mirror of
https://github.com/openai/codex.git
synced 2026-11-14 21:40:39 +00:00
felixxia-oai
f6160ca5b3
## Why Rendering host and executor skill catalogs independently allows their combined metadata to exceed the skills context budget. ## What changed - Allocate one budget across both catalogs when both contain model-visible skills. - Prefer retaining executor skills when budget pressure requires omissions, while preserving an executor omission marker and reporting when all host skills were omitted. - Evaluate absolute and aliased host paths under the shared budget and select the rendering that best preserves executor entries, total entries, and descriptions, in that order. ## Testing Added renderer and production-turn coverage for shared-budget description shortening, host-first omission, executor omission markers, and alias selection. GitOrigin-RevId: 30041736a76aa173cb641aa98a876a306cb437d2
102 lines
3.0 KiB
Rust
102 lines
3.0 KiB
Rust
use codex_extension_api::ExtensionMetrics;
|
|
use codex_otel::THREAD_SKILLS_DESCRIPTION_TRUNCATED_CHARS_METRIC;
|
|
use codex_otel::THREAD_SKILLS_ENABLED_TOTAL_METRIC;
|
|
use codex_otel::THREAD_SKILLS_KEPT_TOTAL_METRIC;
|
|
use codex_otel::THREAD_SKILLS_TRUNCATED_METRIC;
|
|
|
|
use crate::render::SkillMetadataBudget;
|
|
use crate::render::SkillRenderReport;
|
|
|
|
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
|
|
pub(crate) enum CatalogSurface {
|
|
ThreadContext,
|
|
ExecutorWorldState,
|
|
HostWorldState,
|
|
TurnInput,
|
|
}
|
|
|
|
impl CatalogSurface {
|
|
fn as_str(self) -> &'static str {
|
|
match self {
|
|
Self::ThreadContext => "thread_context",
|
|
Self::ExecutorWorldState => "executor_world_state",
|
|
Self::HostWorldState => "host_world_state",
|
|
Self::TurnInput => "turn_input",
|
|
}
|
|
}
|
|
}
|
|
|
|
pub(crate) fn record_catalog_render(
|
|
extension_metrics: Option<&dyn ExtensionMetrics>,
|
|
catalog_surface: CatalogSurface,
|
|
budget: SkillMetadataBudget,
|
|
report: &SkillRenderReport,
|
|
) {
|
|
record_catalog_metrics(
|
|
extension_metrics,
|
|
catalog_surface,
|
|
report.total_count,
|
|
report.included_count,
|
|
report.omitted_count,
|
|
report.truncated_description_chars,
|
|
);
|
|
|
|
trace_catalog_budget_pressure(budget, report);
|
|
}
|
|
|
|
pub(crate) fn trace_catalog_budget_pressure(
|
|
budget: SkillMetadataBudget,
|
|
report: &SkillRenderReport,
|
|
) {
|
|
if report.omitted_count > 0 || report.truncated_description_chars > 0 {
|
|
tracing::info!(
|
|
budget_limit = budget.limit(),
|
|
total_skills = report.total_count,
|
|
included_skills = report.included_count,
|
|
omitted_skills = report.omitted_count,
|
|
truncated_description_chars_per_skill = report.average_truncated_description_chars(),
|
|
truncated_skill_descriptions = report.truncated_description_count,
|
|
"truncated skill metadata to fit skills context budget"
|
|
);
|
|
}
|
|
}
|
|
|
|
pub(crate) fn record_catalog_metrics(
|
|
extension_metrics: Option<&dyn ExtensionMetrics>,
|
|
catalog_surface: CatalogSurface,
|
|
total_count: usize,
|
|
included_count: usize,
|
|
omitted_count: usize,
|
|
truncated_description_chars: usize,
|
|
) {
|
|
let Some(extension_metrics) = extension_metrics else {
|
|
return;
|
|
};
|
|
let samples = [
|
|
(
|
|
THREAD_SKILLS_ENABLED_TOTAL_METRIC,
|
|
i64::try_from(total_count).unwrap_or(i64::MAX),
|
|
),
|
|
(
|
|
THREAD_SKILLS_KEPT_TOTAL_METRIC,
|
|
i64::try_from(included_count).unwrap_or(i64::MAX),
|
|
),
|
|
(
|
|
THREAD_SKILLS_TRUNCATED_METRIC,
|
|
if omitted_count > 0 { 1 } else { 0 },
|
|
),
|
|
(
|
|
THREAD_SKILLS_DESCRIPTION_TRUNCATED_CHARS_METRIC,
|
|
i64::try_from(truncated_description_chars).unwrap_or(i64::MAX),
|
|
),
|
|
];
|
|
let tags = [("catalog_surface", catalog_surface.as_str())];
|
|
for (name, value) in samples {
|
|
extension_metrics.histogram(name, value, &tags);
|
|
}
|
|
}
|
|
|
|
#[cfg(test)]
|
|
#[path = "render_observability_tests.rs"]
|
|
mod tests;
|