bkmrk

# BKMRK — Bookmark Intelligence

You are connected to the user's BKMRK library. BKMRK analyzes bookmarks with Claude AI against the user's coding projects, scoring relevance and generating implementation suggestions.

**Terminology:** The pipeline statuses are `new` → `staged` → `done`. Always use "stage" / "staged" (never "queue" or "queued") when referring to items the user wants to act on next.

## What Gets Analyzed

BKMRK performs deep content extraction across all source types:

- **Tweets** — full tweet text plus all URLs in the tweet
- **X Articles** — full article body extracted via X API (not just the title)
- **Threads** — reconstructed thread text from all replies, plus URLs found in every tweet in the thread (not just the first)
- **YouTube videos** — full transcript extracted (auto-generated or manual captions), analyzed uncapped regardless of video length
- **Blog posts / news articles** — full article text extracted via trafilatura, analyzed uncapped
- **GitHub repos** — README and repo metadata
- **Any URL** — submitted via the API, fetched and extracted automatically

All content is sent to Claude uncapped for analysis — long articles, 2-hour podcast transcripts, and full X Article bodies all get deep, project-specific analysis.

## Authentication

All requests require the user's BKMRK API key as a header:

```
X-API-Key: {BKMRK_API_KEY}
```

The API key is available at https://bkmrkapp.com/settings under "Your API Key."

## What You Can Do

### Browse Library

Browse the analyzed bookmark library with filters. Returns scores, statuses, per-project analyses, and executable prompts. Use this for triaging, browsing, and pipeline management.

```
GET https://bkmrkapp.com/api/agent/library
X-API-Key: {BKMRK_API_KEY}
```

All query parameters are optional:
- `bookmark_id` — Fetch a specific bookmark by UUID (use to check status after submit)
- `status` — Filter by card status: `new`, `staged`, `done`, `trashed`
- `project_id` — Filter by project UUID
-...