2025-03-09 Sun

#AI Coding
- Cursor
  - #Updated CursorAI Project Rule Test with a list of current tests
  - #Filed
- aider/Installation
  - #Filed GitHub/Issue For security reasons, default installation docs should not instruct users to divulge their API keys at the command line · Issue #3475 · Aider-AI/aider
#Agentic Systems
- Anthropic/Blog/24/12/Building Effective Agents
  - #Wordsmithing
    - Agentic Spectrum refers to the spread between Agentic Systems that are more deterministic and pre-specified AI Workflows, and those Agentic Systems where the AI is given freedom to "choose" the next best move or what to do next, AI Agents. It seems that this spectrum is mapped to the space between 0 and 1, from "non-agentic" workflows to "fully agentic" agents.
    - Poka-yoke when you bake something into a process or a technology that prevents a class of errors
      - ⁠⁠Poka-yoke your tools. Change the arguments so that it is harder to make mistakes.
GitHub Pages
- I don't think I quite understood before that GitHub/Pages/User Site was a thing; it's possible to have a domain like <username>.github.io
#Security
- smallstep/Blog/24/05/How to Handle Secrets on the Command Line
  - tldr #Summary
    TL;DR: Best practices for #CLI secrets: (1) Use credential files with proper permissions and disk encryption; (2) Use pipes to pass secrets directly between programs; (3) Avoid EnvVars environment variables when possible; (4) Never pass secrets directly in command arguments; (5) Consider using a keyring facility like Linux/keyring for in-memory storage; (6) For specific tools, use their built-in credential storage (like dotfile/.netrc for curl).
#MCP
- Person/David Ondrej/YouTube/25/03/Build Anything with MCP Agents
uv
- #Filed Diataxis/References
  - uv tool install
  - uv/tool

2 Linked References

Unlinked References