refactor: remove external product references from codebase

Cargo.toml

@@ -69,7 +69,7 @@ members = [
     "src/cortex-skills",
     "src/cortex-prompt-harness",
 
-    # CLI - Features (Phase 3 - Codex-inspired)
+    # CLI - Features (Phase 3)
     "src/cortex-collab",
     "src/cortex-network-proxy",
     "src/cortex-shell-snapshot",
@@ -228,7 +228,7 @@ cortex-ollama = { path = "src/cortex-ollama" }
 cortex-skills = { path = "src/cortex-skills" }
 cortex-prompt-harness = { path = "src/cortex-prompt-harness" }
 
-# Phase 3 - Codex-inspired crates
+# Phase 3 crates
 cortex-collab = { path = "src/cortex-collab" }
 cortex-network-proxy = { path = "src/cortex-network-proxy" }
 cortex-shell-snapshot = { path = "src/cortex-shell-snapshot" }

src/cortex-agents/src/control.rs

@@ -1,7 +1,7 @@
 //! Agent control for multi-agent operations.
 //!
 //! Provides a control-plane interface for managing agent threads,
-//! inspired by OpenAI Codex's AgentControl system.
+//! handling spawning, lifecycle, and inter-agent communication.
 
 use crate::AgentInfo;
 use std::collections::HashMap;

src/cortex-cli/src/run_cmd/mod.rs

@@ -1,6 +1,6 @@
 //! Run command for non-interactive Cortex CLI execution.
 //!
-//! Provides a comprehensive run command similar to OpenCode's run functionality:
+//! Provides a comprehensive run command for batch execution:
 //! - `cortex run [message..]` - Execute with a message prompt
 //! - `--command` - Execute a predefined command
 //! - `--continue/-c` - Continue the last session

src/cortex-collab/src/lib.rs

@@ -1,6 +1,6 @@
 //! Multi-agent collaboration system for Cortex CLI.
 //!
-//! This crate implements multi-agent collaboration features inspired by OpenAI Codex:
+//! This crate implements multi-agent collaboration features:
 //! - Agent spawning and lifecycle management
 //! - Inter-agent communication
 //! - Wait/sync mechanisms with timeout

src/cortex-common/src/model_presets/aliases.rs

@@ -38,10 +38,6 @@ pub const MODEL_ALIASES: &[ModelAlias] = &[
         alias: "o3",
         model: "openai/o3",
     },
-    ModelAlias {
-        alias: "codex",
-        model: "openai/gpt-5.2-codex",
-    },
     // Google models
     ModelAlias {
         alias: "gemini",
@@ -103,7 +99,6 @@ mod tests {
         assert_eq!(resolve_model_alias("opus"), "anthropic/claude-opus-4.5");
         assert_eq!(resolve_model_alias("gpt4"), "openai/gpt-4o");
         assert_eq!(resolve_model_alias("haiku"), "anthropic/claude-haiku-4.5");
-        assert_eq!(resolve_model_alias("codex"), "openai/gpt-5.2-codex");
         assert_eq!(resolve_model_alias("r1"), "deepseek/deepseek-r1");
     }
 

src/cortex-common/src/model_presets/presets.rs

@@ -633,15 +633,6 @@ pub const MODEL_PRESETS: &[ModelPreset] = &[
         supports_tools: true,
         supports_reasoning: false,
     },
-    ModelPreset {
-        id: "openai/gpt-5.2-codex",
-        name: "GPT-5.2 Codex (via Cortex) - DEFAULT",
-        provider: "cortex",
-        context_window: 256_000,
-        supports_vision: true,
-        supports_tools: true,
-        supports_reasoning: true,
-    },
     // Other Cortex models
     ModelPreset {
         id: "openai/gpt-4o",

src/cortex-core/src/widgets/brain.rs

@@ -7,7 +7,7 @@
 //! Uses sine wave radial expansion from center:
 //! - Wave expands outward from brain center
 //! - 2-second cycle (slow, hypnotic)
-//! - Green gradient (#00FFA3 Codex accent)
+//! - Green gradient (#00FFA3)
 //!
 //! # Example
 //! ```
@@ -62,7 +62,7 @@ const WAVE_SPEED: f32 = std::f32::consts::PI * 2.0;
 /// Wave scale: controls the "wavelength" of radial rings
 const WAVE_SCALE: f32 = 6.0;
 
-/// Codex accent green color
+/// Accent green color
 const ACCENT_GREEN: (u8, u8, u8) = (0x00, 0xFF, 0xA3);
 
 /// Dark green base color
@@ -194,7 +194,7 @@ impl Brain {
 
     /// Gets the style for a character based on wave intensity.
     ///
-    /// Creates a gradient from dark green to bright Codex green (#00FFA3).
+    /// Creates a gradient from dark green to bright green (#00FFA3).
     fn get_wave_style(&self, base_density: f32, wave: f32) -> Style {
         // Combine base density with wave for final brightness
         let brightness = base_density * (0.4 + wave * 0.6 * self.intensity);

src/cortex-engine/src/agent/orchestrator.rs

@@ -267,7 +267,7 @@ impl Orchestrator {
 
             // If no tool calls, check if we need to request a summary
             if tool_calls.is_empty() {
-                // OpenCode pattern: If the model finished with no text after doing tool work,
+                // If the model finished with no text after doing tool work,
                 // add a synthetic message asking for a summary and call the model one more time.
                 // This ensures we get a meaningful response instead of empty output.
                 let has_no_text = text.as_ref().map(|t| t.trim().is_empty()).unwrap_or(true);
@@ -277,7 +277,7 @@ impl Orchestrator {
                     // Mark that we've requested a summary to avoid infinite loop
                     ctx.summary_requested = true;
 
-                    // Add synthetic user message asking for summary (OpenCode pattern)
+                    // Add synthetic user message asking for summary
                     let _ = self
                         .context
                         .add_message(Message::user(
@@ -310,7 +310,7 @@ impl Orchestrator {
         }
 
         // Use the last text segment as the primary response for subagents
-        // This follows OpenCode's pattern of using findLast() for text parts
+        // Uses findLast() logic for text parts to get the final segment
         // If there's meaningful last_text_segment, prefer it; otherwise use full response
         let effective_response = if !last_text_segment.trim().is_empty() {
             // The last segment has content - use it as the response

src/cortex-engine/src/config/loader.rs

@@ -2,7 +2,7 @@
 //!
 //! Supports multiple configuration formats:
 //! - TOML (`.toml`) - Primary format
-//! - JSON with comments (`.json`, `.jsonc`) - OpenCode-compatible format
+//! - JSON with comments (`.json`, `.jsonc`) - JSONC format
 
 use std::path::{Path, PathBuf};
 
@@ -694,7 +694,7 @@ model = "sync-model"
     #[test]
     fn test_parse_json_config() {
         let json_content = r#"{
-            // OpenCode-style config with comments
+            // JSON with comments (JSONC) config
             "model": "claude-3-opus",
             "model_provider": "anthropic"
         }"#;

src/cortex-engine/src/config/types.rs

@@ -10,7 +10,7 @@ use crate::custom_command::CustomCommandConfig;
 use crate::plugin::{PluginConfigEntry, PluginSettings};
 
 /// Permission level for granular permission control.
-/// Follows OpenCode-style permission configuration.
+/// Supports three-tier allow/ask/deny permission model.
 #[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Serialize, Deserialize)]
 #[serde(rename_all = "lowercase")]
 pub enum PermissionLevel {
@@ -34,7 +34,7 @@ impl std::fmt::Display for PermissionLevel {
 }
 
 /// Permission configuration structure.
-/// Supports OpenCode-style granular permission control.
+/// Supports granular permission control per operation type.
 #[derive(Debug, Clone, Default, PartialEq, Deserialize)]
 pub struct PermissionConfig {
     /// Default permission level for edit operations.

src/cortex-engine/src/git_snapshot.rs

@@ -1,7 +1,6 @@
 //! Git-based snapshot for fast undo/redo tracking.
 //!
 //! Uses git's plumbing commands (write-tree, read-tree) for instant snapshots.
-//! Inspired by OpenCode's snapshot implementation.
 
 use std::path::{Path, PathBuf};
 use std::process::Command;

src/cortex-engine/src/mcp/oauth.rs

@@ -16,7 +16,7 @@ use tracing::{debug, info, warn};
 use crate::error::{CortexError, Result};
 use cortex_common::create_default_client;
 
-/// OAuth callback port - matches OpenCode's port for consistency.
+/// OAuth callback port for local redirect URI.
 pub const OAUTH_CALLBACK_PORT: u16 = 19876;
 
 /// OAuth callback path.

src/cortex-engine/src/sandbox/landlock.rs

@@ -3,7 +3,7 @@
 //! Uses the cortex-linux-sandbox wrapper binary to apply:
 //! - Landlock filesystem restrictions
 //! - Seccomp network filtering
-//! - Mount namespace for read-only .git/.codex protection
+//! - Mount namespace for read-only .git/.cortex protection
 
 use std::path::PathBuf;
 

src/cortex-engine/src/sandbox/policy.rs

@@ -11,7 +11,7 @@ use serde::{Deserialize, Serialize};
 pub struct WritableRoot {
     /// The writable root path.
     pub root: PathBuf,
-    /// Subpaths that should remain read-only (e.g., .git, .codex).
+    /// Subpaths that should remain read-only (e.g., .git, .cortex).
     pub read_only_subpaths: Vec<PathBuf>,
 }
 
@@ -24,9 +24,9 @@ impl WritableRoot {
         }
     }
 
-    /// Create a writable root with standard protected subpaths (.git, .codex, .cortex).
+    /// Create a writable root with standard protected subpaths (.git, .cortex).
     pub fn with_standard_protections(root: PathBuf) -> Self {
-        let read_only_subpaths = vec![root.join(".git"), root.join(".codex"), root.join(".cortex")];
+        let read_only_subpaths = vec![root.join(".git"), root.join(".cortex")];
         Self {
             root,
             read_only_subpaths,
@@ -212,7 +212,6 @@ impl ProtectedPaths {
         Self {
             read_only: vec![
                 workspace.join(".git"),
-                workspace.join(".codex"),
                 workspace.join(".cortex"),
             ],
             denied: vec![
@@ -253,7 +252,7 @@ mod tests {
 
         assert!(root.is_path_writable(Path::new("/workspace/src/main.rs")));
         assert!(!root.is_path_writable(Path::new("/workspace/.git/config")));
-        assert!(!root.is_path_writable(Path::new("/workspace/.codex/state")));
+        assert!(!root.is_path_writable(Path::new("/workspace/.cortex/state")));
         assert!(!root.is_path_writable(Path::new("/other/file.txt")));
     }
 

src/cortex-engine/src/sandbox/seatbelt.rs

@@ -2,7 +2,7 @@
 //!
 //! Uses sandbox-exec with dynamically generated SBPL policies to provide:
 //! - Filesystem isolation with writable workspace
-//! - .git/.codex protection via require-not clauses
+//! - .git/.cortex protection via require-not clauses
 //! - Optional network blocking
 
 use std::path::PathBuf;

src/cortex-engine/src/tools/handlers/edit_strategies/mod.rs

@@ -1,8 +1,8 @@
 //! Edit replacement strategies for robust code editing.
 //!
-//! This module implements 8 cascading replacement strategies inspired by OpenCode's
-//! edit tool implementation. Each strategy tries progressively more flexible matching
-//! approaches to find and replace text in source code.
+//! This module implements 8 cascading replacement strategies for flexible text matching.
+//! Each strategy tries progressively more flexible matching approaches to find and
+//! replace text in source code.
 //!
 //! Strategies (in cascade order):
 //! 1. SimpleReplacer - Exact string match

src/cortex-engine/src/tools/handlers/subagent/executor.rs

@@ -248,7 +248,7 @@ impl SubagentExecutor {
         };
 
         // Build system prompt - use base prompt WITHOUT task details
-        // Task details will be sent as user message (OpenCode pattern)
+        // Task details will be sent as user message
         let system_prompt = if let Some(ref agent) = custom_agent {
             // Use the custom agent's base system prompt
             agent.system_prompt.clone()
@@ -257,7 +257,7 @@ impl SubagentExecutor {
         };
 
         // Build user message containing the task
-        // This follows OpenCode's pattern where the task is a conversation, not system config
+        // Tasks are conversational - sent as user messages rather than system config
         let user_task_message = if let Some(ref _agent) = custom_agent {
             // For custom agents, format task with context
             let mut message = format!(
@@ -425,7 +425,7 @@ impl SubagentExecutor {
             config.working_dir.clone(),
         );
 
-        // Execute with optional timeout (like OpenCode, no timeout by default)
+        // Execute with optional timeout (no timeout by default)
         let mut turn_result = if let Some(timeout_duration) = config.effective_timeout() {
             // With timeout
             match timeout(timeout_duration, orchestrator.process_turn(&mut turn_ctx)).await {
@@ -437,7 +437,7 @@ impl SubagentExecutor {
                 }
             }
         } else {
-            // No timeout - run until completion (like OpenCode)
+            // No timeout - run until completion
             orchestrator.process_turn(&mut turn_ctx).await
         };
 
@@ -708,7 +708,7 @@ pub struct SubagentTypeInfo {
 
 /// Extract file path from tool output.
 /// Prompt used to request an explicit summary from a subagent when none was provided.
-/// This follows the Codex pattern of ensuring structured output from agents.
+/// Ensures structured output from agents for orchestrator consumption.
 const SUMMARY_REQUEST_PROMPT: &str = r#"You have completed your work but did not provide a summary. Please provide a final summary NOW using EXACTLY this format:
 
 ## Summary for Orchestrator

src/cortex-engine/src/tools/handlers/subagent/types.rs

@@ -81,7 +81,7 @@ impl SubagentType {
 
     /// Get the base system prompt for this subagent type (without task details).
     /// The task details should be sent as a user message, not in the system prompt.
-    /// This follows OpenCode's pattern where the task is a conversation, not configuration.
+    /// Tasks are conversational - passed as user messages rather than system configuration.
     pub fn base_system_prompt(&self) -> String {
         // Common planning instructions that all subagents must follow
         let planning_instructions = r#"
@@ -393,7 +393,7 @@ impl SubagentConfig {
     }
 
     /// Get effective timeout.
-    /// Returns None if no timeout is set (subagent runs until completion, like OpenCode).
+    /// Returns None if no timeout is set (subagent runs until completion).
     pub fn effective_timeout(&self) -> Option<Duration> {
         self.timeout
     }
@@ -420,7 +420,7 @@ impl SubagentConfig {
     }
 
     /// Build the user message containing the task.
-    /// This follows OpenCode's pattern of sending tasks as user messages.
+    /// Tasks are sent as user messages for conversational interaction.
     pub fn build_user_message(&self) -> String {
         let mut message = format!(
             "## Task\n{}\n\n## Instructions\n{}",

src/cortex-engine/src/tools/unified_executor.rs

@@ -520,7 +520,7 @@ impl BatchToolExecutor for UnifiedBatchExecutor {
     ) -> Result<ToolResult> {
         // Prevent recursive batch calls (already handled by BatchToolHandler validation)
         // Also prevent Task from being called in batch for safety
-        // (Task is allowed in OpenCode but we're being more conservative here)
+        // (being conservative to avoid complex nesting scenarios)
 
         self.registry
             .execute_with_context(name, arguments, context.clone())

src/cortex-gui/src/components/SearchEditor.tsx

@@ -1804,7 +1804,7 @@ export function useSearchEditor() {
     setIsOpen(true);
   };
 
-  const openCodeSearchFile = (filePath: string) => {
+  const openSearchFile = (filePath: string) => {
     setInitialQuery("");
     setInitialResults([]);
     setCodeSearchFilePath(filePath);
@@ -1832,7 +1832,7 @@ export function useSearchEditor() {
     initialResults,
     codeSearchFilePath,
     openSearchEditor,
-    openCodeSearchFile,
+    openSearchFile,
     closeSearchEditor,
     SearchEditorComponent: () => (
       <SearchEditor

src/cortex-gui/src/components/ai/SubagentsDialog.tsx

@@ -1,6 +1,6 @@
 /**
  * SubagentsDialog - Manage and spawn AI sub-agents
- * Inspired by OpenCode's subagent system
+ * Subagent management dialog for viewing and controlling active subagents
  */
 
 import {

src/cortex-linux-sandbox/src/mounts.rs

@@ -1,7 +1,7 @@
 //! Mount namespace operations.
 //!
 //! Provides bind-mount operations to make paths read-only within a new
-//! mount namespace. This is used to protect .git and .codex directories
+//! mount namespace. This is used to protect .git and .cortex directories
 //! from modification even within writable roots.
 
 use std::ffi::CString;

src/cortex-linux-sandbox/src/run_main.rs

@@ -101,7 +101,6 @@ pub fn run_main() -> ! {
             root: args.sandbox_policy_cwd.clone(),
             read_only_subpaths: vec![
                 args.sandbox_policy_cwd.join(".git"),
-                args.sandbox_policy_cwd.join(".codex"),
                 args.sandbox_policy_cwd.join(".cortex"),
             ],
         }]

src/cortex-network-proxy/src/lib.rs

@@ -1,6 +1,6 @@
 //! Network proxy with domain filtering and SSRF protection for Cortex CLI.
 //!
-//! This crate implements a network proxy inspired by OpenAI Codex:
+//! This crate implements a network proxy with:
 //! - Domain allowlist/denylist with wildcard patterns
 //! - SSRF protection (blocks private/local IPs)
 //! - Network mode control (Full/Limited/Disabled)

src/cortex-protocol/src/protocol/message_parts.rs

@@ -16,7 +16,7 @@ use super::tokens::TokenUsage;
 // ============================================================
 
 /// Rich message part types supporting various content kinds.
-/// Based on OpenCode's message part system for comprehensive message representation.
+/// Provides comprehensive message representation for structured content.
 #[derive(Debug, Clone, Serialize, Deserialize, JsonSchema)]
 #[serde(tag = "type", rename_all = "snake_case")]
 pub enum MessagePart {