ECC/tests
AlbertChiu777 536221cf7a
fix(hooks): dedupe transcript usage by message.id in cost-tracker (~2.5-3x inflation) (#2483)
Claude Code writes one transcript JSONL line per content block, so a
single API response (one message.id) spans multiple assistant lines that
each repeat the same message.usage. sumUsageFromTranscript summed every
line, inflating token totals and estimated_cost_usd roughly 2.5-3x.

Verified on a real session: 704 assistant lines but only 286 unique
message.ids (2.46 lines/response on average); line-summing reported
$866.52 while the deduped total is $332.62. Usage payloads are identical
across lines of the same id (0/286 varied), so counting once per id is
equivalent to taking the last line per id.

Fix: collect usage into a Map keyed by message.id (last line wins) and
sum unique entries. Lines without a message.id (older transcript shapes)
keep the previous per-line behavior via a synthetic key, so existing
tests and old transcripts are unaffected.

Adds a regression test: a response split into 3 content-block lines with
the same message.id is counted exactly once.

Note: rows already written to ~/.claude/metrics/costs.jsonl by the old
code carry inflated token counts and estimates (except rows whose cost
came from the harness-cost cache, where cost is authoritative but token
counts are still inflated). Downstream consumers may want to annotate
history; this change intentionally does not rewrite the raw log.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-07-27 11:14:47 -07:00
..
ci fix: use scalar Claude agent tools (#2583) 2026-07-26 03:20:15 -07:00
commands fix: accept crlf command frontmatter 2026-04-30 03:41:18 -04:00
docs feat(codex): add ECC navigation guide (#2525) 2026-07-22 20:44:37 -04:00
hooks fix(hooks): dedupe transcript usage by message.id in cost-tracker (~2.5-3x inflation) (#2483) 2026-07-27 11:14:47 -07:00
integration refactor(hooks): consolidate PostToolUse hooks into sync/async dispatchers (#2494) 2026-07-19 15:47:10 -04:00
lib fix: harden local dashboard and data boundaries (#2585) 2026-07-27 11:11:29 -07:00
scripts fix: harden local dashboard and data boundaries (#2585) 2026-07-27 11:11:29 -07:00
__init__.py Small changes 2026-04-12 12:34:45 +05:30
codex-config.test.js fix(codex): broaden context7 config checks 2026-03-29 00:26:16 -04:00
conftest.py docs: salvage focused stale PR contributions 2026-05-11 05:31:12 -04:00
opencode-config.test.js test(opencode): reject any namespace-scoped command agent id, not just the legacy prefix (#2477) 2026-07-10 09:58:03 +05:30
opencode-plugin-hooks.test.js feat(opencode): 全面升级OpenCode集成 (#2251) 2026-06-15 14:01:34 -04:00
opencode-tools.test.js fix(opencode): normalize tool paths across platforms (#2459) 2026-07-26 03:13:49 -07:00
plugin-manifest.test.js fix: add plugin cache health check (#2249) 2026-06-15 14:01:25 -04:00
run-all.js feat: worktree-lifecycle service (deterministic conflict prediction + safe GC) (#2164) 2026-06-07 13:00:08 +08:00
test_astraflow_provider.py feat: add Astraflow provider support 2026-05-11 23:21:46 -04:00
test_atlas_provider.py feat: add Atlas Cloud as LLM/AI provider (#2279) 2026-06-18 16:29:11 -04:00
test_builder.py fix: docs/COMMAND-REGISTRY.json check fails on fresh Windows clone (missing .gitattributes) (#2437) 2026-07-03 20:14:55 -07:00
test_claude_provider.py fix(llm/providers/claude): attach cache_control to system block, not top-level (#2515) 2026-07-17 16:04:32 -04:00
test_executor.py Small changes 2026-04-12 12:34:45 +05:30
test_invariant_runner.py fix: sanitize subprocess call in runner.py (#2149) 2026-06-15 13:49:45 -04:00
test_provider_tools.py fix: harden openai-compatible provider responses 2026-05-18 01:04:28 -04:00
test_resolver.py feat: add Atlas Cloud as LLM/AI provider (#2279) 2026-06-18 16:29:11 -04:00
test_selector.py feat: connect ECC to canonical Ito compute CLI (#2558) 2026-07-23 19:28:55 -07:00
test_templates.py docs: salvage focused stale PR contributions 2026-05-11 05:31:12 -04:00
test_types.py feat: add Astraflow provider support 2026-05-11 23:21:46 -04:00