Before: three drafts, three voices, none of them right
- Reddit downvotes anything that smells like a press release.
- LinkedIn punishes Reddit slang and short, dry text.
- X cuts your headline in half and reposts a screenshot instead.
Case 02
One agent, three voice sets, all in the same chat.
Same one-time sign-in as case 01.
Claude Code labels them per platform and saves each as its own knowledge source.
It calls prostir_agents_build with three platform sets and a routing rule in the system prompt.
Claude Code adds a small JavaScript tool that catches posts longer than X's character limit or shorter than Reddit's minimum.
Your link: prostir.build/yourhandle/three-feeds. Add it once to ChatGPT, Claude, or Codex CLI. Same thing.
Each set is a small batch of real posts that landed on that exact platform.
One chat, one agent, three voice sets.
Build me a social writer on Prostir that picks the voice when I name the platform. I'll paste six reference posts here.
prostir_agents_design_interview Confirming: one agent, three platform sets (reddit, linkedin, x), routing inside the system prompt. Add a length check tool? Recommended.
Yes. Block AI tells like unlock and in-today's-fast-paced-world.
prostir_agents_build Created the agent, attached six knowledge files (two per platform), added length_guard.js, wired the routing rules, published to /yourhandle/three-feeds. ✓ Done.
Try: a LinkedIn post about how I shipped this in five minutes. The router picks the LinkedIn voice, the length tool checks the length, banned words get blocked at draft time.
What gets published
Ready to launch?
Upload what only you have. Write the rules in plain language. Open the link in ChatGPT, Claude, or your site. Ship the agent only you could ship.
Need custom scale, white-label, or dedicated setup? Talk to sales.