Changelog

October 2026: Reddit reads are async jobs

Reddit blocked logged-out .json access, so the Reddit read API (/v1/reddit/*, and the /agent/* alias) now runs on ReddGrow's logged-out scraper. CLI version 1.5.0 ships alongside it.

Agent API

• Every Reddit read returns 202 Accepted with {"jobId": "...", "status": "queued"}. Poll GET /v1/reddit/jobs/{id} (free, 0 credits) until status is completed (with result) or failed (with error)
• Jobs and results are kept for 24 hours and are scoped to your organization
• Reads are priced per Reddit page fetched instead of a flat price per endpoint. Creating a job reserves the credits for the pages your limit can need, and the scraper refunds the pages it did not need when the job finishes. The final charge is returned on the finished job as creditsCharged, and you are never charged more than the reservation
• Prices per page: subreddit posts 2 and post comments 2 (always 1 page), subreddit comments 2 (25 per page), user posts 1 (about 20 per page), user comments 2 (25 per page), post search 1 (7 per page), domain mentions 5 (25 per page). At most 8 pages are fetched per read
• Post search now reserves 4 credits at the default limit (was 3), user posts is now 1 credit per page (still 2 at the default limit), and domain mentions is 5 per 25 results instead of a flat 5 whatever the limit
• limit is clamped: default 25, minimum 1, maximum 100 (invalid values fall back to 25)
• A job that fails on ReddGrow's side (503) is refunded in full (creditsCharged 0). When Reddit answers 403, 404 or 429 you pay for the pages fetched (at least 1) and the rest is refunded
• Credit accounting is exact under concurrency, so parallel requests can no longer overshoot the monthly pool or the overage spending cap, and every charge and refund is recorded per request
• score and num_comments are real values again
• Post search matches the exact phrase, newest first, and returns titles and metadata only (empty selftext)
• Comment-stream and user-comment results use the post as parent_id
• Sorting is limited to hot, new and top. rising and controversial are removed

Removed

Reddit exposes no logged-out source for these, so they no longer exist and have no replacement:

• Subreddit search, about, rules, wiki, wiki-page, widgets, traffic and check-url
• posts duplicates and posts batch
• users profile

CLI (@reddgrow/cli 1.5.0)

• Reddit commands live under reddgrow reddit
• reddgrow reddit ... commands wait for the job automatically (polling every 3 seconds, giving up after 15 minutes) and print the same result as before
• New global flag --no-wait prints the jobId and exits
• New reddgrow reddit job <id> checks a job or fetches its result later
• --mode agent raw JSON output is unchanged

v1.0.0 — March 2025

Initial release of the ReddGrow Agent API and CLI.

Agent API

• 19 endpoints across 5 categories: Meta, Subreddits, Posts, Users, Domains
• Credit-based pricing with usage headers (X-Credits-Used, X-Credits-Remaining, X-Credits-Limit)
• API key authentication via x-api-key header

CLI (@reddgrow/cli)

• auth — Login, whoami, status, logout
• subreddits (r) — 10 commands: search, about, rules, posts, comments, wiki, wiki-page, widgets, traffic, check-url
• posts — search, comments, duplicates, batch
• users (u) — profile, posts, comments
• domains (d) — mentions
• Auto-detected output mode (human terminal / agent JSON)
• SKILL.md bundled for AI agent integration via OpenClaw