Tag: agents

  1. No. 015

    Argument list too long, part 2: the prompt goes in a file

    Part 1 bought room under the 128 KB per-argument limit Linux has with the usual 4 KB pages by counting notes instead of listing them. Part 2 is the real fix: my launcher writes the startup notes to a private file and hands Claude Code only its path. On the way, a test showed why a size cap counted in characters would have let through a string almost twice the limit.

  2. No. 014

    A one-word Claude Code call read 236K tokens

    I switched off every built-in tool on a headless Claude Code helper to make it cheap, and its one-word answer read 236K input tokens, seven times the default. The cause was MCP tool schemas, the fix was one flag, and it started a month of measuring what an agent carries into every turn.

  3. No. 013

    Argument list too long, part 1: from a single argument

    My front door would not restart because of "Argument list too long", and the list held one argument. Months of startup notes had grown into a 172 KB string, past the per-argument limit Linux keeps well under its total for the whole launch (2 MB on my box). Part 1: the limit and the quick fix.

  4. No. 011

    What my coding agent briefs forgot to name

    Five weeks of handing jobs to Codex and Claude workers, and what each failure taught me a brief must name: where to write, which files, when it is done, and how the answer gets back.

  5. No. 008

    My router trusted a usage reading that was seven hours old

    Spreading work across several Claude Code and Codex accounts means watching their five-hour and weekly limits. A cached reading said an account was nearly empty while its live session was past the threshold, and fixing that grew into one line that tells every agent where to go.