The Reddit
mining API.
Pull posts, threads, and clustered unanswered questions from any subreddit as clean JSON. We run the headless browsers and residential proxies — you call an endpoint.
$ curl -s https://www.trufflejar.com/api/v1/posts \
-H "Authorization: Bearer tj_…" \
-G -d subreddit=ClaudeAI -d sort=top -d time=week -d limit=5
{
"subreddit": "ClaudeAI",
"sort": "top",
"posts": [
{ "title": "Fable staying on Max", "score": 3234, "num_comments": 712, … },
{ "title": "New: Teach Claude a skill", "score": 2441, … },
{ "title": "claude doesn't lie anymore", "score": 2253, … }
],
"bandwidth": { "requests": 3, "wire_mb": 0.21, "est_cost_usd": 0.0002 }
}Real output — pulled during our first smoke test.
Get a key
Sign in with your email and create a key. No credit card — the API is free while we build.
Call an endpoint
GET /posts and /thread return in seconds. POST /mine scans a whole subreddit and returns a job you can poll — or wait on the request.
Ship what you find
Unanswered questions, content gaps, and real language from real threads — JSON your product or agent can use directly.
What comes back
- GET /v1/postsa subreddit's posts by top, hot, new, or rising — titles, selftext, scores, and permalinks
- GET /v1/threadone thread with its full comment tree, as Reddit's raw listing JSON
- POST /v1/minescans up to 100 threads, detects questions, flags the unanswered ones, and clusters them by topic
- GET /v1/jobs/{id}status and result for a mine you didn't wait on
Open source at the core
The miner is a single TypeScript file, and it's the same one this API runs. Want it free? Bring your own residential proxy and run it on your machine — grab the code at github.com/johnkueh/trufflejar-miner, or point your coding agent at skill.md and it sets everything up. The hosted API exists because the proxy is the hard part: we clear Reddit's bot challenge, rotate residential IPs when they get flagged, and hand you JSON.