API Reference

Programmatic access to stories, feeds, and the headline debrief tool. All endpoints return JSON and are served at ....

Quick Start

The POST /api/debrief endpoint requires an API key. Pass it via the X-API-Key header or the api_key field in the request body:

Send a debrief request. Pass your key via header or in the request body:

# Option A — X-API-Key header
curl -X POST /api/debrief \
  -H "Content-Type: application/json" \
  -H "X-API-Key: pd_your_key_here" \
  -d '{"headline": "You Won't BELIEVE What Happened!"}'

# Option B — api_key in JSON body
curl -X POST /api/debrief \
  -H "Content-Type: application/json" \
  -d '{"headline": "You Won't BELIEVE What Happened!", "api_key": "pd_your_key_here"}'

Save the raw key immediately — it's returned only once at creation. After that, only the prefix (e.g. pd_aBcDeFg...) is shown.

Request an API Key

Submit a request for an API key. Requests are reviewed manually — you'll receive your key once approved.

Authentication & Rate Limits

The debrief endpoint requires an API key. All other read endpoints are public.

Auth Type	Method	Used By	Rate Limited
API Key	X-API-Key: pd_...or body: {"api_key":"pd_..."}	POST /api/debrief	200 req/min
None	—	Stories, Feeds, Health	No

Rate limiting: Each API key allows 200 requests per minute using a sliding 60-second window. Exceeding the limit returns 429 Too Many Requests. The window slides continuously — no need to wait a full minute for it to reset.

Debrief

GET/api/debrief/prompt

Get the default system prompt used for headline analysis. Useful for reference before overriding with a custom prompt.

Response (200)

{
  "system_prompt": "You are a media literacy analyst. Identify every manipulative tactic used, could be a varying number. Generate exactly 5 alternative neutral headlines (different factual angles, 5-15 words each). Output JSON: {\"variations\": [...], \"clickbait_tactics_found\": [...]}"
}

cURL example

curl "/api/debrief/prompt"

POST/api/debrief

Submit a clickbait headline and receive neutralized, factual alternatives plus an analysis of the manipulation tactics used.

Authentication required: Pass your API key via the X-API-Key header or the api_key field in the JSON body. Rate limited to 200 req/min per key. Header takes priority if both are provided.

Parameters

Name	Type	Required	Description
headline	string	Yes	The clickbait headline to analyze (5-500 characters)
system_prompt	string	No	Custom system prompt (max 5000 chars). Overrides default; bypasses cache.
num_variations	int	No	Number of neutral variations to generate, 1-5 (default: 5)
api_key	string	No	API key (alternative to X-API-Key header). Useful when headers can't be set.

Request body (JSON)

{
  "headline": "You Won't BELIEVE What This Senator Just Did!",
  "system_prompt": "(optional) Custom system prompt override",
  "num_variations": 5,
  "api_key": "(optional) pd_your_key_here — alternative to X-API-Key header"
}

Response (200)

{
  "original": "You Won't BELIEVE What This Senator Just Did!",
  "variations": [
    "Senator [Name] Introduces New Policy on [Topic]",
    "Senate Committee Reviews Proposed Legislation",
    "Senator Takes Action on [Policy Area]",
    "New Senate Proposal Addresses [Issue]",
    "Senator Announces Legislative Initiative"
  ],
  "analysis": {
    "clickbait_tactics_found": [
      "Curiosity gap / vague cliffhanger",
      "ALL CAPS for emotional emphasis",
      "Second-person address to create urgency"
    ]
  }
}

cURL examples

# With X-API-Key header
curl -X POST /api/debrief \
  -H "Content-Type: application/json" \
  -H "X-API-Key: pd_your_key_here" \
  -d '{"headline": "You Won't BELIEVE What This Senator Just Did!"}'

# With api_key in request body (no header needed)
curl -X POST /api/debrief \
  -H "Content-Type: application/json" \
  -d '{"headline": "You Won't BELIEVE What This Senator Just Did!", "api_key": "pd_your_key_here"}'

Error responses

Status	Meaning
401	Missing, invalid, or revoked API key
422	Invalid request body (headline too short/long, etc.)
429	Rate limit exceeded (200 req/min)
502	AI service unavailable — try again

Stories

GET/api/stories

Get paginated story groups with nested articles. Each story group clusters related articles from multiple sources.

Parameters

Name	Type	Required	Description
page	int	No	Page number (default: 1)
per_page	int	No	Items per page, 1-100 (default: 30)
sort	string	No	"latest" (by date) or "sources" (by article count)
search	string	No	Full-text search across headlines and summaries (max 200 chars)
source	string	No	Filter by single source slug (e.g. bbc)
sources	string	No	Comma-separated source slugs (e.g. bbc,nyt,fox)
category	string	No	Filter by source category (e.g. "us_news", "international", "business", "tech_science", "original")
title_format	string	No	"both" (default), "neutral" (only AI-neutralized titles), or "original" (only source titles with bias)

Response (200)

{
  "stories": [
    {
      "id": 42,
      "canonical_title": "Senate Passes Infrastructure Bill",
      "canonical_summary": "The Senate voted 68-29 to pass...",
      "article_count": 5,
      "latest_published_at": "2025-01-15T14:30:00Z",
      "articles": [
        {
          "id": 101,
          "source": { "name": "BBC News", "slug": "bbc", "bias_label": "center" },
          "original_title": "US Senate approves major infrastructure package",
          "neutral_title": "Senate Passes Infrastructure Spending Bill",
          "original_summary": "The US Senate has approved...",
          "link": "https://bbc.com/news/...",
          "published_at": "2025-01-15T14:30:00Z",
          "image_url": "https://..."
        }
      ]
    }
  ],
  "total": 150,
  "page": 1,
  "per_page": 30,
  "has_more": true
}

cURL examples

curl "/api/stories?page=1&per_page=10&sort=latest"

# Search
curl "/api/stories?search=infrastructure"

# Filter by sources
curl "/api/stories?sources=bbc,nyt,fox"

# Filter by category
curl "/api/stories?category=us_news"

# Only neutralized titles (hides original biased titles)
curl "/api/stories?title_format=neutral"

# Only original titles (shows source bias)
curl "/api/stories?title_format=original"

GET/api/stories/{story_group_id}

Get a single story group with all its articles. Supports the same title_format parameter.

cURL example

curl "/api/stories/42"

# With original titles only
curl "/api/stories/42?title_format=original"

Feeds

GET/api/feeds

List all configured RSS feed sources with their name, slug, URL, bias label, and category.

Response (200)

{
  "feeds": [
    {
      "id": 1,
      "name": "BBC News",
      "slug": "bbc",
      "rss_url": "https://feeds.bbci.co.uk/news/world/rss.xml",
      "bias_label": "center",
      "category": "original",
      "is_active": true
    }
  ]
}

cURL example

curl "/api/feeds"

POST/api/feeds/refresh

Trigger a background RSS feed ingestion. Returns immediately — articles are fetched, grouped, and neutralized asynchronously.

Response (202)

{
  "status": "ingestion_started",
  "message": "RSS feed refresh initiated"
}

cURL example

curl -X POST "/api/feeds/refresh"

Health

GET/api/health

Check backend status and database connectivity. Used by Docker healthchecks.

Response (200)

{
  "status": "healthy",
  "db": "connected",
  "timestamp": "2025-01-15T14:30:00Z"
}

cURL example

curl "/api/health"

Error Codes

All errors return JSON with a detail field describing what went wrong.

Code	Meaning
200	Success
202	Accepted — feed refresh queued in background
400	Bad request — validation failed
401	Missing, invalid, or revoked API key
404	Resource not found
422	Unprocessable entity — invalid field values
429	Rate limit exceeded — 200 requests/min per API key
502	Backend or AI service unavailable
504	Request timed out

Example error response

{
  "detail": "Rate limit exceeded (200 requests/min). Try again shortly."
}