mirrors/codex
1
0
Fork 0
mirror of https://github.com/openai/codex.git synced 2026-05-03 04:42:20 +03:00
Code Issues Packages Projects Releases Wiki Activity

feat: Budget skill metadata and surface trimming as a warning (#18298)

Browse Source 
Cap the model-visible skills section to a small share of the context
window, with a fallback character budget, and keep only as many implicit
skills as fit within that budget.

Emit a non-fatal warning when enabled skills are omitted, and add a new
app-server warning notification

Record thread-start skill metrics for total enabled skills, kept skills,
and whether truncation happened

---------

Co-authored-by: Matthew Zeng <mzeng@openai.com>
Co-authored-by: Codex <noreply@openai.com>
This commit is contained in:
xl-openai
2026-04-17 18:11:47 -07:00
committed by GitHub
parent a58a0f083d
commit 3f7222ec76
26 changed files with 883 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
codex-rs/app-server/README.md
View File

@@ -958,6 +958,10 @@ Event notifications are the server-initiated event stream for thread lifecycles,
Thread realtime uses a separate thread-scoped notification surface. `thread/realtime/*` notifications are ephemeral transport events, not `ThreadItem`s, and are not returned by `thread/read`, `thread/resume`, or `thread/fork`.
Recoverable configuration and initialization warnings use the existing `configWarning` notification: `{ summary, details?, path?, range? }`. App-server may emit it during initialization for config parsing and related setup diagnostics.
Generic runtime warnings use the `warning` notification: `{ threadId?, message }`. App-server emits this for non-fatal warnings from the core event stream, including cases where not all enabled skills are included in the model-visible skills list for a session.
### Notification opt-out
Clients can suppress specific notifications per connection by sending exact method names in `initialize.params.capabilities.optOutNotificationMethods`.