The MCP server exposes read-only tools over the Regulatory Pack API. It is for agent hosts that need citation-backed regulatory facts without giving the agent direct database, S3, or pipeline access.Documentation Index
Fetch the complete documentation index at: https://docs.esheria.ai/llms.txt
Use this file to discover all available pages before exploring further.
Install
Host Configurations
Claude Desktop
Cursor
Codex MCP
Internal Agent Host
esheria-mcp, use an absolute path from pipx which esheria-mcp, uv tool dir, or the repository wrapper path.
Tool Catalog
| Tool | Purpose |
|---|---|
esheria_health | Liveness check |
esheria_ready | Runtime dependency readiness |
esheria_list_packs | Discover current packs |
esheria_get_pack | Inspect one pack |
esheria_list_obligations | Retrieve obligations by pack and filters |
esheria_check_applicability | Match obligations to an entity profile |
esheria_verify_claim | Verify a generated claim against published facts |
esheria_get_filing_calendar | Retrieve filing calendar items |
esheria_get_evidence_register | Retrieve evidence/control requirements |
esheria_list_relationships | Retrieve relationship facts |
esheria_query_regulatory_graph | Query cross-pack relationships |
esheria_check_graph_applicability | Run graph-aware applicability |
esheria_get_entity_profile | Summarize entity profile context |
esheria_export_pack | Export a published pack |
esheria_get_citation_context | Retrieve citation context from pack export |
First 5 Tool Calls
Run these after the host connects the server.trace_id; pack and workflow responses preserve readiness labels, limitations, citation IDs, and published-only results where the API returns them.
Agent Instructions
Agent prompts should require the model to:- call Esheria tools before making regulatory assertions
- preserve citation IDs, quote spans, readiness labels, limitations, and trace IDs
- avoid legal-advice framing
- avoid inferring unpublished obligations
- route unsupported or ambiguous answers to human review