Capabilities
Everything pendpost can do, behind your approval gate
pendpost is more than a scheduler. An AI agent can plan, publish, listen, reply, and report across your channels, and every action stays behind the same human approval gate. Here is the full surface, so you can see what is real before you install. Read-only tools can never publish, and beta lanes are labelled as beta.
Publishing and native scheduling
Draft and schedule posts across Instagram, Facebook, LinkedIn, YouTube, X, Telegram, Discord, Mastodon, Nostr, and more, plus long-form blogs on WordPress and Ghost. Reddit, Pinterest, TikTok, and Google Business Profile are in beta. Facebook, YouTube, Mastodon, WordPress, and Ghost schedule natively, so those posts fire even when your machine is off. The rest publish while pendpost is running, which you can cover with the optional 24/7 cloud service or your own always-on host.
The approval gate and progressive autonomy
Every post is fail-closed: with no approval, it does not publish, and by default the approver is you. When you are ready to go hands-off, opt into auto-approve. It is an owner-only policy you can revoke at any time, and revoking it returns every not-yet-published auto-approved post to draft. Anti-ban circuit breakers halt a platform lane on an action block and do not auto-resume until you clear them.
Client review links
Send a client a private review link. They open it in a browser, approve or reject each post with no login and no account, and their decision flows straight into the same fail-closed gate. It is the sign-off step agencies need without inviting the client into your workspace.
Comment inbox and moderation
See every unanswered comment on your recent posts in one comment inbox, and reply from the dashboard or from an agent. Comment moderation lets you hide, hold, or mark a comment as spam. Replies and moderation actions run behind the same approval control as posts.
Radar: intent listening and AI visibility
Radar finds the places your buyers ask questions and drafts approval-gated replies you can send after review. It also tracks your AI-visibility (GEO): whether AI answer engines mention you for the questions that matter, logged over time so you can move the number. A follow-up loop tells you when the original author replied back. Radar listens on Reddit, Hacker News, Bluesky, and Mastodon, which are listening sources, not publish lanes.
A multi-brand workspace
Run multiple brands from one workspace, each with its own connected accounts, plan, and approval state. A cross-client overview shows what is pending, overdue, or blocked across every brand at once. This is in the free self-hosted core, not just the paid cloud.
Insights and a performance digest
Pull fresh insights across every connected lane, then generate a plain-language performance digest in English or Swiss German. It reads back what happened and what moved, so a report is one command rather than an afternoon.
Relationship memory
pendpost keeps a relationship memory of the people who keep engaging with you, across posts and platforms, so replies can stay personal instead of starting cold every time.
Deep per-platform actions
Each lane exposes more than a plain post. A selection:
- Google Business Profile: read and reply to Google reviews, add media, and set business attributes.
- Nostr: send Lightning zaps and manage your relays (NIP-65) and lists (NIP-51).
- Ghost: publish a post and send it as a newsletter, and manage Ghost members including bulk CSV import.
- Pinterest: create and organise Pinterest boards and sections.
- YouTube: build playlists and add videos to them.
- Discord: create Discord scheduled events in your server.
- Mastodon: pin a post and follow accounts.
- Profiles: edit your profile across X, Mastodon, Nostr, Telegram, and YouTube.
- Media: carousels with a grid-crop cover preview so the profile-grid crop matches what you approve.
Local-first and MCP-native
pendpost binds 127.0.0.1, keeps your keys in your own .env, and never phones home. Every capability here is an MCP tool and a dashboard control, kept at parity by a test, so an agent and a person reach the same surface. Captions run through a humanizer brand-lint before they can publish. For the full setup and per-lane detail, see thedocumentation.