rising
A claude code plugin that turns claude into a personal Reddit content engine
for founders, marketers, and devrel folks. Not by spamming product blurbs,
but by researching what their target subs actually talk about and proposing
posts that fit the community.
The name comes from Reddit's "rising" sort — the posts on their way up. That's
what this plugin helps you write.
Reddit punishes self-promotion. So this plugin doesn't post for you. It does
the research that good Reddit content requires (writing style, sub trends,
performance history, sub rules) and proposes 3 posts a day that you then
review, edit, and post manually.
What it does
One-time setup (/reddit-setup):
- Asks for your product brief, Reddit username, and goal
- Pulls your past Reddit posts and analyzes your writing style
- Discovers ~10 subreddits split into "posting" (where to post) and
"non-posting" (research only — competitors / trend signal)
- Caches each posting sub's rules
- Categorizes your past posts (conversational / soft-promo / promotional)
and computes per-category engagement stats so the engine knows what's
worked for you
Daily run (/reddit-daily):
- Refreshes engagement on your recent posts
- Dispatches one research agent in parallel for every configured
subreddit — each pulls top-of-week + hot, summarizes themes
- Synthesizes 3 post proposals — title, body, target sub, category, and
reasoning that cites which trend it's responding to and why this format
has worked for you before
- Saves to
drafts/YYYY-MM-DD.md and pastes them into chat for review
To log a post after you've made it, use /reddit-add-post <url> separately —
the daily run never blocks on user input.
Other commands:
/reddit-add-post <url> — log a post you made (appends to history,
refreshes performance)/reddit-add-sub <angle> — discover and add fresh subreddits when your
current list goes stale. Give it an angle ("AI music creators", "my
vibecoding subs are tapped out") and it researches new candidates,
asks you which to keep, caches their rules, and appends to your list/reddit-schedule "weekdays at 9am" — automate the daily run
Install
/plugin marketplace add agamjn/rising
/plugin install rising
Quickstart (5 minutes)
# Make a folder for your product
mkdir -p ~/reddit/my-product
cd ~/reddit/my-product
# Set up
/reddit-setup
# (answers product URL, Reddit username, and goal interactively)
# Get your first batch of proposals
/reddit-daily
# (open drafts/<today>.md, pick a proposal, post on Reddit)
# (afterward, log it back so the engine learns)
/reddit-add-post https://reddit.com/r/<sub>/comments/<id>/<slug>/
# Optional: automate daily
/reddit-schedule "weekdays at 9am"
Want to see what files this produces? See examples/tensorfuse/
for a fully-populated example product folder, including a real-looking
drafts/2026-05-09.md.
⚠️ Reddit profile visibility caveat
The setup pulls your posts via Reddit's public JSON API, which only works
on public profiles. Most users hide their profile by default.
The recommended flow:
- Before running
/reddit-setup, make your Reddit profile public:
- Click your profile avatar (top right) → User Settings
- Open the Profile tab → scroll to Content and Activity and
turn on the toggles that make your posts and comments visible
- Open the Privacy tab → toggle ON Show up in search results
- Run
/reddit-setup — it pulls your full history and saves it to
posts-history.md in your project folder
- Hide your profile again (flip those toggles back off)
- From then on, log new posts manually with
/reddit-add-post <url> —
the engine grows your history without your profile being public
Your posts-history.md is the source of truth after setup. Reddit's API
is only used for the initial pull and for fetching individual post URLs
you provide.
Architecture (for the curious)
- 4 slash commands as entry points
- 5 sub-agents — most heavy lifting happens here, dispatched in parallel
for fan-out work like per-subreddit research
- 1 skill (
reddit-writing-style) — reusable style-extraction workflow
- No MCP server — slash commands curl Reddit's public
*.json endpoints
directly (Bash + curl, with a Reddit-conformant User-Agent). Sub-agents are
pure analyzers reading from a local .rising/cache/ directory. No OAuth,
no install steps, no auth tokens
- Per-project data — everything lives in your CWD. Want a separate
product? Make a separate folder
The plugin doesn't post to Reddit. Ever. There is no Reddit write API
integration and no way to add one without modifying the plugin's source.
Why no auto-posting?
Three reasons:
- Reddit ToS. Automated posting via the API requires OAuth + a
registered Reddit app, and even then the rules around automation are
strict. Most accounts that try this get shadowbanned within weeks.
- Quality control. The whole point of doing Reddit right is human
judgment — picking the moment, tweaking the tone, responding to
comments. Auto-posting throws away the part that makes the strategy
work.
- Account safety. A founder's Reddit account is a long-term asset.
Letting an LLM post unsupervised is a fast path to a permaban for
spam.
The engine does the 90% of work that's tedious (research, drafting). You
do the 10% that has to be human (review, post, engage with comments).
Configuration files (in your CWD after setup)
| File | Purpose |
|---|
product.md | Your product brief |
goal.md | What you want from Reddit |
writing-style.md | Your voice, distilled from past posts |
posts-history.md | Append-only log of every post you've made |
post-categories.md | Classification + per-category performance stats |
subreddits.md | Posting subs + non-posting subs, with rationale |
subreddits/<sub>/rules.md | Cached rules per posting sub |
drafts/YYYY-MM-DD.md | Daily output — 3 proposals |
.rising/config.json | Your Reddit username (for the API User-Agent) |
.rising/cache/ | Raw Reddit JSON for the current run — gitignored, safe to delete |
Edit any of these freely between runs. The engine uses whatever it finds.
License
MIT. See LICENSE.
Contributing
Issues and PRs welcome. The plugin is opinionated — see the agent and
command files for the design choices. The goal is to keep it simple:
no MCP server, no auth, no DB, just markdown files and the public Reddit
API.
