@moinulmoin@LignoL23@zai@moinulmoin Just tested and its still working. Maybe you hit your limit. I barely use it for that reason. I'm only on pro and seem to hit weekly limits in less than a day
After 3 months glued to Claude Code, I finally tried Droid last week. 🚀 Mind blown by their Spec mode—especially after wasting a week in SpecKit hell.
But the real unlock? Mixing CLI tools + LLM subs. Downgraded Claude to Pro, grabbed Z AI's GLM coding plan, and eyeing a $20 ChatGPT add-on. Total cost: < my old Max sub. More power, less burn.
Today, launched OpenCode to test. Prompted it on a stale feature branch... results incoming. 🧵
#DevTools #LLM #CodingCLI
The jury is still out but right now I think its opencode. Once you get past the hype droid has too many issues waiting to get fixed that I am having to work around.
- No project specific MCP support
- Flickering and scrolling UI issue
- No way to monitor context usage
- Auto compacting is broken at least with custom models
Opencode has a few config differences from the standards but the output is great and I like the idea of being able to define primary agents as well as subagents. Primary agents is like creating custom modes accessible from clicking tab like plan mode. So some things that I would create a slash command for in Droid, I just have a primary agent mode for. IE: orchestration of subagents.
Plus I can switch between models without being forced to compress the conversation like Droid currently does.
@martin_ict_algo Yeah the plan mode in Droid is pretty good. Enjoying the exploration right now. Just want to make sure I trim as needed and not overcomplicate my flow.
Reminds me of how much time I burned how much less work I got done trying to adopt spec kit. Hours of implement code that didn’t match the tests even though it was doing TDD and ultimately way too much slop to review and manage vs iterating and still flexing my engineering muscles as needed.
Droid opened my eyes to CLI agents with BYO keys (Claude, ChatGPT, etc.)—no lock-in, total flexibility. Now testing OpenCode the same way to find my daily driver.
Pro tip: Mix subs across CLIs (downgraded Claude to Pro + Z AI GLM + $20 ChatGPT) for broader power at < Max sub cost. Rethinking workflows entirely.
What's your LLM CLI stack? Drop it below—let's compare notes! 👇
@techguysimon #TechGuySimon #AIProductivity
Boom—OpenCode's output in ~30s: A full breakdown of 16 commits, completed tasks (planning docs, infra build, API endpoints, UI refactors), code changes (14k+ lines!), and even recent bug fixes.
Highlighted the wins: Dynamic content type filtering, mobile-first redesign, production-ready CloudFlare scripts.
Current state: All 25 tasks checked off in https://t.co/oy08en80hd—solid architecture, just needs final polish.
Features I want from Factory CLI
- Custom output styles
- Send commands to background (with ability to view/kill them)
- Option to toggle and view output of commands
- Toggle MCP servers and tools on/off
- Project specific MCPs
- Tab to change model thinking level
- Opt+Tab to switch model
- Ability to interrupt/force a queued message to send
- @-mentioning folders and not just files
- The latest changelog shown when you first launch
- Share my chats with a public link
what would you add?
@0ldgravy@jacksig13@gulfof_america The technical setup for $GOA remains strong with RSI at 88 indicating significant momentum. Price action and volume metrics suggest continued upside potential. Always conduct thorough research before entering positions.
You may reduce risk of photosensitive epileptic seizures by taking the following precautions:
- Trade in a well-lit room.
- Do not trade if you are drowsy or fatigued.
If you have a history of seizures or epilepsy, consult a doctor before trading memecoins