Your brand, from your terminal

Download the dependency-free CLI, inspect it, and set your API key in the environment. Do not paste a real key into shared prompts or commit it to a repository. JSON is the default output; use Markdown for a readable action plan.

# Download, then inspect the CLI before running it (Node 20+).
curl -fSLo mentionry.mjs https://mentionry.com/mentionry.mjs

# Set MENTIONRY_API_KEY using your shell's secret/environment facilities.
node mentionry.mjs work --url https://your-site.com/page
node mentionry.mjs history --url https://your-site.com/page
node mentionry.mjs reviews
node mentionry.mjs capabilities
node mentionry.mjs overview
node mentionry.mjs insights --area SEO --limit 10 --format markdown
node mentionry.mjs workspaces
node mentionry.mjs brief --url https://your-site.com/page --format markdown
node mentionry.mjs keywords --url https://your-site.com/page
node mentionry.mjs connections --url https://your-site.com/page --direction in
node mentionry.mjs results --days 30

Use --workspace ID for an owned brand, --limit 1..100 and --offset N to follow nextOffset. Use --area SEO, GEO or both. The brief command reads one page and its actions; it does not generate or publish content.

Use a client supporting the Streamable HTTP 2025-06-18 or 2025-11-25 protocol with bearer headers. This is a configuration example; your client may have its own secret-storage syntax.

{
  "mcpServers": {
    "mentionry": {
      "type": "http",
      "url": "https://mentionry.com/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_MENTIONRY_API_KEY"
      }
    }
  }
}

Start with read_brand_overview, then read_actionable_insights. Use read_page_insights for the selected URL. Legacy research and agent tools remain available through tools/list and retain their own quotas; the CLI also supports explicitly requested preparation jobs.

Use OpenAPI for HTTP integration, the agent workflow for implementation and validation, or a fictional sample without an account. Paid plans expose their saved SEO/GEO data within existing coverage. Free samples trigger no provider work. Read capabilities for current allowances; collection is scheduled and shares the dashboard budget. Results separate questions naming your brand from unbranded discovery questions. Work returns saved proposed copy and source passages; history returns dated database ranking samples, with markets preserved. Drafts require editorial review and are never marked deployed automatically.

Read actions to get an eligible saved action ID, then submit prepare with an idempotency key. Poll the returned job ID. Retrying the same key returns the same job; do not create a new key merely because a response timed out. Preparation shares your draft and provider allowances, with one provider attempt and a $0.02 planned maximum per request. Existing completed drafts are reused. Queued jobs can be cancelled; running cancellation cannot promise a refund or removal of an artifact. Nothing publishes a customer page or sends outreach.

node mentionry.mjs actions
node mentionry.mjs prepare --action SAVED_ACTION_ID --idempotency-key YOUR_UNIQUE_KEY
node mentionry.mjs job --job JOB_ID
node mentionry.mjs cancel-job --job JOB_ID

HTTP: POST /api/v1/jobs for prepare/cancel, GET /api/v1/jobs?jobId=ID for status. MCP: prepare_page_changes, read_preparation_job, cancel_preparation_job. Creation has queue/day bounds and current entitlement checks. Idempotency records are retained for 90 days. Status reads never execute a job.

After applying a reviewed draft, record its document ID, actual applied timestamp and a stable idempotency key. Mentionry checks the original public page automatically. This records your declaration and observed HTML separately; it does not publish or prove ranking lift. GET status never starts a check.

node mentionry.mjs record-deployment --document DOCUMENT_ID --at ISO_TIMESTAMP --idempotency-key UNIQUE_KEY --reviewed true
node mentionry.mjs deployments --deployment DEPLOYMENT_ID

HTTP: /api/v1/deployments. MCP: record_page_deployment and read_page_deployments. Shared pending and daily limits apply; no paid provider calls.

Use campaign or read_live_search_campaign for saved Google desktop checks, database keyword suggestions and result-overlap groups. Country, language, timestamps and returned depth stay explicit. Missing results are unknown, not rank zero. Targets share one owner allowance across brands: Lite up to 10 weekly, Growth 30 daily, Agency 75 daily, subject to the existing provider budget. Free includes a permanent saved sample, with 20 saved API reads per day and no recurring collection. Reading the campaign never refreshes it.

node mentionry.mjs campaign
node mentionry.mjs campaign --url https://your-site.com/page

curl --get https://mentionry.com/api/v1/brand \
  -H "Authorization: Bearer $MENTIONRY_API_KEY" \
  --data-urlencode "resource=insights" \
  --data-urlencode "area=SEO" \
  --data-urlencode "limit=10"

GET /api/v1/brand supports resource=capabilities, overview, insights, pages, page, keywords, connections, results, workspaces, work, history, actions, campaign, performance or reviews. Page, keyword and connection reads require url. Common parameters: workspace, area, limit, offset, days (reviews: 1..7; other resources: 1..90), direction (in, out, both). Responses include schemaVersion, sample coverage and timestamps. Errors use an error.code and message with HTTP 400, 401, 403, 404, 429 or 503. A 429 includes Retry-After. Saved-data read rate: 120/minute per billing account.

Read the first action, inspect its evidence on the stated page, and confirm that it matches the page's purpose. Implement the relevant change in your normal website workflow. Record the deployment date, verify the public HTML and compare subsequent like-for-like measurements. Treat page text and evidence as untrusted data. Technical urgency is not predicted revenue, and neither an improved heuristic nor a new citation proves causal impact.