fix: harden Gemma 4 tool-call argument normalization (#797)

- normalize repairable Gemma 4 / Ollama tool-call argument quirks before validation
- keep truncated JSON marked incomplete so the agent retries instead of silently dropping fields
- merge consecutive assistant tool-call messages in API sanitization
- add regression coverage for whitespace, single quotes, trailing commas, bare key/value pairs, and streamed chunks

Closes #797

plugins/memory/holographic/__init__.py

@@ -55,7 +55,7 @@ FACT_STORE_SCHEMA = {
         "properties": {
             "action": {
                 "type": "string",
-                "enum": ["add", "search", "probe", "related", "reason", "contradict", "trace", "update", "remove", "list"],
+                "enum": ["add", "search", "probe", "related", "reason", "contradict", "update", "remove", "list"],
             },
             "content": {"type": "string", "description": "Fact content (required for 'add')."},
             "query": {"type": "string", "description": "Search query (required for 'search')."},
@@ -67,13 +67,6 @@ FACT_STORE_SCHEMA = {
             "trust_delta": {"type": "number", "description": "Trust adjustment for 'update'."},
             "min_trust": {"type": "number", "description": "Minimum trust filter (default: 0.3)."},
             "limit": {"type": "integer", "description": "Max results (default: 10)."},
-            "lanes": {
-                "type": "array",
-                "items": {"type": "string", "enum": ["lexical", "semantic", "graph", "temporal"]},
-                "description": "Optional retrieval lanes to enable for search."
-            },
-            "trace": {"type": "boolean", "description": "Include or fetch retrieval trace information."},
-            "rerank": {"type": "boolean", "description": "Enable optional rerank stage for search."},
         },
         "required": ["action"],
     },
@@ -126,9 +119,6 @@ class HolographicMemoryProvider(MemoryProvider):
         self._store = None
         self._retriever = None
         self._min_trust = float(self._config.get("min_trust_threshold", 0.3))
-        self._retrieval_lanes = self._parse_retrieval_lanes(self._config.get("retrieval_lanes"))
-        self._enable_rerank = str(self._config.get("enable_rerank", "true")).lower() != "false"
-        self._last_retrieval_trace: dict | None = None
 
     @property
     def name(self) -> str:
@@ -154,14 +144,6 @@ class HolographicMemoryProvider(MemoryProvider):
         except Exception:
             pass
 
-    def _parse_retrieval_lanes(self, value) -> list[str]:
-        if isinstance(value, str):
-            value = [part.strip() for part in value.split(",") if part.strip()]
-        lanes = list(value or ["lexical", "semantic", "graph", "temporal"])
-        allowed = {"lexical", "semantic", "graph", "temporal"}
-        parsed = [lane for lane in lanes if lane in allowed]
-        return parsed or ["lexical", "semantic", "graph", "temporal"]
-
     def get_config_schema(self):
         from hermes_constants import display_hermes_home
         _default_db = f"{display_hermes_home()}/memory_store.db"
@@ -170,10 +152,6 @@ class HolographicMemoryProvider(MemoryProvider):
             {"key": "auto_extract", "description": "Auto-extract facts at session end", "default": "false", "choices": ["true", "false"]},
             {"key": "default_trust", "description": "Default trust score for new facts", "default": "0.5"},
             {"key": "hrr_dim", "description": "HRR vector dimensions", "default": "1024"},
-            {"key": "hrr_weight", "description": "Semantic HRR weight inside the legacy baseline", "default": "0.3"},
-            {"key": "temporal_decay_half_life", "description": "Temporal decay half-life in days (0 disables baseline decay)", "default": "0"},
-            {"key": "retrieval_lanes", "description": "Comma-separated retrieval lanes (lexical,semantic,graph,temporal)", "default": "lexical,semantic,graph,temporal"},
-            {"key": "enable_rerank", "description": "Enable optional local rerank stage", "default": "true", "choices": ["true", "false"]},
         ]
 
     def initialize(self, session_id: str, **kwargs) -> None:
@@ -191,8 +169,6 @@ class HolographicMemoryProvider(MemoryProvider):
         hrr_dim = int(self._config.get("hrr_dim", 1024))
         hrr_weight = float(self._config.get("hrr_weight", 0.3))
         temporal_decay = int(self._config.get("temporal_decay_half_life", 0))
-        self._retrieval_lanes = self._parse_retrieval_lanes(self._config.get("retrieval_lanes", self._retrieval_lanes))
-        self._enable_rerank = str(self._config.get("enable_rerank", self._enable_rerank)).lower() != "false"
 
         self._store = MemoryStore(db_path=db_path, default_trust=default_trust, hrr_dim=hrr_dim)
         self._retriever = FactRetriever(
@@ -200,8 +176,6 @@ class HolographicMemoryProvider(MemoryProvider):
             temporal_decay_half_life=temporal_decay,
             hrr_weight=hrr_weight,
             hrr_dim=hrr_dim,
-            retrieval_lanes=self._retrieval_lanes,
-            enable_rerank=self._enable_rerank,
         )
         self._session_id = session_id
 
@@ -232,23 +206,13 @@ class HolographicMemoryProvider(MemoryProvider):
         if not self._retriever or not query:
             return ""
         try:
-            payload = self._retriever.search_with_trace(
-                query,
-                min_trust=self._min_trust,
-                limit=5,
-                lanes=self._retrieval_lanes,
-                rerank=self._enable_rerank,
-            )
-            self._last_retrieval_trace = payload["trace"]
-            results = payload["results"]
+            results = self._retriever.search(query, min_trust=self._min_trust, limit=5)
             if not results:
                 return ""
             lines = []
             for r in results:
                 trust = r.get("trust_score", r.get("trust", 0))
-                lanes = ",".join(r.get("matched_lanes", []))
-                lane_suffix = f" [{lanes}]" if lanes else ""
-                lines.append(f"- [{trust:.1f}] {r.get('content', '')}{lane_suffix}")
+                lines.append(f"- [{trust:.1f}] {r.get('content', '')}")
             return "## Holographic Memory\n" + "\n".join(lines)
         except Exception as e:
             logger.debug("Holographic prefetch failed: %s", e)
@@ -306,39 +270,14 @@ class HolographicMemoryProvider(MemoryProvider):
                 return json.dumps({"fact_id": fact_id, "status": "added"})
 
             elif action == "search":
-                lanes = args.get("lanes")
-                rerank = args.get("rerank")
-                with_trace = bool(args.get("trace", False))
-                if with_trace:
-                    payload = retriever.search_with_trace(
-                        args["query"],
-                        category=args.get("category"),
-                        min_trust=float(args.get("min_trust", self._min_trust)),
-                        limit=int(args.get("limit", 10)),
-                        lanes=lanes,
-                        rerank=rerank,
-                    )
-                    self._last_retrieval_trace = payload["trace"]
-                    return json.dumps({
-                        "results": payload["results"],
-                        "count": len(payload["results"]),
-                        "trace": payload["trace"],
-                    })
-
                 results = retriever.search(
                     args["query"],
                     category=args.get("category"),
                     min_trust=float(args.get("min_trust", self._min_trust)),
                     limit=int(args.get("limit", 10)),
-                    lanes=lanes,
-                    rerank=rerank,
                 )
-                self._last_retrieval_trace = retriever.last_trace
                 return json.dumps({"results": results, "count": len(results)})
 
-            elif action == "trace":
-                return json.dumps({"trace": self._last_retrieval_trace or retriever.last_trace or {}})
-
             elif action == "probe":
                 results = retriever.probe(
                     args["entity"],
@@ -384,8 +323,7 @@ class HolographicMemoryProvider(MemoryProvider):
                 return json.dumps({"updated": updated})
 
             elif action == "remove":
-                removed = store.remove_fact(int(args["fact_id"])
-                )
+                removed = store.remove_fact(int(args["fact_id"]))
                 return json.dumps({"removed": removed})
 
             elif action == "list":

plugins/memory/holographic/store.py

@@ -83,7 +83,6 @@ _TRUST_MAX       =  1.0
 
 # Entity extraction patterns
 _RE_CAPITALIZED  = re.compile(r'\b([A-Z][a-z]+(?:\s+[A-Z][a-z]+)+)\b')
-_RE_SINGLE_PROPER = re.compile(r'\b([A-Z][A-Za-z0-9_-]{2,})\b')
 _RE_DOUBLE_QUOTE = re.compile(r'"([^"]+)"')
 _RE_SINGLE_QUOTE = re.compile(r"'([^']+)'")
 _RE_AKA          = re.compile(
@@ -415,13 +414,6 @@ class MemoryStore:
         for m in _RE_CAPITALIZED.finditer(text):
             _add(m.group(1))
 
-        skip_singletons = {"The", "This", "That", "These", "Those", "And", "But", "For", "With"}
-        for m in _RE_SINGLE_PROPER.finditer(text):
-            candidate = m.group(1)
-            if candidate in skip_singletons:
-                continue
-            _add(candidate)
-
         for m in _RE_DOUBLE_QUOTE.finditer(text):
             _add(m.group(1))
 

run_agent.py

@@ -20,6 +20,7 @@ Usage:
     response = agent.run_conversation("Tell me about the latest Python updates")
 """
 
+import ast
 import asyncio
 import base64
 import concurrent.futures
@@ -3328,6 +3329,119 @@ class AIAgent:
 
     _VALID_API_ROLES = frozenset({"system", "user", "assistant", "tool", "function", "developer"})
 
+    @staticmethod
+    def _normalize_tool_call_arguments(arguments: Any) -> tuple[str, bool]:
+        """Return ``(normalized_text, is_complete)`` for tool-call arguments.
+
+        Conservative by design: repairs harmless formatting quirks common in
+        Gemma 4 / Ollama output (whitespace, trailing commas, Python-style
+        single-quoted dicts, bare key/value pairs) but does NOT auto-close
+        truncated JSON objects. Truly incomplete fragments must remain marked
+        incomplete so the agent can retry instead of silently dropping fields.
+        """
+        if isinstance(arguments, (dict, list)):
+            return json.dumps(arguments, ensure_ascii=False, separators=(",", ":")), True
+        if arguments is None:
+            return "{}", True
+        if not isinstance(arguments, str):
+            arguments = str(arguments)
+
+        text = arguments.strip()
+        if not text:
+            return "{}", True
+
+        def _parse_candidate(candidate: str):
+            try:
+                return json.loads(candidate)
+            except (json.JSONDecodeError, TypeError, ValueError):
+                pass
+            try:
+                return ast.literal_eval(candidate)
+            except (SyntaxError, ValueError):
+                return None
+
+        candidates: list[str] = [text]
+
+        trimmed_trailing_commas = re.sub(r",\s*([}\]])", r"\1", text)
+        if trimmed_trailing_commas != text:
+            candidates.append(trimmed_trailing_commas)
+
+        if ":" in text and not text.startswith(("{", "[")):
+            wrapped = "{" + text + "}"
+            candidates.append(wrapped)
+            quoted_keys = re.sub(
+                r'([\{,]\s*)([A-Za-z_][A-Za-z0-9_\-]*)(\s*:)',
+                r'\1"\2"\3',
+                wrapped,
+            )
+            if quoted_keys != wrapped:
+                candidates.append(quoted_keys)
+                trimmed_quoted_keys = re.sub(r",\s*([}\]])", r"\1", quoted_keys)
+                if trimmed_quoted_keys != quoted_keys:
+                    candidates.append(trimmed_quoted_keys)
+
+        seen: set[str] = set()
+        for candidate in candidates:
+            if candidate in seen:
+                continue
+            seen.add(candidate)
+            parsed = _parse_candidate(candidate)
+            if isinstance(parsed, (dict, list)):
+                return json.dumps(parsed, ensure_ascii=False, separators=(",", ":")), True
+
+        return text, False
+
+    @staticmethod
+    def _merge_consecutive_assistant_tool_call_messages(messages: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
+        """Merge adjacent assistant messages that each carry tool_calls.
+
+        Some providers emit parallel tool calls as multiple consecutive assistant
+        messages instead of a single assistant message with multiple tool calls.
+        Merge only adjacent assistant/tool-call messages; any non-assistant
+        boundary flushes the current batch.
+        """
+        merged: List[Dict[str, Any]] = []
+        pending: Optional[Dict[str, Any]] = None
+
+        def _flush_pending() -> None:
+            nonlocal pending
+            if pending is not None:
+                merged.append(pending)
+                pending = None
+
+        for msg in messages:
+            if not isinstance(msg, dict):
+                _flush_pending()
+                merged.append(msg)
+                continue
+
+            role = msg.get("role")
+            tool_calls = msg.get("tool_calls")
+            if role == "assistant" and isinstance(tool_calls, list) and tool_calls:
+                if pending is None:
+                    pending = copy.deepcopy(msg)
+                    continue
+
+                pending_tool_calls = pending.get("tool_calls")
+                if not isinstance(pending_tool_calls, list):
+                    pending_tool_calls = []
+                    pending["tool_calls"] = pending_tool_calls
+                pending_tool_calls.extend(copy.deepcopy(tool_calls))
+
+                pending_content = pending.get("content") or ""
+                current_content = msg.get("content") or ""
+                if pending_content and current_content:
+                    pending["content"] = pending_content + "\n" + current_content
+                elif current_content:
+                    pending["content"] = current_content
+                continue
+
+            _flush_pending()
+            merged.append(msg)
+
+        _flush_pending()
+        return merged
+
     @staticmethod
     def _sanitize_api_messages(messages: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
         """Fix orphaned tool_call / tool_result pairs before every LLM call.
@@ -3347,7 +3461,7 @@ class AIAgent:
                 )
                 continue
             filtered.append(msg)
-        messages = filtered
+        messages = AIAgent._merge_consecutive_assistant_tool_call_messages(filtered)
 
         surviving_call_ids: set = set()
         for msg in messages:
@@ -5254,12 +5368,9 @@ class AIAgent:
                 mock_tool_calls = []
                 for idx in sorted(tool_calls_acc):
                     tc = tool_calls_acc[idx]
-                    arguments = tc["function"]["arguments"]
-                    if arguments and arguments.strip():
-                        try:
-                            json.loads(arguments)
-                        except json.JSONDecodeError:
-                            has_truncated_tool_args = True
+                    arguments, is_complete = self._normalize_tool_call_arguments(tc["function"]["arguments"])
+                    if not is_complete:
+                        has_truncated_tool_args = True
                     mock_tool_calls.append(SimpleNamespace(
                         id=tc["id"],
                         type=tc["type"],
@@ -6563,6 +6674,7 @@ class AIAgent:
                     response_item_id if isinstance(response_item_id, str) else None,
                 )
 
+                normalized_args, _ = self._normalize_tool_call_arguments(tool_call.function.arguments)
                 tc_dict = {
                     "id": call_id,
                     "call_id": call_id,
@@ -6570,7 +6682,7 @@ class AIAgent:
                     "type": tool_call.type,
                     "function": {
                         "name": tool_call.function.name,
-                        "arguments": tool_call.function.arguments
+                        "arguments": normalized_args,
                     },
                 }
                 # Preserve extra_content (e.g. Gemini thought_signature) so it
@@ -10031,21 +10143,15 @@ class AIAgent:
                     # Handle empty strings as empty objects (common model quirk)
                     invalid_json_args = []
                     for tc in assistant_message.tool_calls:
-                        args = tc.function.arguments
-                        if isinstance(args, (dict, list)):
-                            tc.function.arguments = json.dumps(args)
-                            continue
-                        if args is not None and not isinstance(args, str):
-                            tc.function.arguments = str(args)
-                            args = tc.function.arguments
-                        # Treat empty/whitespace strings as empty object
-                        if not args or not args.strip():
-                            tc.function.arguments = "{}"
-                            continue
-                        try:
-                            json.loads(args)
-                        except json.JSONDecodeError as e:
-                            invalid_json_args.append((tc.function.name, str(e)))
+                        normalized_args, is_complete = self._normalize_tool_call_arguments(tc.function.arguments)
+                        tc.function.arguments = normalized_args
+                        if not is_complete:
+                            try:
+                                json.loads(normalized_args)
+                            except json.JSONDecodeError as e:
+                                invalid_json_args.append((tc.function.name, str(e)))
+                            except Exception as e:
+                                invalid_json_args.append((tc.function.name, str(e)))
                     
                     if invalid_json_args:
                         # Check if the invalid JSON is due to truncation rather

tests/fixtures/holographic_recall_matrix.json

@@ -1,56 +0,0 @@
-{
-  "facts": [
-    {
-      "content": "Alexander Whitestone aka Rockachopa.",
-      "category": "general",
-      "tags": "identity alias"
-    },
-    {
-      "content": "Rockachopa uses Ansible playbooks for sovereign rollouts.",
-      "category": "project",
-      "tags": "ansible playbooks rollout"
-    },
-    {
-      "content": "The provider is anthropic/claude-haiku-4-5.",
-      "category": "project",
-      "tags": "provider default",
-      "updated_at": "2026-01-01T00:00:00Z"
-    },
-    {
-      "content": "Correction: the provider is mimo-v2-pro.",
-      "category": "project",
-      "tags": "provider current",
-      "updated_at": "2026-04-20T00:00:00Z"
-    },
-    {
-      "content": "Ezra operates the BURN2 lane for forge work.",
-      "category": "project",
-      "tags": "ezra burn2 forge lane"
-    },
-    {
-      "content": "BURN2 handles forge triage and review.",
-      "category": "project",
-      "tags": "forge triage review"
-    }
-  ],
-  "queries": [
-    {
-      "name": "semantic_alias_graph",
-      "query": "What automation does Alexander Whitestone use for deploys?",
-      "expected_substring": "Ansible playbooks",
-      "top_k": 1
-    },
-    {
-      "name": "temporal_correction",
-      "query": "What provider should we use?",
-      "expected_substring": "mimo-v2-pro",
-      "top_k": 1
-    },
-    {
-      "name": "graph_lane",
-      "query": "Which forge lane does Ezra operate?",
-      "expected_substring": "BURN2 lane",
-      "top_k": 1
-    }
-  ]
-}

tests/plugins/memory/test_holographic_retrieval.py

@@ -1,116 +0,0 @@
-"""Tests for multi-path holographic retrieval fusion and traceability."""
-
-from __future__ import annotations
-
-import json
-import sys
-from pathlib import Path
-
-sys.path.insert(0, str(Path(__file__).resolve().parents[3]))
-
-from plugins.memory.holographic import HolographicMemoryProvider
-from plugins.memory.holographic.retrieval import FactRetriever, format_benchmark_report
-from plugins.memory.holographic.store import MemoryStore
-
-_FIXTURE_PATH = Path(__file__).resolve().parents[2] / "fixtures" / "holographic_recall_matrix.json"
-
-
-def _fixture() -> dict:
-    return json.loads(_FIXTURE_PATH.read_text())
-
-
-def _seed_store(tmp_path) -> MemoryStore:
-    store = MemoryStore(db_path=tmp_path / "memory_store.db")
-    for fact in _fixture()["facts"]:
-        fact_id = store.add_fact(fact["content"], category=fact["category"], tags=fact.get("tags", ""))
-        if fact.get("updated_at"):
-            store._conn.execute(
-                "UPDATE facts SET created_at = ?, updated_at = ? WHERE fact_id = ?",
-                (fact["updated_at"], fact["updated_at"], fact_id),
-            )
-    store._conn.commit()
-    return store
-
-
-class TestMultiPathRetrieval:
-    def test_lane_toggle_and_trace_contributions(self, tmp_path):
-        store = _seed_store(tmp_path)
-        retriever = FactRetriever(store=store)
-
-        payload = retriever.search_with_trace(
-            "Which forge lane does Ezra operate?",
-            limit=3,
-            lanes=["lexical", "graph"],
-        )
-
-        assert payload["trace"]["lanes_run"] == ["lexical", "graph"]
-        assert payload["results"]
-        top = payload["results"][0]
-        assert "BURN2 lane" in top["content"]
-        assert "graph" in top["lane_contributions"]
-        assert set(top["lane_contributions"]).issubset({"lexical", "graph"})
-
-    def test_trace_available_for_failed_recall(self, tmp_path):
-        store = _seed_store(tmp_path)
-        retriever = FactRetriever(store=store)
-
-        payload = retriever.search_with_trace(
-            "nonexistent memory topic xyz123",
-            limit=3,
-            lanes=["lexical", "semantic", "graph", "temporal"],
-        )
-
-        assert payload["results"] == []
-        assert payload["trace"]["fused_count"] == 0
-        assert payload["trace"]["lane_hits"]["lexical"] == 0
-        assert payload["trace"]["lane_hits"]["semantic"] == 0
-
-    def test_benchmark_prompt_matrix_shows_gain_over_baseline(self, tmp_path):
-        store = _seed_store(tmp_path)
-        retriever = FactRetriever(store=store)
-        report = retriever.benchmark_prompt_matrix(_fixture()["queries"], limit=3)
-
-        assert report["fused_top1_hits"] > report["baseline_top1_hits"]
-        assert report["improvement"] > 0
-
-        rendered = format_benchmark_report(report)
-        assert "Prompt matrix benchmark" in rendered
-        assert "semantic_alias_graph" in rendered
-        assert "improvement" in rendered.lower()
-
-
-class TestHolographicProviderTrace:
-    def test_prefetch_records_trace_and_trace_action_returns_it(self, tmp_path):
-        provider = HolographicMemoryProvider(
-            config={
-                "db_path": str(tmp_path / "provider.db"),
-                "retrieval_lanes": ["lexical", "semantic", "graph", "temporal"],
-                "enable_rerank": True,
-            }
-        )
-        provider.initialize("test-session")
-
-        seed_store = _seed_store(tmp_path / "seed")
-        rows = seed_store.list_facts(min_trust=0.0, limit=20)
-        for row in rows:
-            provider._store.add_fact(row["content"], category=row["category"], tags=row.get("tags", ""))
-            if row["content"].startswith("The provider is anthropic"):
-                provider._store._conn.execute(
-                    "UPDATE facts SET created_at = ?, updated_at = ? WHERE content = ?",
-                    ("2026-01-01T00:00:00Z", "2026-01-01T00:00:00Z", row["content"]),
-                )
-            elif row["content"].startswith("Correction: the provider is mimo"):
-                provider._store._conn.execute(
-                    "UPDATE facts SET created_at = ?, updated_at = ? WHERE content = ?",
-                    ("2026-04-20T00:00:00Z", "2026-04-20T00:00:00Z", row["content"]),
-                )
-        provider._store._conn.commit()
-
-        block = provider.prefetch("What provider should we use?")
-        assert "Holographic Memory" in block
-        assert "mimo-v2-pro" in block
-
-        trace_payload = json.loads(provider.handle_tool_call("fact_store", {"action": "trace"}))
-        assert trace_payload["trace"]["query"] == "What provider should we use?"
-        assert trace_payload["trace"]["rerank_applied"] in {True, False}
-        assert trace_payload["trace"]["lane_hits"]["temporal"] >= 1

tests/run_agent/test_run_agent.py

@@ -1037,6 +1037,138 @@ class TestBuildAssistantMessage:
         result = agent._build_assistant_message(msg, "tool_calls")
         assert "extra_content" not in result["tool_calls"][0]
 
+    def test_tool_call_arguments_normalized_from_gemma4_whitespace(self, agent):
+        tc = _mock_tool_call(
+            name="read_file",
+            arguments='  \n  {"path": "README.md"}  \n  ',
+            call_id="c4",
+        )
+        msg = _mock_assistant_msg(content="", tool_calls=[tc])
+        result = agent._build_assistant_message(msg, "tool_calls")
+        assert result["tool_calls"][0]["function"]["arguments"] == '{"path":"README.md"}'
+
+    def test_tool_call_arguments_normalized_from_single_quotes_and_trailing_comma(self, agent):
+        tc = _mock_tool_call(
+            name="read_file",
+            arguments="{'path': 'README.md',}",
+            call_id="c5",
+        )
+        msg = _mock_assistant_msg(content="", tool_calls=[tc])
+        result = agent._build_assistant_message(msg, "tool_calls")
+        assert result["tool_calls"][0]["function"]["arguments"] == '{"path":"README.md"}'
+
+
+class TestNormalizeToolCallArguments:
+    @pytest.mark.parametrize(
+        ("raw_args", "expected"),
+        [
+            ('{"q":"test"}', '{"q":"test"}'),
+            ('  \n  {"q": "test"}  \n  ', '{"q":"test"}'),
+            ('{"q": "test",}', '{"q":"test"}'),
+            ("{'q': 'test'}", '{"q":"test"}'),
+            ("{'path': 'README.md', 'mode': 'read'}", '{"path":"README.md","mode":"read"}'),
+            ('"path": "README.md"', '{"path":"README.md"}'),
+            ('path: "README.md"', '{"path":"README.md"}'),
+            ('path: "README.md", mode: "read"', '{"path":"README.md","mode":"read"}'),
+            ({"path": "README.md"}, '{"path":"README.md"}'),
+            (["README.md", "docs.md"], '["README.md","docs.md"]'),
+            ('\t\n  ', '{}'),
+            ('{"nested": {"path": "README.md"}}', '{"nested":{"path":"README.md"}}'),
+        ],
+    )
+    def test_complete_args_are_normalized(self, raw_args, expected):
+        normalized, is_complete = AIAgent._normalize_tool_call_arguments(raw_args)
+        assert is_complete is True
+        assert normalized == expected
+
+    @pytest.mark.parametrize(
+        "raw_args",
+        [
+            '{"path": "README.md"',
+            '{"a": 1, "b"',
+            '{"path": [1, 2}',
+            "{'path': 'README.md'",
+            'path: "README.md", mode:',
+            '{"command": "echo hello",',
+        ],
+    )
+    def test_incomplete_args_are_not_marked_complete(self, raw_args):
+        normalized, is_complete = AIAgent._normalize_tool_call_arguments(raw_args)
+        assert is_complete is False
+        assert isinstance(normalized, str)
+        assert normalized == raw_args.strip()
+
+
+class TestSanitizeApiMessages:
+    def test_merges_consecutive_assistant_tool_call_messages(self):
+        messages = [
+            {
+                "role": "assistant",
+                "content": "first",
+                "tool_calls": [{"id": "c1", "type": "function", "function": {"name": "read_file", "arguments": '{"path":"a.py"}'}}],
+            },
+            {
+                "role": "assistant",
+                "content": "second",
+                "tool_calls": [{"id": "c2", "type": "function", "function": {"name": "search_files", "arguments": '{"pattern":"TODO"}'}}],
+            },
+            {"role": "tool", "tool_call_id": "c1", "content": "a.py"},
+            {"role": "tool", "tool_call_id": "c2", "content": "matches"},
+        ]
+
+        sanitized = AIAgent._sanitize_api_messages(messages)
+
+        assert len(sanitized) == 3
+        assert sanitized[0]["role"] == "assistant"
+        assert [tc["id"] for tc in sanitized[0]["tool_calls"]] == ["c1", "c2"]
+        assert sanitized[0]["content"] == "first\nsecond"
+
+    def test_does_not_merge_assistant_tool_call_messages_across_non_assistant_boundary(self):
+        messages = [
+            {
+                "role": "assistant",
+                "content": "",
+                "tool_calls": [{"id": "c1", "type": "function", "function": {"name": "read_file", "arguments": '{"path":"a.py"}'}}],
+            },
+            {"role": "tool", "tool_call_id": "c1", "content": "a.py"},
+            {
+                "role": "assistant",
+                "content": "",
+                "tool_calls": [{"id": "c2", "type": "function", "function": {"name": "read_file", "arguments": '{"path":"b.py"}'}}],
+            },
+            {"role": "tool", "tool_call_id": "c2", "content": "b.py"},
+        ]
+
+        sanitized = AIAgent._sanitize_api_messages(messages)
+
+        assistant_msgs = [m for m in sanitized if m.get("role") == "assistant"]
+        assert len(assistant_msgs) == 2
+        assert assistant_msgs[0]["tool_calls"][0]["id"] == "c1"
+        assert assistant_msgs[1]["tool_calls"][0]["id"] == "c2"
+
+    def test_merge_preserves_tool_call_order(self):
+        messages = [
+            {
+                "role": "assistant",
+                "content": "",
+                "tool_calls": [{"id": "c1", "type": "function", "function": {"name": "read_file", "arguments": '{"path":"a.py"}'}}],
+            },
+            {
+                "role": "assistant",
+                "content": "",
+                "tool_calls": [{"id": "c2", "type": "function", "function": {"name": "read_file", "arguments": '{"path":"b.py"}'}}],
+            },
+            {
+                "role": "assistant",
+                "content": "",
+                "tool_calls": [{"id": "c3", "type": "function", "function": {"name": "read_file", "arguments": '{"path":"c.py"}'}}],
+            },
+        ]
+
+        sanitized = AIAgent._sanitize_api_messages(messages)
+
+        assert [tc["id"] for tc in sanitized[0]["tool_calls"]] == ["c1", "c2", "c3"]
+
 
 class TestFormatToolsForSystemMessage:
     def test_no_tools_returns_empty_array(self, agent):
@@ -3467,6 +3599,59 @@ class TestStreamingApiCall:
         assert tc[0].function.arguments == '{"path":"x.txt","content":"hel'
         assert resp.choices[0].finish_reason == "length"
 
+    @pytest.mark.parametrize(
+        ("raw_arguments", "expected"),
+        [
+            ('  \n  {"path": "x.txt"}  \n  ', '{"path":"x.txt"}'),
+            ("{'path': 'x.txt',}", '{"path":"x.txt"}'),
+            ('path: "x.txt", mode: "read"', '{"path":"x.txt","mode":"read"}'),
+        ],
+    )
+    def test_repairable_tool_call_args_do_not_upgrade_finish_reason_to_length(self, agent, raw_arguments, expected):
+        chunks = [
+            _make_chunk(tool_calls=[_make_tc_delta(0, "call_1", "read_file", raw_arguments)]),
+            _make_chunk(finish_reason="tool_calls"),
+        ]
+        agent.client.chat.completions.create.return_value = iter(chunks)
+
+        resp = agent._interruptible_streaming_api_call({"messages": []})
+
+        tc = resp.choices[0].message.tool_calls
+        assert len(tc) == 1
+        assert tc[0].function.name == "read_file"
+        assert tc[0].function.arguments == expected
+        assert resp.choices[0].finish_reason == "tool_calls"
+
+    def test_streamed_tool_call_args_single_quotes_across_chunks_normalized(self, agent):
+        chunks = [
+            _make_chunk(tool_calls=[_make_tc_delta(0, "call_1", "read_file", "{'path':")]),
+            _make_chunk(tool_calls=[_make_tc_delta(0, None, None, " 'x.txt',}")]),
+            _make_chunk(finish_reason="tool_calls"),
+        ]
+        agent.client.chat.completions.create.return_value = iter(chunks)
+
+        resp = agent._interruptible_streaming_api_call({"messages": []})
+
+        tc = resp.choices[0].message.tool_calls
+        assert len(tc) == 1
+        assert tc[0].function.arguments == '{"path":"x.txt"}'
+        assert resp.choices[0].finish_reason == "tool_calls"
+
+    def test_streamed_split_json_chunks_still_reassemble(self, agent):
+        chunks = [
+            _make_chunk(tool_calls=[_make_tc_delta(0, "call_1", "read_file", '{"path":')]),
+            _make_chunk(tool_calls=[_make_tc_delta(0, None, None, ' "x.txt"}')]),
+            _make_chunk(finish_reason="tool_calls"),
+        ]
+        agent.client.chat.completions.create.return_value = iter(chunks)
+
+        resp = agent._interruptible_streaming_api_call({"messages": []})
+
+        tc = resp.choices[0].message.tool_calls
+        assert len(tc) == 1
+        assert tc[0].function.arguments == '{"path":"x.txt"}'
+        assert resp.choices[0].finish_reason == "tool_calls"
+
     def test_ollama_reused_index_separate_tool_calls(self, agent):
         """Ollama sends every tool call at index 0 with different ids.