api: fix large output crash, truncate tool results >2k chars with sha256

Author: psyb0t

api_server.py

@@ -144,6 +144,9 @@ def _build_env():
     }
 
 
+_STREAM_LIMIT = 100 * 1024 * 1024  # 100MB — claude lines can be huge
+
+
 async def _stream(workspace: str, req: RunRequest):
     env = _build_env()
 
@@ -155,6 +158,7 @@ async def _stream(workspace: str, req: RunRequest):
                 stderr=asyncio.subprocess.STDOUT,
                 cwd=workspace,
                 env=env,
+                limit=_STREAM_LIMIT,
             )
             busy_workspaces[workspace] = proc
             if proc.stdout:
@@ -170,6 +174,7 @@ async def _stream(workspace: str, req: RunRequest):
             stderr=asyncio.subprocess.STDOUT,
             cwd=workspace,
             env=env,
+            limit=_STREAM_LIMIT,
         )
         busy_workspaces[workspace] = proc
 
@@ -187,6 +192,7 @@ async def _stream(workspace: str, req: RunRequest):
                 stderr=asyncio.subprocess.STDOUT,
                 cwd=workspace,
                 env=env,
+                limit=_STREAM_LIMIT,
             )
             busy_workspaces[workspace] = proc
             if proc.stdout:

jsonverbose.py

@@ -6,9 +6,12 @@
 that combines the final result with the full conversation history.
 """
 
+import hashlib
 import json
 import sys
 
+_CONTENT_TRUNCATE = 2000  # chars to keep before truncating
+
 
 def _extract_tool_uses(content):
     """Extract tool_use entries from assistant message content."""
@@ -36,26 +39,39 @@ def _extract_text(content):
     return out
 
 
+def _truncate_content(text: str) -> dict:
+    """Return content dict, truncating if over limit with sha256 + length."""
+    if len(text) <= _CONTENT_TRUNCATE:
+        return {"content": text}
+    sha = hashlib.sha256(text.encode()).hexdigest()
+    return {
+        "content": text[:_CONTENT_TRUNCATE],
+        "truncated": True,
+        "total_length": len(text),
+        "sha256": sha,
+    }
+
+
 def _extract_tool_results(content):
     """Extract tool_result entries from user message content."""
     out = []
     for block in content:
         if block.get("type") != "tool_result":
             continue
-        entry = {
-            "type": "tool_result",
-            "tool_use_id": block.get("tool_use_id", ""),
-            "is_error": block.get("is_error", False),
-        }
         raw = block.get("content", "")
         if isinstance(raw, str):
-            entry["content"] = raw
+            text = raw
         elif isinstance(raw, list):
-            # content can be a list of blocks
             texts = [b.get("text", "") for b in raw if b.get("type") == "text"]
-            entry["content"] = "\n".join(texts) if texts else str(raw)
+            text = "\n".join(texts) if texts else str(raw)
         else:
-            entry["content"] = str(raw)
+            text = str(raw)
+        entry = {
+            "type": "tool_result",
+            "tool_use_id": block.get("tool_use_id", ""),
+            "is_error": block.get("is_error", False),
+            **_truncate_content(text),
+        }
         out.append(entry)
     return out
 

tests/test_api.sh

@@ -260,6 +260,27 @@ test_api_json_verbose() {
     _api_stop "${API_CONTAINER}-jv"
 }
 
+# ── large output (>64KB line) doesn't crash ─────────────────────────────────
+
+test_api_large_output() {
+    _api_start "${API_CONTAINER}-large" || return 1
+
+    # generate a file >64KB and ask claude to read it — tool result will be one big JSON line
+    local big_file="/tmp/bigfile_$$.txt"
+    python3 -c "print('X' * 70000)" > "$big_file"
+    curl -sf -X PUT "$API_BASE/files/bigtest.txt" --data-binary @"$big_file" >/dev/null
+    rm -f "$big_file"
+
+    local out code
+    code=$(curl -s -o /dev/null -w "%{http_code}" -X POST "$API_BASE/run" \
+        -H "Content-Type: application/json" \
+        -d "{\"prompt\": \"read the file /workspaces/bigtest.txt and tell me how many characters it has\", \"model\": \"$TEST_MODEL\", \"noContinue\": true}")
+    assert_eq "$code" "200" "large output does not crash (no 500)" || { _api_stop "${API_CONTAINER}-large"; return 1; }
+
+    echo "OK: api_large_output"
+    _api_stop "${API_CONTAINER}-large"
+}
+
 ALL_TESTS+=(
     test_api_endpoints
     test_api_run
@@ -272,4 +293,5 @@ ALL_TESTS+=(
     test_api_append_system_prompt
     test_api_continue_fallback
     test_api_json_verbose
+    test_api_large_output
 )