Commit Graph
8 Commits
Author SHA1 Message Date
BaaRai 60728092e1 fix(db): stop billing cache writes twice in profile cost
Norma normalizes InputTokens to input + cache read + cache write, so both cache buckets are subsets of it. The fresh share billed at the input price is now input minus reads minus writes, still clamped at zero for gateways that report disjoint buckets.
2026-11-15 18:17:33 +01:00
BaaRai 1da0d07222 db: versioned migrations runner with schema_migrations
Move the llm_records and llm_usage DDL out of Go constants into embedded
files db/migrations/0001_llm_records.sql and 0002_llm_usage.sql (byte-for-byte
the old constants). Open now applies pending migrations after schema.sql and
before seedBuiltins, under the existing schema advisory lock and on the
lock-holding connection. Each migration runs in its own transaction with its
schema_migrations row (version, name, SHA-256 checksum, applied_at).

Open fails closed when the database has a version missing from the binary, or
an applied migration whose file name or checksum changed. A failing migration
is rolled back and blocks startup; earlier ones stay applied.

EnsureLLMRecordsTable and EnsureLLMUsageTable re-run the embedded files and
keep their transaction-level schema lock; Open never calls them.
2026-11-14 12:52:34 +01:00
BaaRai c96c3ea3a9 fix: cost ledger charges fresh input only (cache read is part of input tokens)
input_tokens includes the cached subset under Anthropic and OpenAI usage semantics, so pricing input and cache read at their full prices double-counted the cache. The fresh share is now input minus cache reads, clamped at zero for gateways reporting disjoint buckets.
2026-10-12 08:53:20 +02:00
BaaRai 67ff6ffffa feat: cost ledger (per-profile EUR pricing on token usage)
The token metering is real now (input, output, cache read/write per
profile), but it answers "how many tokens" while the question an operator
actually asks is "what did this run cost". Converting requires per-profile
prices, which nowhere exist.

Changes:
- llm_profiles gains four optional price columns (EUR per 1M tokens:
  input, output, cache read, cache write; 0/NULL = unset), migrated
  idempotently and round-tripped through SaveProfile / the profile API
  in one small auditable write
- the usage aggregation computes cost_eur per profile from the stored
  prices, in Go after the archived-task merge so cold-storage tokens are
  priced identically; profiles without prices keep cost_eur nil instead
  of showing a fake zero
- web types carry cost_eur and the price fields; the dashboard card and
  profile-dialog inputs consume them in a follow-up (the API serves the
  data already)
- integration test: priced profile + usage rows -> expected EUR to the
  cent (2M in @1.5 + 0.5M out @4 + 1M cache-read @0.15 = 5.15), plus a
  prices round-trip check

Note: the Go side activates on the next build; the schema migration is
idempotent at startup.
2026-10-12 08:20:37 +02:00
Autumn-27 8bc4d58c85 feat(intercept): 模型兜底审批单独计量 token 并在配置页展示
兜底审批(LLM judge)调模型判定漏网命令,其 token 消耗此前无法单独查看:
usage 行要么无 worker 归属、要么被并进复用的模型 profile 桶里。

- llmrec 增加 WithWorker 上下文覆盖,reviewer 调用前标注 worker='judge',
  让审批调用的计量行可被单独筛出(复用现有 llm_usage.worker 列,无需迁移)。
- db.JudgeUsageStats 返回累计总量 + 近 N 天每日序列(WHERE worker='judge')。
- 新增 GET /api/intercept/judge/usage。
- 配置页「模型兜底审批」开关下方新增用量卡片:调用次数 / 输入 / 输出 /
  缓存读写 五项统计 + 近 30 天每日消耗迷你柱图 + 刷新。
2026-10-04 02:43:26 -04:00
pphua 7df4406d44 fix: prevent OOM when archiving large tasks 2026-08-28 15:25:20 +08:00
pphua 809e94f6e4 归档 2026-08-28 12:56:29 +08:00
pphua ddb09287a7 feat(metrics): add persistent LLM usage analytics 2026-08-24 19:30:07 +08:00