NCBM b1223c8189 core/agent: calibrate soft-compact with last-request prompt_tokens ...

After the first model call, scale char-based token estimates by
API/resolved prompt size so budget checks track near-real tokens.
/compact uses the same calibration when available.

2026-07-15 11:11:15 +08:00

..

test_budget.py

ci/lint: per-file ignores for tests in ruff and basedpyright

2026-07-15 10:35:30 +08:00

test_compact_summarize.py

core/agent+config: soft context budget in estimated tokens

2026-07-15 10:58:01 +08:00

test_compact.py

core/agent: calibrate soft-compact with last-request prompt_tokens

2026-07-15 11:11:15 +08:00

test_confirm.py

core/agent: optional confirm hook on ToolRegistry execute

2026-07-14 23:30:10 +08:00

test_loop.py

core/agent+cli: treat last-request prompt_tokens as context size

2026-07-15 11:07:52 +08:00

test_parallel_tools.py

core/agent: parallel tool calls and tool-result char budget

2026-07-14 23:24:11 +08:00

test_stream_merge.py

core/lmproto+agent: use normal stream API for chat loop

2026-07-14 22:40:15 +08:00

test_tools.py

test/agent: cover tools, loop, and memory-backed ChatAgent

2026-07-14 17:45:47 +08:00

test_usage.py

core/agent+cli: separate last-request usage from billed turn totals

2026-07-15 11:04:41 +08:00