Developer access

Using Verachi MCP

Point your AI agent at meeting notes or a repo and bulk-extract decisions into a Verachi project.

UpdatedMay 7, 2026

On this page

MCP lets an AI agent work with Verachi records through a stateless HTTP endpoint.

Use it for record automation: decisions, projects, documents, guidelines, and risks. Do not use it for chat execution or workspace administration.

Before you start

You need:

A Verachi workspace.
An API key with the scopes your agent needs.
An MCP-compatible client.

Create the API key from Settings > Security & data > API keys.

Connect your client

Endpoint:

https://app.verachi.io/api/mcp

For Codex, add this to ~/.codex/config.toml:

[mcp_servers.verachi]
url = "https://app.verachi.io/api/mcp"
http_headers = { Authorization = "Bearer YOUR_API_KEY" }

For JSON-based clients:

{
  "mcpServers": {
    "verachi": {
      "url": "https://app.verachi.io/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Verify the connection

Ask your client:

Use auth_whoami to confirm the Verachi connection.

The response should show the tenant ID, user ID, and granted scopes.

Common workflow

Ask the agent to extract decisions from files and create draft records:

Read docs/meeting-notes/ and extract the decisions.

For each decision:
- write a clear title
- summarize what was decided and why
- cite the source file

Then use Verachi MCP to:
1. list projects and find "MY_PROJECT"
2. create each decision
3. link each decision to that project
4. report what was created

Review created records in Verachi before publishing or relying on them.

Tools by area

Identity

Tool	Purpose
`auth_whoami`	Check tenant, user, and scopes

Decisions

Tool	Purpose
`list_decisions`	Search and filter decisions
`search_decisions_picker`	Lightweight decision picker search
`get_decision`	Read one decision and linked artifacts
`create_decision`	Create a decision
`update_decision`	Update mutable decision fields
`delete_decision`	Delete a decision
`publish_decision`	Publish a draft
`archive_decision`	Archive a decision
`unarchive_decision`	Restore an archived decision
`supersede_decision`	Mark one decision as replaced by another
`link_decision`	Link external evidence
`get_decisions_graph`	Fetch a workspace decision graph
`get_decision_graph`	Fetch a focused decision graph
`list_decision_projects`	List project links
`update_decision_projects`	Replace project links
`list_decision_comments`	List comments
`create_decision_comment`	Add a comment or review action
`update_decision_comment`	Edit your comment
`delete_decision_comment`	Delete your comment
`list_decision_documents`	List document links
`link_decision_document`	Link a document URL or ID
`unlink_decision_document`	Remove a document link
`list_decision_attachments`	List uploaded file attachments
`list_pipeline_runs`	Inspect similarity pipeline runs

Projects

Tool	Purpose
`list_projects`	List visible projects
`get_project`	Read one project
`create_project`	Create a project
`update_project`	Update project fields
`delete_project`	Soft-delete a project
`archive_project`	Archive a project
`unarchive_project`	Restore an archived project
`list_project_teams`	List team assignments
`assign_project_team`	Assign a team
`remove_project_team`	Remove a team assignment
`list_project_dependencies`	List upstream dependencies
`create_project_dependency`	Add a project dependency
`delete_project_dependency`	Remove a dependency

Documents

Tool	Purpose
`list_documents`	List documents, optionally by attachment
`get_document`	Read metadata and optional extracted text
`create_document_from_text`	Create a text document, optionally attached to a record
`link_document`	Attach a document to a project or decision
`unlink_document`	Remove a document attachment

Guidelines

Tool	Purpose
`list_guideline_sets`	List guideline sets
`get_guideline_set`	Read a set, checks, and assignments
`create_guideline_set`	Create a custom set
`update_guideline_set`	Update metadata or checks
`list_project_guidelines`	List project assignments
`assign_guideline_set_to_project`	Assign a set to a project
`update_guideline_assignment_status`	Pause, resume, or archive an assignment

Risks

Tool	Purpose
`list_risks`	List risks with filters
`get_risk`	Read one risk and evidence
`create_risk`	Create a manual risk
`append_risk_evidence`	Add evidence to a risk
`update_risk_state`	Mark mitigated, snooze, mark not a risk, or reopen; requires a note, and snooze requires a date

Not exposed through MCP

Use the app or REST API for:

chat and agent execution
integration connection setup
integration rules
audit trail administration
notifications
AI settings
organization, team, and user management
API-key management