English
David
Report

David✓

@dasfacc

introspecting about agents, economics and health – vibes are my own

1.1K Following232 Followers3.6K Posts
  1. David✓@dasfacc
    2026 is the year of accountability
    21111.7K
  7. David✓@dasfacc
    we improved cache reads for any @pydantic ai agent, with no code changes needed! measured on haiku 5.5: resuming a run from a StepStore: 67% → 98% right after Code Mode's dynamic catalog finds a tool: 15% → 97%
    Download
    04181.5K
  8. David✓@dasfacc
    pydantic ai now returns structured citations Anthropic, OpenAI, Gemini, Bedrock, OpenRouter and xAI all come back in the same shape
    Download
    3015670
  9. David✓@dasfacc
    David
    RT @mitsuhiko: As fun as AI is, we built so much bad software over the last 30 years, and now the universe is punishing us for our sins.
    0110047K
  10. David✓@dasfacc
    SmartSearch lets your @pydantic ai agent search code by what it does, in natural language a judge model scores snippets on a shortlist and passes back ranked excerpts to the agent with file and line ranges
    Download
    108271
  11. David✓@dasfacc
    your claude can now answer questions about a video it still can't watch one, but we got your back FileUnderstanding uses gemini to describe it once, and claude (or any model) reads the description
    Download
    00136
  12. David✓@dasfacc
    you know what would be cool if an agent would look at my queued messages on grok build and identify what stuff it can start parallel work on what say you @poteto
    002204
  13. David✓@dasfacc
    im pretty sure codex got pwned
    Download
    001181
  14. David✓@dasfacc
    you can't blame people for what they did when they were llm'less
    The Hacker News @TheHackersNews
    🚨 CRITICAL: GitLab flaw (CVSS 10.0) affects 12 years of releases. CVE-2026-110264 could allow unauthenticated remote code execution on self-managed servers. Read -
    002230
  15. The Hacker News@TheHackersNews
    🚨 CRITICAL: GitLab flaw (CVSS 10.0) affects 12 years of releases. CVE-2026-110264 could allow unauthenticated remote code execution on self-managed servers. Read -
    2652.9K14.9K3.6M
  16. David✓@dasfacc
    why am I getting 100s of these
    Download
    10179
  17. David✓@dasfacc
    hey codex please summarize the anthropic IPO filings for me
    Lain on the Blockchain @CryptoCyberia
    Anthropic lost 80,000,000,000.0000000000000000000000 btw
    102227
  18. Lain on the Blockchain✓@CryptoCyberia
    Anthropic lost 80,000,000,000.0000000000000000000000 btw
    1962393K
  19. David✓@dasfacc
    I know exactly what this repo is missing more labels
    103189
  20. David✓@dasfacc
    microvm my balls, this is a lot better
    Marcelo Trylesinski @marcelotryle
    Monty can pause agent-written code, turn it into bytes, and resume it later ⏸️ - dump() serializes the whole interpreter - load_snapshot() resumes it in another process or machine - Wait hours for a human approval with nothing running
    002168
  21. Marcelo Trylesinski✓@marcelotryle
    Monty can pause agent-written code, turn it into bytes, and resume it later ⏸️ - dump() serializes the whole interpreter - load_snapshot() resumes it in another process or machine - Wait hours for a human approval with nothing running
    Download
    44622.8K
  22. David✓@dasfacc
    you'd think they would've learnt by now
    Socket @SocketSecurity
    🚨 LiteLLM, a Python package for connecting to AI model providers with 92M+ monthly downloads, has been compromised. Version 1.107.0 ships credential-stealing malware through PyPI wheels and Docker images, targeting cloud credentials, developer secrets, and AI tools.
    004267
  23. Socket@SocketSecurity
    🚨 LiteLLM, a Python package for connecting to AI model providers with 92M+ monthly downloads, has been compromised. Version 1.107.0 ships credential-stealing malware through PyPI wheels and Docker images, targeting cloud credentials, developer secrets, and AI tools.
    Download
    1103261.9K303K
  25. Show more posts