Move host skill prompt injection into the skills extension (#37503)

## What changed

- Add `HostSkillsSnapshot::load_skill_prompts` so the skills extension owns reading and rendering selected host skills.
- Keep explicit invocation telemetry, unreadable-skill warnings, plugin prompt ordering, and suppression of host prompts superseded by provider prompts.
- Remove the legacy prompt injection and fragment implementation from `codex-core-skills`.

## Testing

- Cover prompt truncation without an installed skills extension, skill/plugin ordering, selective provider supersession, unreadable skills, and contextual fragment detection.

GitOrigin-RevId: 9c8b84ad54b90d174abc287ab49cc5a231e9e1aa

codex-rs/core/src/skills.rs

@@ -3,6 +3,7 @@ use crate::session::session::Session;
 use crate::session::turn_context::TurnContext;
 use codex_analytics::InvocationType;
 use codex_analytics::SkillInvocation;
+use codex_analytics::TrackEventsContext;
 use codex_analytics::build_track_events_context;
 use codex_extension_api::SkillInvocationInput;
 use codex_extension_api::SkillInvocationKind;
@@ -15,15 +16,12 @@ use tokio::sync::Mutex;
 
 pub use codex_core_skills::SkillError;
 pub use codex_core_skills::SkillLoadOutcome;
-pub use codex_core_skills::build_skill_name_counts;
-pub use codex_core_skills::detect_implicit_skill_invocation_for_command;
-pub use codex_core_skills::injection;
-pub use codex_core_skills::injection::SkillInjections;
-pub use codex_core_skills::injection::build_skill_injections;
 pub use codex_core_skills::model;
 pub use codex_skills::SkillMetadata;
 pub use codex_skills::SkillPolicy;
+pub use codex_skills::build_skill_name_counts;
 pub use codex_skills::collect_explicit_skill_mentions;
+pub use codex_skills::detect_implicit_skill_invocation_for_command;
 pub use codex_skills_extension::HostSkillsLoadInput;
 pub use codex_skills_extension::HostSkillsService;
 pub use codex_skills_extension::bundled_skills_enabled_from_stack;
@@ -43,6 +41,51 @@ pub(crate) fn skills_load_input_from_config(
     )
 }
 
+pub(crate) fn emit_explicit_skill_invocations(
+    sess: &Session,
+    turn_context: &TurnContext,
+    mentioned_skills: &[SkillMetadata],
+    injected_skills: &[SkillMetadata],
+    tracking: TrackEventsContext,
+) {
+    let injected_skill_paths = injected_skills
+        .iter()
+        .map(|skill| &skill.path_to_skills_md)
+        .collect::<HashSet<_>>();
+    for skill in mentioned_skills {
+        let skill_name_tag = sanitize_metric_tag_value(skill.name.as_str());
+        let status = if injected_skill_paths.contains(&skill.path_to_skills_md) {
+            "ok"
+        } else {
+            "error"
+        };
+        turn_context.session_telemetry.counter(
+            "codex.skill.injected",
+            /*inc*/ 1,
+            &[
+                ("status", status),
+                ("skill", skill_name_tag.as_str()),
+                ("invoke_type", "explicit"),
+            ],
+        );
+    }
+
+    let invocations = injected_skills
+        .iter()
+        .map(|skill| SkillInvocation {
+            skill_name: skill.name.clone(),
+            skill_scope: skill.scope,
+            skill_path: skill.path_to_skills_md.to_path_buf(),
+            plugin_id: skill.plugin_id.clone(),
+            remote_plugin_id: skill.remote_plugin_id.clone(),
+            invocation_type: InvocationType::Explicit,
+        })
+        .collect();
+    sess.services
+        .analytics_events_client
+        .track_skill_invocations(tracking, invocations);
+}
+
 pub(crate) async fn maybe_emit_implicit_skill_invocation(
     sess: &Session,
     turn_context: &TurnContext,