v0.1.61

Changelog

• 73a44eb Document list [refs] marker and inject call collapsing

v0.1.60

Changelog

• 1a7b039 Address code review findings: DRY, robustness, test coverage
• e0779a2 Collapse cc-session inject calls and mark referencing sessions in list
• 352bb0b Fix benchmark tests failing after date window expiry
• 2c79626 Fix refs false-positive, benchmark panic, and dead field

v0.1.59

Changelog

• e2fe9f1 Clarify benchmark docs after review
• cbb0af7 Harden benchmark command tests
• ef6f59a Model paginated inject setup costs
• bdd0e05 Tighten benchmark cost model tests
• 01e9c5b Update benchmark docs for paginated inject cost
• be8f223 Use fractional calls per turn in benchmark costs

v0.1.58

Changelog

• 1c45920 Rewrite SKILL.md to make inject the clear default over read

v0.1.57

Changelog

• 2b61bd0 Fix code review findings: error handling, naming, DRY, and test assertions

v0.1.56

Changelog

• 6f01a8a Split model.go and event.go to follow SRP

v0.1.55

Changelog

• 0795ab0 Fix tool result matching to use tool_use_id instead of blind last-pending

v0.1.54

Changelog

• 780beac Skip caller session test on Windows where Claude Code paths don't apply

v0.1.53

Changelog

• cf23210 Add unit tests for cost model formulas in internal/benchmark
• d3ffa6f Refactor benchmark into internal package and gate usage logging on Claude Code caller

v0.1.52

Changelog

• 35183f5 Support flexible truthy values for no_usage config field