fix(codex): context meter tracks the full-history resend + recover from context-window rejections (#174)

src/agent_compact.zig

@@ -252,6 +252,11 @@ pub fn compact(self: *Agent) anyerror!usize {
     // (WriteFailed) rather than returning a clean overflow, so compaction could
     // never run once near the cap. Old tool outputs are superseded by the summary
     // anyway; truncating them keeps the request sendable + all pairing intact.
+    // #174: on the Responses path, prior-turn reasoning items go first — they
+    // dominate the resend bloat on long high-effort sessions, and the backend
+    // itself discards them from chained context, so dropping them can't lose
+    // anything the server would have kept.
+    _ = dropPriorTurnReasoning(self);
     _ = trimOldestToolOutputs(self);
     try self.messages.append(try textMessage(self.arena, "user", compact_instruction));
     errdefer _ = self.messages.pop();
@@ -305,6 +310,66 @@ pub fn cleanUserTurn(m: Value) bool {
     }
 }
 
+/// #174: drop Responses `reasoning` items older than the last user message,
+/// in place (no allocation). These are exactly the items the backend discards
+/// from chained context (previous_response_id), so removing them can't lose
+/// anything the server would have kept — and on a long high-effort session
+/// their encrypted blobs dominate the full-resend size. Reasoning at or after
+/// the last user message stays: the API requires the current turn's reasoning
+/// between a function_call and its output. Returns how many were dropped.
+pub fn dropPriorTurnReasoning(self: *Agent) usize {
+    if (self.provider.kind != .responses) return 0;
+    var last_user: usize = 0;
+    for (self.messages.items, 0..) |m, i| {
+        if (m != .object) continue;
+        const role = m.object.get("role") orelse continue;
+        if (role == .string and std.mem.eql(u8, role.string, "user")) last_user = i;
+    }
+    var w: usize = 0;
+    for (self.messages.items, 0..) |m, i| {
+        const old_reasoning = i < last_user and m == .object and blk: {
+            const t = m.object.get("type") orelse break :blk false;
+            break :blk t == .string and std.mem.eql(u8, t.string, "reasoning");
+        };
+        if (old_reasoning) continue;
+        self.messages.items[w] = m;
+        w += 1;
+    }
+    const dropped = self.messages.items.len - w;
+    self.messages.shrinkRetainingCapacity(w);
+    return dropped;
+}
+
+test "dropPriorTurnReasoning (#174): prior-turn reasoning goes, current turn + non-responses stay" {
+    var arena_state = std.heap.ArenaAllocator.init(std.testing.allocator);
+    defer arena_state.deinit();
+    const a = arena_state.allocator();
+
+    var msgs = std.json.Array.init(a);
+    try msgs.append(try textMessage(a, "user", "turn one"));
+    try msgs.append(try std.json.parseFromSliceLeaky(Value, a, "{\"type\":\"reasoning\",\"encrypted_content\":\"OLD1\"}", .{}));
+    try msgs.append(try textMessage(a, "assistant", "reply one"));
+    try msgs.append(try std.json.parseFromSliceLeaky(Value, a, "{\"type\":\"reasoning\",\"encrypted_content\":\"OLD2\"}", .{}));
+    try msgs.append(try textMessage(a, "user", "turn two"));
+    try msgs.append(try std.json.parseFromSliceLeaky(Value, a, "{\"type\":\"reasoning\",\"encrypted_content\":\"CURRENT\"}", .{}));
+    try msgs.append(try std.json.parseFromSliceLeaky(Value, a, "{\"type\":\"function_call\",\"name\":\"bash\",\"call_id\":\"c1\",\"arguments\":\"{}\"}", .{}));
+
+    var agent: Agent = undefined;
+    agent.provider = .{ .id = "codex", .kind = .responses, .auth = .bearer, .url = "", .api_key = "", .model = "gpt-5", .context = 100_000 };
+    agent.messages = msgs;
+
+    try std.testing.expectEqual(@as(usize, 2), dropPriorTurnReasoning(&agent));
+    try std.testing.expectEqual(@as(usize, 5), agent.messages.items.len);
+    // current-turn reasoning (after the last user message) survives, in order
+    const kept = agent.messages.items[3].object.get("encrypted_content").?.string;
+    try std.testing.expectEqualStrings("CURRENT", kept);
+
+    // non-responses providers are untouched
+    agent.provider.kind = .openai;
+    try std.testing.expectEqual(@as(usize, 0), dropPriorTurnReasoning(&agent));
+    try std.testing.expectEqual(@as(usize, 5), agent.messages.items.len);
+}
+
 /// Index to cut history at for an emergency trim: the first clean user turn
 /// at or after the midpoint, so messages[cut..] is always a valid
 /// conversation start (never an orphaned tool_result). null when there is no

src/agent_request.zig

@@ -71,6 +71,22 @@ test "isAuthError (#148): auth failures only, not credits/rate/other" {
     try std.testing.expect(!isAuthError("context length exceeded"));
 }
 
+test "fullInputEstimateTokens (#174): counts retained reasoning the chained usage never reports" {
+    var arena_state = std.heap.ArenaAllocator.init(std.testing.allocator);
+    defer arena_state.deinit();
+    const a = arena_state.allocator();
+    var msgs = std.json.Array.init(a);
+    var agent: Agent = undefined;
+    agent.messages = msgs;
+    try std.testing.expectEqual(@as(u64, 0), fullInputEstimateTokens(&agent) / 100); // empty history ≈ nothing
+    // a fat encrypted-reasoning item — exactly what a WS-chained total_tokens excludes
+    const blob = "{\"type\":\"reasoning\",\"encrypted_content\":\"" ++ ("A" ** 8192) ++ "\"}";
+    try msgs.append(try std.json.parseFromSliceLeaky(Value, a, blob, .{}));
+    agent.messages = msgs;
+    const est = fullInputEstimateTokens(&agent);
+    try std.testing.expect(est > 2000); // ~8KB serialized / 4 bytes-per-token
+}
+
 pub fn request(self: *Agent, tools: ?[]const u8) !std.json.ObjectMap {
     var force = self.strict and tools != null;
     var stream_usage = true; // openai stream_options; dropped if rejected
@@ -223,6 +239,15 @@ pub fn request(self: *Agent, tools: ?[]const u8) !std.json.ObjectMap {
                         if (self.tracer) |tr| tr.note("ws", "server rejected previous_response_id — re-anchoring with full input");
                         continue :rebuild;
                     }
+                    // #174: a context-window rejection means the true input
+                    // size blew past the wall while the chained meter lagged —
+                    // and the rejected request never returns usage to correct
+                    // it, so the meter would stay stuck under compact@ and the
+                    // session would wedge (every retry resends the same
+                    // oversized history). Pin the meter to the window so the
+                    // ApiError compact-and-recover path engages.
+                    if (std.mem.indexOf(u8, msg, "exceeds the context window") != null)
+                        self.last_context_tokens = self.provider.context;
                     if (self.tracer) |tr| tr.api(self.label, self.provider.model, ms, body.len, resp_body.len, 0, 0, true);
                     try self.sayApiError("codex api error: {s}", .{msg});
                     return error.ApiError;
@@ -425,6 +450,21 @@ pub fn errorMessage(obj: std.json.ObjectMap) ?[]const u8 {
     return null;
 }
 
+/// #174: ~4-bytes/token estimate of the FULL history serialized as Responses
+/// `input` items — the cost of the next full-history resend (runTurn closes
+/// the WS per turn, so every turn's first request replays everything). The
+/// chained WS usage can sit far below this: with previous_response_id the
+/// server discards prior-turn reasoning from context, while a resend pays for
+/// every retained encrypted reasoning item again. Counting discard writer —
+/// no allocation.
+pub fn fullInputEstimateTokens(self: *Agent) u64 {
+    var buf: [512]u8 = undefined;
+    var d: Io.Writer.Discarding = .init(&buf);
+    var s: std.json.Stringify = .{ .writer = &d.writer };
+    s.write(Value{ .array = self.messages }) catch return 0;
+    return d.fullCount() / 4;
+}
+
 pub fn recordUsageResponses(self: *Agent, response: std.json.ObjectMap, req_body_len: usize) void {
     self.last_cache_read = 0;
     // Fallback estimate (~4 bytes/token) from the serialized request body,
@@ -456,6 +496,15 @@ pub fn recordUsageResponses(self: *Agent, response: std.json.ObjectMap, req_body
             self.last_context_tokens = est;
         }
     }
+    // #174: the server's chained number undercounts what the next full-history
+    // resend will cost, and the resend is the request that gets rejected — so
+    // the meter must never sit below the local full-input estimate. Same
+    // correction codex CLI applies (get_non_last_reasoning_items_tokens).
+    // Without it a long Extra-high session reads "95k/270k" right up until the
+    // backend rejects the resend for exceeding the window, and auto-compaction
+    // (gated on this meter) never rescues it.
+    const full_est = fullInputEstimateTokens(self);
+    if (full_est > self.last_context_tokens) self.last_context_tokens = full_est;
     var cached: i64 = 0;
     if (u.get("input_tokens_details")) |d| if (d == .object) {
         cached = usageInt(d.object, "cached_tokens");