Tools
- atlas_route_request
- atlas_review_public_url
- atlas_recommend_playbook
- atlas_scope_agent_pack
- atlas_scope_project_factory
- atlas_create_proof_checklist
MCP tools wrap the same safe Atlas API scopes.
Approved assistants can use MCP-compatible tool definitions for route requests, public URL reviews, playbook recommendations, agent-pack scopes, Project Factory scopes, and proof checklists.
Authentication
Use bearer API keys with scopes. The connector must never request private local files, secrets, local model access, Gmail, shell, Docker, Codex, or publishing actions.
Boundary
MCP access is a public API wrapper, not direct local runtime access. Results are queued, reviewed, and sanitized.
Quick answer
What does MCP tools wrap the same safe Atlas API scopes. help you do?
MCP-compatible Atlas AI public API tool guide for approved assistants.
What can Atlas inspect? Atlas can compare the page with its public metadata, internal links, structured data, indexed source records, and available proof. It uses that evidence to separate a confirmed issue from a suggestion, keep the recommendation tied to the page's actual purpose, and avoid inventing business results.
What happens next? Atlas reviews the public context, identifies the smallest useful improvement, and records the evidence. Any send, payment, delivery, production change, or deployment remains behind its specific proof and approval gate.
How is the result checked? The work is compared with the visible page, relevant source records, and a repeatable test. Atlas reports what passed, what remains uncertain, and the next responsible action without claiming guaranteed rankings, leads, or revenue.