> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://api.docs.siftsy.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://api.docs.siftsy.com/_mcp/server.

## 1.6.0

Routes for combing through a workspace, finding what stands out, and exporting it next to your own data. See [Exploring and exporting](/guides/core-concepts/exploring).

* **`POST /posts/match`.** Resolve up to 100 post links from a campaign tracker, creator list, or post-performance report to posts on your team. Each link comes back in order as `matched`, `processing`, `failed`, `not_in_workspace`, `invalid_url`, or `unsupported_platform`, and `meta.matchedPostIds` drops straight into `selection.postIds`. Nothing is added or charged.
* **`POST /posts/scan`.** One row per post across a selection, against the selection's comment-weighted baseline, with flags for negative sentiment, posts well below or above the baseline, commenters arguing with each other, off-topic threads, content-vs-topic gaps, and highly engaged negative posts. Sort, filter by flag, page, or take every row as CSV.
* **`POST /breakdowns`.** Roll a selection up by account, platform, content tag, content type, collection, week, or month — the same math as the dashboard's combined-by-account rows — with each group's share of the selection and its difference from it.
* **Comparisons.** `POST /analyses` takes `compare` — 2 to 4 labeled selections — instead of `selection`. Each side gets its own scope, content, and comment analysis; `comparison.deltas` gives each side against side 1. `GET /analyses/{id}/comments?side=N` samples one side.
* **Cited answers.** `answer: { question }` (or `answer: true` for an overview) writes the finding in words, with every claim citing real comments that resolve in `commentAnalysis.answer.citations`. Always a job, in the `analysis_search` rate class. If the writer fails, the numbers are still ready and `answer.error` explains.
* **Scores over Time.** `include: ["timeline"]` adds the dashboard chart's series over the matched comments, by day, week, or month, by comment date or post date, in your time zone.
* **Languages.** `filters.languages` narrows to comments that contain (`contains`) or are only in (`only`) the listed languages; `include: ["languages"]` adds a language breakdown.
* **`GET /analyses/{id}/export`.** Flat `posts`, `sides`, `timeline`, `languages`, and `citations` tables as JSON or CSV. The `posts` table carries the same join keys as scan and match rows. Every analysis links it as `export.url`.
* **Answers need `comments:read`.** Requesting `answer` without it is a `403`, and a key without it that reads an analysis gets citations as ids and post links only.
* **`GET /analyses?kind=comparison`** lists comparisons.
* New rate classes: `posts_scan`, `breakdowns`, `posts_match` (20 / minute, 500 / day each) and `analysis_export` (30 / minute, 500 / day). `GET /me` lists them under `rateLimit.routeClasses`.

## MCP (beta)

The Analyst API is also available as an MCP, in beta, at `POST /mcp` on `https://api.siftsy.app`. Same key, same scopes, same rate classes. Tool names and shapes can still change; `/analyst/v1` stays the stable warehouse contract. See [MCP (beta)](/guides/get-started/mcp).