Contextra MCP Server

Contextra

Contextra is an AI-powered sales assistance platform for managing targets (prospects), products, and pitches. The integration authenticates with email and password (Bearer token exchange on each call). Upload targets with LinkedIn profiles for background enrichment, pair existing products with targets to generate pitches (ChatGPT or Gemini), and read pipeline state through paginated listings and single-entity fetches.

Data model

• Targets — prospects with name, company, location, LinkedIn URL, optional email, and generatedData from enrichment.
• Products — catalog items in the account; batch pitch creation references products by name (resolved server-side to IDs).
• Pitches — generated outreach tied to a product and target; generatedData.status progresses QUEUED → GENERATING → GENERATED (or FAILED / NOT_GENERATED).

Typical workflow

1. Ingest prospects — batch upload targets (name, company, location, linkedInUrl required; email optional). Duplicates by name or LinkedIn URL land in the failed array; successful rows start enrichment asynchronously.
2. Review targets — paginated target listing (sort by name or createdAt); fetch one target by ID for full generatedData.
3. Create pitches — batch upload pitches with productName and targetName matching existing account entities (max 1,000 per request). Optional model (chat-gpt or gemini) and per-row or bulk send flags. Generation runs in the background.
4. Review pitches — paginated pitch listing supports search by pitch name; fetch by ID for full body, subject, and generation status.

Pagination and IDs

List endpoints use page (1-indexed) and page_size (1–100). Response payloads include pagination metadata (totalPages, totalItems on targets). Use listing pages to discover IDs before single-entity reads.

Gotchas

• Rate limit: platform throttles to 60 calls/minute with ~1s spacing between calls.
• Async generation: pitch content may not be ready immediately after batch upload — poll single pitch reads until generatedData.status is GENERATED or FAILED.
• Name resolution: pitch batch upload fails rows when product or target names do not match an existing entity or when the product+target pair is already pitched.