Atlasly MCP Tools: Public Discovery and Site-Research Limits
A dated reference to Atlasly MCP public tool discovery, practical tool groups, input schemas, source coverage and account-linked actions.

Quick answer
Atlasly MCP at https://mcp.atlasly.app/mcp returned 27 public tool definitions through tools/list on 12 September 2026. Use that live list for current names and input schemas. Five additional account-linked operations exist in source but were not verified in an authorized live session. Discovery confirms a tool is advertised; it does not prove every source, jurisdiction or analysis result.
This is a dated discovery reference for Atlasly at atlasly.app, the UK-focused site-analysis service. It is not a complete accuracy benchmark or a guarantee of US planning coverage.
Public initialization and tools/list returned successfully during the review. No site-analysis tool was called and no private project was created. The table below groups the actual public names so a team can choose what to inspect next. For connection permissions, read the authentication guide.
Find the schema before supplying a site
Initialize the MCP connection, then request tools/list. Initialization is not the tool manifest. Use each returned inputSchema to find required fields, permitted values and limits; do not guess a radius or copy a partial JSON fragment from an older article.
The current source adds an optional site_ref to many tools so a returned site reference can carry location context into the next check. Preserve that reference where returned, inspect its boundary confidence, and avoid conflicting address or coordinate overrides. A resolved address point is not a surveyed ownership boundary.
Public tool groups observed on 12 September 2026
| Purpose | Names returned by public discovery |
|---|---|
| Discovery and source readiness | search_tools; get_lpa_adapter_status; get_official_source_adapters |
| Site screening and comparison | site_verdict; compare_sites; boundary_doctor; analyze_site; find_development_sites |
| Planning research | check_planning_constraints; check_permitted_development; get_local_plan_policies; get_planning_history; find_precedents; assess_planning_approval_likelihood |
| Environmental and heritage research | get_flood_risk; calculate_biodiversity_net_gain; run_sequential_test; get_heritage_impact_table |
| Costs and regulatory scoping | estimate_planning_costs; check_building_regulations_triggers |
| Briefing, maps and research support | draft_site_description; generate_site_map; search_aec_knowledge; plan_site_intelligence; get_map_layer_plan |
| Terrain and movement | get_site_topography; check_walkability_transport |
This is the observed public list, not a promise that the list will remain unchanged or that all advertised operations have passed a live result check.
Interpret the output before using it
Read the returned status, errors, warnings, source links and coverage notes as well as any summary. Keep the requested location and study boundary with the result. A missing dataset or failed upstream request must remain unknown, rather than becoming evidence that a constraint is absent.
The get_site_topography implementation retrieves point elevation; it does not supply surveyed contours, slope profiles or a full terrain model. Planning likelihood, biodiversity estimates, permitted-development checks and sequential-test support are screening aids. Their names do not establish a planning decision, statutory metric assessment, lawful-development certificate or professional sign-off.
Many relevant public datasets are England-specific. Confirm the applicable nation, authority, policy edition and source for each check. The methodology and limits guide explains the next level of verification.
Account-linked operations need a separate check
Source defines five additional operations behind account permissions:
- create_project_from_site saves a project.
- run_site_package starts an account-linked analysis job.
- get_job_status reads a job’s status.
- generate_site_brief creates a project brief.
- enable_site_watch configures recurring site checks.
These were absent from the unauthenticated tools/list response. Their current authorized discovery, account allowances and end-to-end behavior remain unverified in this review. Do not assume that calling a missing name automatically signs a user in. Authorize through the client flow, discover again, and inspect the schema before requesting an action. Do not repeat a write after a timeout without checking whether it already succeeded.
A useful research sequence
For an illustrative England feasibility review, first establish the site boundary and authority. Inspect source readiness, then select the available planning, flood, heritage and transport checks that answer the brief. Keep each material result beside its primary-source reference and record unavailable evidence.
Only compare two sites after confirming comparable study areas and source coverage. Resolve findings that could change the brief through current official data and the appropriate specialist. This proposed sequence is not a tested claim that every listed operation returns complete evidence for every site.
Versioning, logging and publication limits
The reviewed session negotiated MCP protocol 2025-03-26. Protocol negotiation does not freeze Atlasly’s tool catalog or validate every client. Re-discover after reconnecting or receiving a tool-list change notice, and use the current schema.
Source contains tool-call logging that can include site addresses and coordinates. This article does not promise anonymous telemetry, no personal data, or a 30-day retention period. Review actual client disclosures and terms before sending sensitive material.
The public REST hostname previously used in examples did not resolve during the same review. See developer documentation for its publication status; a reachable MCP endpoint does not verify a separate REST API.
Illustrative example
Illustrative evidence register: record discovery as successful, each uncalled assessment as not tested, and each unavailable source as unknown. This separates connectivity from the evidence needed for a project decision; it is not a reported customer deployment.
Frequently asked
How many tools were verified?
Public tools/list returned 27 definitions on 12 September 2026. Their presence was verified, but their analysis outputs were not executed or benchmarked in this review.
Does initialize return the complete tool list?
No. Initialize negotiates connection capabilities and protocol. Request tools/list for the tool names and input schemas visible to the current session.
Does get_site_topography return contours?
The reviewed implementation provides point elevation, not surveyed contours, slope profiles or a complete terrain surface.
Are the private tools available anonymously?
They were absent from unauthenticated discovery. Five account-linked operations exist in source; verify permissions and discover their current schemas after an authorized connection.
Can I assume full UK or US coverage?
No. Confirm source availability and jurisdiction for each check. Much planning and environmental evidence is England-specific, and US planning coverage was not established.
Conclusion
Use live discovery to choose a tool, then inspect its returned evidence and limitations before relying on the finding. Tool availability, successful execution and a defensible project conclusion are separate checks.

About the author
Shatakshi Patil
Architect writing about pre-construction due diligence, planning context, and site intelligence workflows for design teams using Atlasly.
Sources and references
Authoritative references for the planning policies, regulations, and standards referenced in this article. Always check the publisher for the latest version.
Related articles
Auto-suggested
Atlasly vs PlanningBot
Atlasly vs PlanningBot: Compare Evidence, Reports and Site Workflows
A dated, source-backed comparison of Atlasly and PlanningBot: published capabilities, coverage checks, report requirements and how to evaluate both on the same site.
Read
how to use Atlasly in Claude ChatGPT
How To Use Atlasly Inside Claude, ChatGPT, and Other AI Assistants
Connect Atlasly’s hosted MCP endpoint in a supported Claude or ChatGPT client, verify discovered tools and distinguish public checks from account-linked operations.
Read
export DXF DWG from AI chat
Export Site Analysis to DXF: A CAD Handover Checklist
How to request and verify a site-analysis CAD export, including actual file formats, units, coordinates, layer coverage and checks before AutoCAD or Revit coordination.
Read
AI UK planning constraints NPPF
How To Ask AI For UK Planning Constraints: NPPF, Conservation Areas, Flood Zones, And Getting It Right
Use Atlasly to organise early UK planning questions, then verify current policy editions, source records, coverage and missing information before relying on an answer.
Read