slowbench
RetrievalInferenceEvalsAgentsFindingsSeriesBenchmarksArchive

One session, eighty days

Where agent context goes

A coding agent stopped answering for nineteen minutes. The investigation that followed found that context is not spent gradually on conversation — it goes in jumps, and the jumps come from a handful of commands nobody remembers running.

6 posts

Where this stands

Held to be true
Context fills from tool output, not from conversation, and the cost is paid on every subsequent turn rather than once. Continuing a session feels free at each individual moment, which is why the ceiling arrives without warning.
What would break it
A long session where conversation, not tool output, dominates the transcript by size.
Still open
What a good session boundary actually is. The current answer is a percentage on a status line, which reports the problem rather than preventing it.
  1. 1
    Agents9 min

    An 80-day agent session and the wall at the end of it

    Claude Code context limit on an 80-day session: fifty compactions, each starting from a full million tokens. Then one morning it stopped answering for nineteen minutes.

  2. 2
    Agents10 min

    Context is burned by tool output, not by conversation

    Claude Code context usage over a two-month transcript: the largest entries were single command results, around 300,000 tokens each.

  3. 3
    Agents8 min

    Building the guardrails I needed after a session died

    Claude Code hooks and a status line for context usage and compaction. Both read the transcript rather than an API field, and that choice is the post.

  4. 4
    Agents8 min

    I instrumented everything except where the decisions happen

    LLM agent observability with a connected, green tracing dashboard, and zero traces from the three components that actually make choices.

  5. 5
    Agents9 min

    LangGraph yes, LangChain no

    LangGraph vs LangChain in an agent pipeline: one earns its place by doing something I would otherwise write badly. The other is a spec I satisfy without importing.

  6. 6
    Agents4 min

    The numbers halved and the fix never ran

    Deployment verification the hard way: wall clock down 53 percent, tool calls down 47, and the line the fix exists to produce absent in both runs. The config was a symlink shared by every worktree.