Log inSign up
Graham Lipsman
5,196 posts
@glipsman

Graham Lipsman

@glipsman
Designer/Developer/Entrepreneur. Built the first Figma to code MCP. Taking it to the next level @ framelink.ai
glipsman.com
Joined July 2010
215
Following
1,471
Followers
RepliesRepliesRepostsRepostsMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
  • Pinned
    @glipsman
    Graham Lipsman
    @glipsman
    Feb 13, 2025
    I wrote a Cursor<>Figma MCP server to make it easier for AI copilots to implement designs. It's not perfect, but the resulting HTML and CSS is a lot cleaner than giving AI a screenshot and having it guess. Link in the replies, take it for a spin and let me know what you think!
    Image
    00:00
    24
  • @glipsman
    Graham Lipsman
    @glipsman
    Sep 4
    15k tokens per second is unbelievable. Highly recommend seeing it for yourself for a glimpse of the very, very weird future of swarm intelligence that's in store. chatjimmy.ai
    Image
    00:00
    10
  • @glipsman
    Graham Lipsman
    @glipsman
    Sep 1
    Many people are working on realtime visualizations of active agents which seems like a great way to burn a ton of tokens and that's about it. Much better to focus on either side of the problem: effectively describing and dispatching work or verifying outputs.
  • @glipsman
    Graham Lipsman
    @glipsman
    Aug 27
    My codebase-visualizer-as-dwarf-fortress is shaping up nicely—it's actually useful for understanding a codebase. This codebase is craaaazy—I've always known it but now I can point directly to where the issues are.
    Image
    @glipsman
    Graham Lipsman
    @glipsman
    Aug 24
    Image
    Not sure if this is anything yet
    44
  • @glipsman
    Graham Lipsman
    @glipsman
    Aug 27
    "We did not see them successfully edit the source of truth in our transcripts"
    Image
    GIF
    @METR_Evals
    METR
    @METR_Evals
    Aug 26
    Replying to @METR_Evals
    Agents also tried to retroactively edit transcripts. They modified some accessible action logs, but these weren’t the source of transcripts (agents realized this but tried anyway). We did not see them successfully edit the source of truth in our transcripts (scoped to Jul 7-13).
Advertisement
Advertisement