Skip to content

v0.12.6

Choose a tag to compare

@coder-eval-release-bot coder-eval-release-bot released this 23 Sep 18:42
· 41 commits to main since this release

Bug Fixes

  • claude-code: End a turn the CLI did not cap at max_turns
    (#194,
    17fb10f)

  • claude-code: Kill the CLI on a cooperative stop
    (#194,
    17fb10f)

  • claude-code: Kill the CLI on a max_turns backstop stop without a turn timeout
    (#194,
    17fb10f)

  • codex: End a capped turn before the next call can run a tool
    (#194,
    17fb10f)

  • delegate: Count a tool-only reply as a model call
    (#194,
    17fb10f)

  • delegate: Keep counting model calls when a tool never returns
    (#194,
    17fb10f)

  • evalboard: Compare visible turns, not SDK num_turns, against expected_turns
    (#194,
    17fb10f)

  • pricing: Add Opus 5.5 and GPT-6, correct rates that moved publicly
    (#195,
    11b902f)

  • run-limits: Count max_turns as model API calls on every harness
    (#194,
    17fb10f)

  • run-limits: Count max_turns as model API calls on every harness and enforce it on claude-code
    (#194,
    17fb10f)

  • run-limits: Stop a turn mid-flight once a token or USD budget is crossed
    (#194,
    17fb10f)

  • simulation: End a dialog with agent_max_turns when the agent hits its own cap
    (#194,
    17fb10f)

Documentation

  • parity: Record the claude-code max_turns backstop
    (#194,
    17fb10f)

  • run-limits: Say max_turns restarts per iteration and expected_turns counts tool calls
    (#194,
    17fb10f)