Browse the docs
Integrations

Coding agents (MCP)

Verity is an MCP server. Connect your coding agent once, then ask it to fix a report: it reads what the reporter said, the environment, the steps before the report, console errors, failed requests and the screenshot, and works from that evidence in your codebase.

Setup

  1. 1. In Verity, open Configure → Agent access. Owners and admins can create tokens.
  2. 2. Create a token and copy it. It is shown once.
  3. 3. Add Verity to your agent with the endpoint and the token.

Claude Code

claude mcp add --transport http verity https://api.feedback.mom/v1/agent/mcp \
  --header "Authorization: Bearer vrt_..."

Cursor and other MCP clients

{
  "mcpServers": {
    "verity": {
      "url": "https://api.feedback.mom/v1/agent/mcp",
      "headers": { "Authorization": "Bearer vrt_..." }
    }
  }
}

The Agent access page shows both snippets with your endpoint and new token filled in.

Tools

list_reports
Reports newest first, filtered by status (open by default), type, project or text in the message.
get_report
One report as a brief the agent can act on, with the screenshot attached as an image. It is the same brief as “Copy for agent” in the inbox.
update_report_status
Moves a report to new, triaged or closed, for example closed once the fix is merged. Only offered to tokens created with Allow status changes.

Try: “Fix the latest Verity bug”, or paste a report link from the inbox.

Access and privacy

  • Read-only by default. A token created with Allow status changes can also set a report’s status, and nothing else.
  • Reporter names and email addresses are never shared with agents.
  • Tokens belong to one organization, are stored only as a hash and can be revoked at any time.
  • Every report an agent reads, and every status it sets, is recorded in the audit log.

REST API

Scripts and CI jobs can use the same token with GET /v1/agent/reports and GET /v1/agent/reports/:id, and tokens allowed to change status can send PATCH /v1/agent/reports/:id with { "status": "closed" }. Requests are limited to 120 per minute per token.