poytr1

6 ms·

1. 1.▲
   by poytr1 10d ago

   Yes, I’m the author of this paper. One correction: the paper is inaccurate on this point. If you look at the implementation( https://github.com/agentconnect-md/lsp-vs-grep-token-study/b... ), the harness is actually
2. 2.▲
   by poytr1 13d ago

   I had a similar intuition that we could definitely optimize token efficiency and performance for non-general tasks. For example, I was handling a lot of infra operations, many of which follow fixed workflows like locating a specific log in
3. 3.▲
   by poytr1 13d ago

   Interesting tool, it seems that dedicated skills for these tools are important. Even explicit system-prompt instructions—such as “you must use tool X for code exploration”—often fail to behave as expected. Another tricky part is that how to
4. 4.▲
   by poytr1 14d ago

   I am thinking of where the initial training data came from. For example, Claude Code likely collected a substantial number of real-world coding trajectories through its CLI. However, trajectories involving tools such as LSP, MCP, or AST-gre