Grayscale/nanobot
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor(runner): update message processing to preserve historical context

Browse Source 
- Adjusted message handling in AgentRunner to ensure that historical messages remain unchanged during context governance.
- Introduced tests to verify that backfill operations do not alter the saved message boundary, maintaining the integrity of the conversation history.
This commit is contained in:
Xubin Ren
2026-04-10 04:46:48 +00:00
parent 27e7a338a3
commit 363a0704db
2 changed files with 171 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
nanobot/agent/runner.py
+8 -4
View File
@@ -101,10 +101,14 @@ class AgentRunner:
for iteration in range(spec.max_iterations):
try:
messages = self._backfill_missing_tool_results(messages)
messages = self._microcompact(messages)
messages = self._apply_tool_result_budget(spec, messages)
messages_for_model = self._snip_history(spec, messages)
# Keep the persisted conversation untouched. Context governance
# 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._backfill_missing_tool_results(messages)
messages_for_model = self._microcompact(messages_for_model)
messages_for_model = self._apply_tool_result_budget(spec, messages_for_model)
messages_for_model = self._snip_history(spec, messages_for_model)
except Exception as exc:
logger.warning(
"Context governance failed on turn {} for {}: {}; using raw messages",