, and who is the operator?"
- "What AI use cases exist for an MRO planning role, and what data do they need?"
- "Screen 'automated NOTAM triage' against the EASA AI framework."
Why It Matters for Operators¶
An agent working on an aviation problem - reconciling an MRO work order to the right airframe, checking a customer's booking against the operating carrier, scoping an AI project for an airport operations team - can produce a confident answer easily and a defensible one with difficulty. These tools are built around that failure mode: every field says where it came from and when it was valid, alternates are listed rather than silently dropped, and a question the evidence cannot settle returns "unresolved" instead of a plausible number.
Verification (Sep 10, 2026 night sweep)¶
Endpoint https://mcp.airsidelabs.com/mcp live-probed over JSON-RPC: keyless initialize answered with server info (Airside Labs Aviation Tools v1.1.0, matching the listing), and keyless tools/list captured all 22 tools exactly as documented. Fully live-verified.
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.