Annotate agent plans, html, and review code visually. Iterate with your agent. Share with your team • https://t.co/3Gg5SohuqEgithub.com/backnotprop/pl…Joined January 2026
@PhilShteuck@emil_priver fwiw there are users who like using speckit/superpowers/similar with plannotator...
the most simple example is annotating those files, but you can setup review gates and what not. It only adds a surface to review & integrates feedback with the agent!
a flow I really like is using prompting my idea and telling the llm to use @plannotator annotate and give me a todo of a solution. If I feel that I want to see code examples do I let the LLM do it's thing and then I use @plannotator diff review to see the actual changes
and then I throw all the llm code away and built it myself :D
I was sleeping on @plannotator holy shit
the concept is simple but the depth in execution is something you feel once you've tried it
this one's going to stick
Simplifier: You can now use plannotator to edit markdown files, collaborate with an in browser agent.
Its a fun experience if you are actively engaged in creating specs/adrs/context/docs/etc.
The TUI-in-a-GUI code is an open source package, use it for your own apps. Link below
Plannotator 0.21.0 is out
Edit files. Co-context with an embedded cli agent.
- Direct document editing
- Launch an agent to build context with together
- HTML files have first class support
- Open any file you are viewing in external app
- Bedrock/Vertex support
- Stronger
Plannotator 0.21.0 is out
Edit files. Co-context with an embedded cli agent.
- Direct document editing
- Launch an agent to build context with together
- HTML files have first class support
- Open any file you are viewing in external app
- Bedrock/Vertex support
- Stronger sharing configs
- Ergonomics UX polish across the app.
If you want a persistent experience, just run
`plannotator annotate folder/`
(It'll launch the same way you launch any IDE.)
If you are a plannotator user and want to compound (make agents better according to how you work) like @dillon_mulroy, you can do so based on your feedback that has been stored on your disk for any code or annotation reviews.
1. tell your agent to look in ~/.plannotator/drafts/
2. analyze the json and parse all "codeAnnotations" (code review) or "annotations" (plan feedback)
3. follow the process: x.com/dillon_mulroy/…
4. (optionally 😄) sign up here: plannotator.ai/workspaces
@plannotator I've actually filmed a video walkthrough of me using Plannotator with Pi, and building a RAG application powered by @elastic. I'd love to know what you think of it.
youtu.be/ZdkPaXhly34
Code Review: The feedback prompt to your agent has been modified to force it to verify and acknowledge findings, and discuss with you before changing code.
Hope this is not too disruptive but latest gen models have been very eager to jump into changes!
This feedback is still customizable
Plannotator 0.20.2 is out
Code Review:
- (perf) Pierre worker-pool highlighting
- (perf) large PRs: paged-diff fallback + local recompute - unified agent engine setting
Pi:
- other extensions can drive plan mode (enter/exit/toggle)
@Ookay2077@pyrons_@theo@trq212 Codex is the hardest harness to integrate with. You basically need to use skills only and hope the agent can make the right calls.
Codex hooks lacks ergonomics.
Codex has no ability to invoke slash commands.
Codex skills do not support bash/shell execution.
@pyrons_@pierrecomputer probably wouldn't review a PR this big, but shipping an abstraction of their code viewer virtualization - should fix your issues! ty pierre 🪨
54K Followers 4K FollowingI started and sold 2 small companies: Museum Hack and FDS Avionics. Now I write, invest in stocks, and meet interesting people // https://t.co/4hnVTJWwfa
1 Followers 267 FollowingAmazon marketplace operator. Organic ranking strategy, advertising, product launches, brand portfolio management. Ran a 13-person agency. Now building with
643 Followers 3K Followingfounding swe @AutoRFP (since Oct 2025). ex-swe @canva (2021-2025). I like building things. husband to @ashleymjtech. up-and-coming girl dad
326 Followers 3K FollowingInterested in all things Product Design • Designer that codes on the side • Exploring the world of tools for thought and productivity
3K Followers 421 FollowingLiterally nobody. I JS things @pierredotco. Contributed to ViM once. Previously at https://t.co/kfemvP0EDB https://t.co/QxvkrGs5G3 https://t.co/mIzvuWNaka
210K Followers 0 FollowingThe free and flexible app for your private thoughts. For help and deeper discussions, join our community: https://t.co/wHB7xZ3AjA
590K Followers 51K FollowingSan Francisco/Silicon Valley AI | Robots, holodecks, BCIs, analysis of new things | Ex-Microsoft, Rackspace, Fast Company | Wrote eight books about the future.
540K Followers 24 FollowingThe AI that does things. Emails, calendar, home automation, from your favorite chat app. Your machine, your rules.
New shell, same lobster soul. 🦞
554K Followers 2K FollowingPolyagentmorous ClawFather. Came back from retirement to mess with AI and help a lobster take over the world.
@OpenClaw🦞 + @OpenAI
54K Followers 1K FollowingArmin's handler at https://t.co/B05ybKGkzx. Old man yelling at Claudes.
https://t.co/Q1wG57v1yc
https://t.co/mnOoWUr0TO
https://t.co/8i5vIRE0Wn
1.5M Followers 2 FollowingClaude is an AI assistant built by @anthropicai to be safe, accurate, and secure. Talk to Claude on https://t.co/ZhTwG8d1e5 or download the app.