Add body_after_prefix auto-compact token limit scope

codex-rs/app-server-protocol/src/protocol/v2/config.rs

@@ -3,6 +3,7 @@ use super::AskForApproval;
 use super::SandboxMode;
 use super::shared::default_enabled;
 use codex_experimental_api_macros::ExperimentalApi;
+use codex_protocol::config_types::AutoCompactTokenLimitScope;
 use codex_protocol::config_types::ForcedLoginMethod;
 use codex_protocol::config_types::ReasoningSummary;
 use codex_protocol::config_types::Verbosity;
@@ -251,6 +252,7 @@ pub struct Config {
     pub review_model: Option<String>,
     pub model_context_window: Option<i64>,
     pub model_auto_compact_token_limit: Option<i64>,
+    pub model_auto_compact_token_limit_scope: Option<AutoCompactTokenLimitScope>,
     pub model_provider: Option<String>,
     #[experimental(nested)]
     pub approval_policy: Option<AskForApproval>,

codex-rs/app-server-protocol/src/protocol/v2/tests.rs

@@ -1511,6 +1511,7 @@ fn config_granular_approval_policy_is_marked_experimental() {
         review_model: None,
         model_context_window: None,
         model_auto_compact_token_limit: None,
+        model_auto_compact_token_limit_scope: None,
         model_provider: None,
         approval_policy: Some(AskForApproval::Granular {
             sandbox_approval: false,
@@ -1551,6 +1552,7 @@ fn config_approvals_reviewer_is_marked_experimental() {
         review_model: None,
         model_context_window: None,
         model_auto_compact_token_limit: None,
+        model_auto_compact_token_limit_scope: None,
         model_provider: None,
         approval_policy: None,
         approvals_reviewer: Some(ApprovalsReviewer::AutoReview),
@@ -1585,6 +1587,7 @@ fn config_nested_profile_granular_approval_policy_is_marked_experimental() {
         review_model: None,
         model_context_window: None,
         model_auto_compact_token_limit: None,
+        model_auto_compact_token_limit_scope: None,
         model_provider: None,
         approval_policy: None,
         approvals_reviewer: None,
@@ -1641,6 +1644,7 @@ fn config_nested_profile_approvals_reviewer_is_marked_experimental() {
         review_model: None,
         model_context_window: None,
         model_auto_compact_token_limit: None,
+        model_auto_compact_token_limit_scope: None,
         model_provider: None,
         approval_policy: None,
         approvals_reviewer: None,

codex-rs/config/src/config_toml.rs

@@ -38,6 +38,7 @@ use codex_model_provider_info::ModelProviderInfo;
 use codex_model_provider_info::OLLAMA_CHAT_PROVIDER_REMOVED_ERROR;
 use codex_model_provider_info::OLLAMA_OSS_PROVIDER_ID;
 use codex_model_provider_info::OPENAI_PROVIDER_ID;
+use codex_protocol::config_types::AutoCompactTokenLimitScope;
 use codex_protocol::config_types::ForcedLoginMethod;
 use codex_protocol::config_types::Personality;
 use codex_protocol::config_types::ReasoningSummary;
@@ -156,6 +157,10 @@ pub struct ConfigToml {
     /// Token usage threshold triggering auto-compaction of conversation history.
     pub model_auto_compact_token_limit: Option<i64>,
 
+    /// Controls whether the auto-compaction limit applies to the full context or
+    /// only to tokens after the carried prefix in the current compaction window.
+    pub model_auto_compact_token_limit_scope: Option<AutoCompactTokenLimitScope>,
+
     /// Default approval policy for executing commands.
     pub approval_policy: Option<AskForApproval>,
 

codex-rs/core-api/src/lib.rs

@@ -58,6 +58,7 @@ pub use codex_models_manager::manager::SharedModelsManager;
 pub use codex_protocol::ThreadId;
 pub use codex_protocol::config_types::AltScreenMode;
 pub use codex_protocol::config_types::ApprovalsReviewer;
+pub use codex_protocol::config_types::AutoCompactTokenLimitScope;
 pub use codex_protocol::config_types::CollaborationModeMask;
 pub use codex_protocol::config_types::ShellEnvironmentPolicy;
 pub use codex_protocol::config_types::WebSearchMode;

codex-rs/core/config.schema.json

@@ -309,6 +309,25 @@
         }
       ]
     },
+    "AutoCompactTokenLimitScope": {
+      "description": "Selects which part of the active context is charged against `model_auto_compact_token_limit`.",
+      "oneOf": [
+        {
+          "description": "Count the full active context against the limit.",
+          "enum": [
+            "total"
+          ],
+          "type": "string"
+        },
+        {
+          "description": "Count sampled output and later growth after the carried window prefix.",
+          "enum": [
+            "body_after_prefix"
+          ],
+          "type": "string"
+        }
+      ]
+    },
     "AutoReviewToml": {
       "properties": {
         "policy": {
@@ -4544,6 +4563,14 @@
       "format": "int64",
       "type": "integer"
     },
+    "model_auto_compact_token_limit_scope": {
+      "allOf": [
+        {
+          "$ref": "#/definitions/AutoCompactTokenLimitScope"
+        }
+      ],
+      "description": "Controls whether the auto-compaction limit applies to the full context or only to tokens after the carried prefix in the current compaction window."
+    },
     "model_catalog_json": {
       "allOf": [
         {

codex-rs/core/src/client.rs

@@ -931,7 +931,7 @@ impl Drop for ModelClientSession {
 }
 
 impl ModelClientSession {
-    fn reset_websocket_session(&mut self) {
+    pub(crate) fn reset_websocket_session(&mut self) {
         self.websocket_session.connection = None;
         self.websocket_session.last_request = None;
         self.websocket_session.last_response_rx = None;

codex-rs/core/src/config/config_tests.rs

@@ -7711,6 +7711,7 @@ async fn test_precedence_fixture_with_o3_profile() -> std::io::Result<()> {
             review_model: None,
             model_context_window: None,
             model_auto_compact_token_limit: None,
+            model_auto_compact_token_limit_scope: AutoCompactTokenLimitScope::Total,
             service_tier: None,
             model_provider_id: "openai".to_string(),
             model_provider: fixture.openai_provider.clone(),
@@ -8164,6 +8165,7 @@ async fn test_precedence_fixture_with_gpt3_profile() -> std::io::Result<()> {
         review_model: None,
         model_context_window: None,
         model_auto_compact_token_limit: None,
+        model_auto_compact_token_limit_scope: AutoCompactTokenLimitScope::Total,
         service_tier: None,
         model_provider_id: "openai-custom".to_string(),
         model_provider: fixture.openai_custom_provider.clone(),
@@ -8331,6 +8333,7 @@ async fn test_precedence_fixture_with_zdr_profile() -> std::io::Result<()> {
         review_model: None,
         model_context_window: None,
         model_auto_compact_token_limit: None,
+        model_auto_compact_token_limit_scope: AutoCompactTokenLimitScope::Total,
         service_tier: None,
         model_provider_id: "openai".to_string(),
         model_provider: fixture.openai_provider.clone(),
@@ -8483,6 +8486,7 @@ async fn test_precedence_fixture_with_gpt5_profile() -> std::io::Result<()> {
         review_model: None,
         model_context_window: None,
         model_auto_compact_token_limit: None,
+        model_auto_compact_token_limit_scope: AutoCompactTokenLimitScope::Total,
         service_tier: None,
         model_provider_id: "openai".to_string(),
         model_provider: fixture.openai_provider.clone(),

codex-rs/core/src/config/mod.rs

@@ -77,6 +77,7 @@ use codex_model_provider_info::built_in_model_providers;
 use codex_model_provider_info::merge_configured_model_providers;
 use codex_models_manager::ModelsManagerConfig;
 use codex_protocol::config_types::AltScreenMode;
+use codex_protocol::config_types::AutoCompactTokenLimitScope;
 use codex_protocol::config_types::ForcedLoginMethod;
 use codex_protocol::config_types::Personality;
 use codex_protocol::config_types::ReasoningSummary;
@@ -527,6 +528,10 @@ pub struct Config {
     /// Token usage threshold triggering auto-compaction of conversation history.
     pub model_auto_compact_token_limit: Option<i64>,
 
+    /// Controls whether `model_auto_compact_token_limit` applies to the full
+    /// active context or only tokens after the carried compaction-window prefix.
+    pub model_auto_compact_token_limit_scope: AutoCompactTokenLimitScope,
+
     /// Key into the model_providers map that specifies which provider to use.
     pub model_provider_id: String,
 
@@ -3373,6 +3378,9 @@ impl Config {
             review_model,
             model_context_window: cfg.model_context_window,
             model_auto_compact_token_limit: cfg.model_auto_compact_token_limit,
+            model_auto_compact_token_limit_scope: cfg
+                .model_auto_compact_token_limit_scope
+                .unwrap_or_default(),
             model_provider_id,
             model_provider,
             cwd: resolved_cwd,