Issues

• tooldefer: hosted OpenAI tool search for Responses clients that have no search of their own

  priority: low
  rossoctl/context-guru#932 · Filter by author amiddavid opened on Nov 5, 2026

• tooldefer: measure Codex's additional_tools item on gpt-5.6 (about 23 KB a request)

  priority: high
  rossoctl/context-guru#931 · Filter by author amiddavid opened on Nov 5, 2026

• keepalive: own-tool waits outside strategy windows are often not pinged, after #858

  rossoctl/context-guru#922 · Filter by author OsherElhadad opened on Nov 4, 2026
  1

• Dashboard: the first request of a session books cost $0 (token_accounting=partial), so total cost under-counts the bill

  rossoctl/context-guru#920 · Filter by author boazdavid opened on Nov 4, 2026
  1

  

• Accountant integration: keepalive_saved_usd is credited by the library and re-gated by a separate SQL model; use one

  rossoctl/context-guru#919 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration: the keep-alive ping pricing has no recover, so a panic would end the process

  rossoctl/context-guru#918 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration (#893): state file has no proxy-side version; save is not durable; nested cache_control still moves the hash

  rossoctl/context-guru#917 · Filter by author boazdavid opened on Nov 4, 2026
  

• Dashboard: /api/stats cache key rounds since/until to 30 s, so a narrower window returns the wider window's totals

  rossoctl/context-guru#916 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration (#893 Major 5): the first request on an unseen warm cache books a baseline about 2x too high

  rossoctl/context-guru#915 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration (#893 Major 4): saved tokens are booked on a request the provider rejected

  rossoctl/context-guru#914 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration (#893 B3): a manual ping stores no facts, so the next request may be classified cold although the cache was read

  rossoctl/context-guru#913 · Filter by author boazdavid opened on Nov 4, 2026
  1

  

• Accountant integration (#893 B4): the keep-alive rescue gate needs a component row, so a keep-alive-only setup still loses the rescue

  rossoctl/context-guru#912 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration (#893 B2): the double-count fix also drops measured tool-filter requests and tokens, and items no longer sum to the total

  rossoctl/context-guru#911 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration: summarize books its model cost but no saving (ops_dropped)

  rossoctl/context-guru#910 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration: in-session cachesplit credit is $0 on the live path (pre-#768 credited it)

  rossoctl/context-guru#909 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration: tool defer fails check_sum and conservation on every request

  rossoctl/context-guru#908 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration: the plausibility gate drops the whole result when removed tool declarations exceed message tokens

  rossoctl/context-guru#907 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration: the default tail breakpoint makes every request fall back (check_sum) and book no saving

  rossoctl/context-guru#906 · Filter by author boazdavid opened on Nov 4, 2026
  1

  

• keepalive: a proxy restart drops pending ping plans, so waiting threads lose their cache

  priority: low
  rossoctl/context-guru#905 · Filter by author amiddavid opened on Nov 4, 2026

• Pricing: apply long-context tariffs (above 100k and 200k prompt tokens)

  rossoctl/context-guru#901 · Filter by author boazdavid opened on Nov 4, 2026
  

• Accountant integration: price the OpenAI Responses wire (every saving is $0 there)

  rossoctl/context-guru#899 · Filter by author boazdavid opened on Nov 4, 2026
  

• ReadSearchesSSE counts searches on the Anthropic wire only

  rossoctl/context-guru#895 · Filter by author amiddavid opened on Nov 4, 2026

• Prices above 100k/200k prompt tokens are ignored: long requests are costed at the base rate

  bugdashboard
  rossoctl/context-guru#892 · Filter by author OsherElhadad opened on Nov 4, 2026

• dash: mark estimated savings in the request list and the totals (follow-up to #768)

  rossoctl/context-guru#890 · Filter by author amiddavid opened on Nov 4, 2026

• Integrate the proxy with a central metrics server (OTel push to Prometheus)

  rossoctl/context-guru#886 · Filter by author boazdavid opened on Nov 4, 2026