fix(providers): dedupe tool_use ids to prevent Anthropic 400s
Anthropic rejects any request where two tool_use blocks share an id
("messages.N.content.M: tool_use ids must be unique"). A mis-assembled
stream could surface the same tool_use block twice in one assistant turn;
the runner persisted it verbatim, so the malformed message was re-sent on
every subsequent turn and permanently bricked the session — the agent
silently stopped replying.
Fix at two layers:
- AnthropicProvider._parse_response: drop duplicate tool_use ids (keep
first) as the response enters nanobot, so corruption is never persisted.
- AgentRunner._dedup_tool_calls: a new context-governance pass that dedupes
assistant tool_calls and tool results by id before each send, healing any
history that was already corrupted.
Add regression tests covering both the dedup and the no-op fast path.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
committed by
Xubin Ren
co-authored by
Claude Opus 4.8
parent
160cec2396
commit
6689e2d377
+56
-2
@@ -373,7 +373,8 @@ class AgentRunner:
|
||||
# may repair or compact historical messages for the model, but
|
||||
# those synthetic edits must not shift the append boundary used
|
||||
# later when the caller saves only the new turn.
|
||||
messages_for_model = self._drop_orphan_tool_results(messages)
|
||||
messages_for_model = self._dedup_tool_calls(messages)
|
||||
messages_for_model = self._drop_orphan_tool_results(messages_for_model)
|
||||
messages_for_model = self._backfill_missing_tool_results(messages_for_model)
|
||||
messages_for_model = self._microcompact(messages_for_model)
|
||||
messages_for_model = self._apply_tool_result_budget(spec, messages_for_model)
|
||||
@@ -388,7 +389,8 @@ class AgentRunner:
|
||||
spec.session_key or "default",
|
||||
)
|
||||
try:
|
||||
messages_for_model = self._drop_orphan_tool_results(messages)
|
||||
messages_for_model = self._dedup_tool_calls(messages)
|
||||
messages_for_model = self._drop_orphan_tool_results(messages_for_model)
|
||||
messages_for_model = self._backfill_missing_tool_results(messages_for_model)
|
||||
except Exception:
|
||||
messages_for_model = messages
|
||||
@@ -1355,6 +1357,58 @@ class AgentRunner:
|
||||
return truncate_text(content, spec.max_tool_result_chars)
|
||||
return content
|
||||
|
||||
@staticmethod
|
||||
def _dedup_tool_calls(
|
||||
messages: list[dict[str, Any]],
|
||||
) -> list[dict[str, Any]]:
|
||||
"""Remove duplicate tool_call / tool_result ids from the history.
|
||||
|
||||
Anthropic rejects any request where two tool_use blocks share an id
|
||||
("tool_use ids must be unique"). An accidental duplicate (e.g. from a
|
||||
mis-assembled stream) otherwise poisons the session permanently, since
|
||||
the bad message is re-sent on every turn. Keep the first occurrence of
|
||||
each id across all assistant tool_calls, and likewise keep only the
|
||||
first tool result per id so pairing stays 1:1.
|
||||
"""
|
||||
seen_call_ids: set[str] = set()
|
||||
seen_result_ids: set[str] = set()
|
||||
updated: list[dict[str, Any]] | None = None
|
||||
for idx, msg in enumerate(messages):
|
||||
role = msg.get("role")
|
||||
replacement: dict[str, Any] | None = None
|
||||
drop = False
|
||||
|
||||
if role == "assistant" and msg.get("tool_calls"):
|
||||
kept = []
|
||||
changed = False
|
||||
for tc in msg.get("tool_calls") or []:
|
||||
tid = tc.get("id") if isinstance(tc, dict) else None
|
||||
if tid and tid in seen_call_ids:
|
||||
changed = True
|
||||
continue
|
||||
if tid:
|
||||
seen_call_ids.add(tid)
|
||||
kept.append(tc)
|
||||
if changed:
|
||||
replacement = dict(msg)
|
||||
replacement["tool_calls"] = kept
|
||||
elif role == "tool":
|
||||
tid = msg.get("tool_call_id")
|
||||
if tid:
|
||||
if str(tid) in seen_result_ids:
|
||||
drop = True
|
||||
else:
|
||||
seen_result_ids.add(str(tid))
|
||||
|
||||
if (replacement is not None or drop) and updated is None:
|
||||
updated = [dict(m) for m in messages[:idx]]
|
||||
if drop:
|
||||
continue
|
||||
if updated is not None:
|
||||
updated.append(replacement if replacement is not None else dict(msg))
|
||||
|
||||
return messages if updated is None else updated
|
||||
|
||||
@staticmethod
|
||||
def _drop_orphan_tool_results(
|
||||
messages: list[dict[str, Any]],
|
||||
|
||||
@@ -10,6 +10,8 @@ import string
|
||||
from collections.abc import Awaitable, Callable
|
||||
from typing import Any
|
||||
|
||||
from loguru import logger
|
||||
|
||||
from nanobot.providers.base import (
|
||||
LLMProvider,
|
||||
LLMResponse,
|
||||
@@ -522,11 +524,19 @@ class AnthropicProvider(LLMProvider):
|
||||
content_parts: list[str] = []
|
||||
tool_calls: list[ToolCallRequest] = []
|
||||
thinking_blocks: list[dict[str, Any]] = []
|
||||
seen_tool_ids: set[str] = set()
|
||||
|
||||
for block in response.content:
|
||||
if block.type == "text":
|
||||
content_parts.append(block.text)
|
||||
elif block.type == "tool_use":
|
||||
# Anthropic requires every tool_use id to be unique. A mis-assembled
|
||||
# stream can occasionally surface the same block twice; keep the first
|
||||
# and drop the duplicate so it never poisons the persisted history.
|
||||
if block.id in seen_tool_ids:
|
||||
logger.warning("dropping duplicate tool_use id from response: {}", block.id)
|
||||
continue
|
||||
seen_tool_ids.add(block.id)
|
||||
tool_calls.append(ToolCallRequest(
|
||||
id=block.id,
|
||||
name=block.name,
|
||||
|
||||
@@ -184,6 +184,63 @@ async def test_backfill_missing_tool_results_inserts_error():
|
||||
assert backfilled[0]["name"] == "read_file"
|
||||
|
||||
|
||||
def test_dedup_tool_calls_removes_duplicate_ids():
|
||||
from nanobot.agent.runner import AgentRunner
|
||||
|
||||
messages = [
|
||||
{"role": "user", "content": "hi"},
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "",
|
||||
"tool_calls": [
|
||||
{"id": "a", "type": "function", "function": {"name": "x", "arguments": "{}"}},
|
||||
{"id": "b", "type": "function", "function": {"name": "y", "arguments": "{}"}},
|
||||
# Duplicate of "b" — would trigger "tool_use ids must be unique".
|
||||
{"id": "b", "type": "function", "function": {"name": "y", "arguments": "{}"}},
|
||||
],
|
||||
},
|
||||
{"role": "tool", "tool_call_id": "a", "name": "x", "content": "ra"},
|
||||
{"role": "tool", "tool_call_id": "b", "name": "y", "content": "rb"},
|
||||
# Duplicate result for "b".
|
||||
{"role": "tool", "tool_call_id": "b", "name": "y", "content": "rb-dup"},
|
||||
]
|
||||
|
||||
cleaned = AgentRunner._dedup_tool_calls(messages)
|
||||
|
||||
assert cleaned == [
|
||||
{"role": "user", "content": "hi"},
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "",
|
||||
"tool_calls": [
|
||||
{"id": "a", "type": "function", "function": {"name": "x", "arguments": "{}"}},
|
||||
{"id": "b", "type": "function", "function": {"name": "y", "arguments": "{}"}},
|
||||
],
|
||||
},
|
||||
{"role": "tool", "tool_call_id": "a", "name": "x", "content": "ra"},
|
||||
{"role": "tool", "tool_call_id": "b", "name": "y", "content": "rb"},
|
||||
]
|
||||
|
||||
|
||||
def test_dedup_tool_calls_noop_when_unique():
|
||||
from nanobot.agent.runner import AgentRunner
|
||||
|
||||
messages = [
|
||||
{"role": "user", "content": "hi"},
|
||||
{
|
||||
"role": "assistant",
|
||||
"content": "",
|
||||
"tool_calls": [
|
||||
{"id": "a", "type": "function", "function": {"name": "x", "arguments": "{}"}},
|
||||
],
|
||||
},
|
||||
{"role": "tool", "tool_call_id": "a", "name": "x", "content": "ra"},
|
||||
]
|
||||
|
||||
# No duplicates -> identical list object returned (cheap no-op path).
|
||||
assert AgentRunner._dedup_tool_calls(messages) is messages
|
||||
|
||||
|
||||
def test_drop_orphan_tool_results_removes_unmatched_tool_messages():
|
||||
from nanobot.agent.runner import AgentRunner
|
||||
|
||||
|
||||
Reference in New Issue
Block a user