Add Codex Responses support and session debugging

Teach provider config and auth resolution about the Codex Responses
dialect, including user-facing `style = "openai_responses"` with
`dialect = "codex"`. Serialize and parse Responses traffic with
provider-specific reasoning replay, assistant phase metadata, and robust
function-call assembly keyed by `output_index` so streamed tool inputs
survive proxy quirks and empty terminal payloads.

Also persist thinking origins and message metadata across sessions, add
the Anthropic interleaved-thinking header switch, write per-session
debug logs, and improve the TUI and scripts for inspecting tool output
and session costs.

1 files changed, 38 insertions, 2 deletions

diff --git a/src/models_toml.zig b/src/models_toml.zig
index 955640d..622bdf6 100644
--- a/src/models_toml.zig
+++ b/src/models_toml.zig
@@ -21,7 +21,8 @@
 //!
 //!     # anthropic_messages-only knobs:
 //!     api_version          = <string>  # Anthropic-Version header; default "2023-06-01"
-//!     thinking             = <string>  # disabled | enabled | adaptive  (default: disabled)
+//!     thinking             = <string>  # disabled | enabled | adaptive  (default: disabled,
+//!                                      # or adaptive when `effort` is present)
 //!     effort               = <string>  # low | medium | high | xhigh | max  (default: medium)
 //!                                      # only used when thinking = "adaptive"
 //!     thinking_budget_tokens = <int>   # max reasoning tokens for thinking = "enabled"
@@ -268,13 +269,15 @@ fn ingestModel(
         break :blk null;
     };
 
+    const has_effort = v.get("effort") != null;
+
     const thinking: Thinking = blk: {
         if (v.get("thinking")) |t| {
             if (t.asString()) |s| {
                 break :blk std.meta.stringToEnum(Thinking, s) orelse .disabled;
             }
         }
-        break :blk .disabled;
+        break :blk if (has_effort) .adaptive else .disabled;
     };
 
     const effort: Effort = blk: {
@@ -527,6 +530,39 @@ test "parseInto: anthropic adaptive thinking with effort" {
     try testing.expectEqual(false, def.thinking_interleaved); // default
 }
 
+test "parseInto: anthropic effort implies adaptive thinking" {
+    var models = emptyModels(testing.allocator);
+    defer models.deinit();
+
+    const src =
+        \\[anthropic.opus]
+        \\model  = "claude-opus-4-8"
+        \\effort = "high"
+    ;
+    try parseInto(&models, src);
+
+    const def = models.defs.get("anthropic", "opus").?;
+    try testing.expectEqual(Thinking.adaptive, def.thinking);
+    try testing.expectEqual(Effort.high, def.effort);
+}
+
+test "parseInto: explicit thinking overrides effort-implied adaptive thinking" {
+    var models = emptyModels(testing.allocator);
+    defer models.deinit();
+
+    const src =
+        \\[anthropic.haiku]
+        \\model    = "claude-haiku-4-5-20251001"
+        \\thinking = "disabled"
+        \\effort   = "high"
+    ;
+    try parseInto(&models, src);
+
+    const def = models.defs.get("anthropic", "haiku").?;
+    try testing.expectEqual(Thinking.disabled, def.thinking);
+    try testing.expectEqual(Effort.high, def.effort);
+}
+
 test "parseInto: anthropic thinking disabled" {
     var models = emptyModels(testing.allocator);
     defer models.deinit();