Knowns CLI
AI-first CLI for task management and documentation.
90 followers
AI-first CLI for task management and documentation.
90 followers
AI-first CLI for task management and documentation. Link docs in tasks, AI reads context automatically. Stop repeating patterns every session.
Free
Launch Team / Built With
CloudflareStill paying to deploy your apps? Try Cloudflare Workers for free
Promoted
@howznguyen Context loss is still one of the biggest productivity killers with AI coding.
Letting AI read tasks and docs directly instead of relying on copy paste feels like the right mental model. Closing the loop by updating docs is a strong differentiator, not just convenience.
I am seeing the same pattern while building Curatora. The real wins come when AI reduces meta work and keeps systems in sync. This feels like a solid step in that direction.
@imtiyazmohammed
Totally agree with this.
Beyond just git storage, I'm actually planning to build a self-hosted solution that serves as a central hub for all project documentation. The idea is that the entire team can contribute and keep things updated continuously.
Think about syncing docs between Backend and Frontend, or between Backend and Mobile App — right now that's a manual headache. Having a single source of truth that AI can read from (and write back to) would eliminate so much friction.
And it's not just technical docs — I want to include tasks, plans, milestones, and even audit logs for team members. Basically everything that gives AI the full context it needs to actually be useful, without the constant copy-paste dance.
Sounds like Curatora is tackling similar problems. Would love to see how you're approaching the "keeping systems in sync" part — that's where most solutions fall short.
Context loss is expensive. Building systems that maintain context by design feels like the unlock.
@howznguyen Persistent, self-updating context is exactly what’s been missing from most AI dev workflows.
Closing that loop between tasks, docs, and memory is a problem we’ve been thinking a lot about while building GTWY.
Context loss is THE killer problem for AI-assisted development. @howznguyen Your insight about letting AI read tasks and docs directly resonates deeply—we waste so much time copy-pasting context that gets stale immediately.
The vision of connecting documentation across Backend/Frontend/Mobile is fascinating. Right now, most teams have scattered docs in Notion, code comments, and Slack threads. A self-hosted solution that maintains a central source of truth for AI agents to query would be incredibly powerful.
I'm curious: How does Knowns handle version control for documentation? When docs update, does the AI automatically get the latest context, or is there a manual sync step? For fast-moving teams, keeping AI context fresh without manual intervention is critical.
@new_user___01920256530b9f092c9e057
You nailed it — context loss is the silent killer of AI productivity.
Version Control
Git IS the version control. All docs live as Markdown in .knowns/ folder. Commit your code, docs get versioned too. No separate system, no cloud lock-in, no sync conflicts.
Does AI Get Latest Context Automatically?
Yes. AI reads directly from your filesystem. Save a file → next AI request sees it instantly. No stale snapshots, no copy-paste ritual.
The `@doc/patterns/auth` reference? It resolves to the actual file on disk. Always fresh.
What About Saving New Knowledge?
This part is intentionally not automatic. Why? If AI saves everything, your docs become garbage.
You decide what's worth keeping. After AI solves a tricky problem, just ask: "Save this pattern to docs." You control the signal-to-noise ratio.
The loop: AI reads automatically → You curate what gets saved.
---
We're also working on self-hosted team sync — shared visibility without giving up local-first control. Files stay the source of truth, server just syncs.
@doc @howznguyen This is brilliant! The Git-based approach with docs in .knowns/ folder is exactly the right architecture - no proprietary formats, just Markdown files that developers already understand. Love that @doc references resolve to actual files on disk, keeping everything in sync automatically.
The intentional curation aspect is really smart too. Having AI read everything but letting developers decide what's worth saving solves the signal-to-noise problem beautifully. It's the opposite of tools that try to auto-save everything and create doc graveyard.
Self-hosted team sync sounds perfect for teams worried about IP leakage. Looking forward to trying this out!
How do you prevent the AI from hallucinating or adding incorrect info to the documentation during the update process?
@landy2
Great question — this is a real concern.
The safeguard: Git.
Knowns stores everything as plain Markdown files in `.knowns/`. Every AI edit is just a file change — you review it in your normal git diff before committing. Nothing gets "auto-merged" into your knowledge base without your approval.
Typical workflow:
AI proposes doc update
You see the diff (`git diff`)
Accept, edit, or reject
You stay in control. AI suggests, Git tracks, you decide.
We deliberately avoided any "magic sync" that bypasses human review. The friction of a commit is actually a feature here.
Context-aware CLIs get really interesting once memory persists across sessions instead of resetting every run.
We’ve seen that payoff when turning task history into durable context while building GTWY.
@human_gtwy
Exactly! That's the unlock — memory that survives the session.
Most AI tools treat each conversation as a blank slate. Knowns flips that: tasks, docs, and decisions persist in `.knowns/` files. When you say "work on task 42", AI reads the full history — what was tried, what failed, what patterns to follow. The real magic: when AI learns something mid-session, it writes back to docs. Knowledge compounds instead of evaporating.
Curious about GTWY — how are you persisting task history there?
Always interested in how others solve the context problem.
@howznguyen Exactly. GTWY stores task history as durable context, not chat. Each run reads + writes back, so memory compounds instead of resetting.
Wow. Wow wow wow. This is incredible. Installing straight away - your demo video was incredible 🤌🏻
@la7
Thank you! 🙏 That means a lot.
Let me know how it goes — happy to help if you hit any bumps during setup. And if you find a workflow that works well for you, I'd love to hear about it!
Howz is my friend, and I'm using Knowns in my stealth startup now. It definitely saved us a lot of time and effort since our team is using Claude Code a lot.
No more copy-pasting context, no more repeating the same prompt, enforcing architecture.
I would recommend this 100%.
@harryvu176
Thanks bro! 🙏
Working with you and your team on that stealth project has been incredibly valuable. The feedback from your team helped me improve so many things — from the CLI experience to how documents are structured and synced.
Building a tool is one thing, but seeing it actually save time for a real team shipping real products? That's the best validation I could ask for.
Appreciate the support, and can't wait to see what you guys are building! 🚀
Congrats on launch — love how Knowns turns CLI tasks into real AI-aware workflow.
@zeiki_yu
Thanks so much! 🙌
That's exactly the vision — making CLI the bridge between your workflow and AI's understanding. Instead of AI being a separate tool you have to constantly feed information to, it becomes part of the workflow itself.
And this is just the beginning — in the future, Knowns will also connect everyone across the entire project. Imagine the whole team sharing context, docs, and tasks in one place that AI can read and update. No more silos, no more sync issues.
Appreciate the support! 🚀