fix(gateway): break compression-exhaustion infinite loop and auto-reset session - #10063
Merged
Merged
Conversation
…et session (#9893) When compression fails after max attempts, the agent returns {completed: False, partial: True} but was missing the 'failed' flag. The gateway's agent_failed_early guard checked for 'failed' AND 'not final_response', but _run_agent_blocking always converts errors to final_response — making the guard dead code. This caused the oversized session to persist, creating an infinite fail loop where every subsequent message hits the same compression failure. Changes: - run_agent.py: add 'failed: True' and 'compression_exhausted: True' to all 5 compression-exhaustion return paths - gateway/run.py (_run_agent_blocking): forward 'failed' and 'compression_exhausted' flags through to the caller - gateway/run.py (_handle_message_with_agent): fix agent_failed_early to check bool(failed) without the broken 'not final_response' clause; auto-reset the session when compression is exhausted so the next message starts fresh - Update tests to match new guard logic and add TestCompressionExhaustedFlag test class Closes #9893
18 tasks
teknium1
pushed a commit
that referenced
this pull request
Jun 13, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the #9893/#10063 auto-reset fix. Fixes #35809
AIalliAI
pushed a commit
to AIalliAI/Hermes
that referenced
this pull request
Jun 14, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
T02200059
pushed a commit
to T02200059/hermes-agent
that referenced
this pull request
Jun 18, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
waefrebeorn
pushed a commit
to waefrebeorn/slermes
that referenced
this pull request
Jul 2, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
habarmc1223-sudo
pushed a commit
to habarmc1223-sudo/hermes-agent-fluxmem
that referenced
this pull request
Jul 8, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
santhreal
pushed a commit
to santhreal/hermes-agent
that referenced
this pull request
Jul 13, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
donbowman
pushed a commit
to donbowman/hermes-agent
that referenced
this pull request
Jul 13, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
Gravezzz
pushed a commit
to Gravezzz/hermes-agent
that referenced
this pull request
Jul 21, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
1 task
leewenjie
pushed a commit
to leewenjie/hermes-agent
that referenced
this pull request
Aug 7, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
melon-xf
added a commit
to melon-xf/hermes-agent
that referenced
this pull request
Sep 3, 2026
…tion auto-reset After compression exhaustion the auto-reset created a fresh session but discarded reset_session()'s return value and left the Telegram topic binding pointing at the oversized compressed child. The next inbound message in that topic healed the binding forward and switch_session'd the freshly-reset lane back onto the bloated transcript, re-triggering compression exhaustion in a loop with a new session id each time. Capture the fresh entry and re-sync the topic binding to it so the next message starts clean. No-op on non-topic lanes. Regression of the NousResearch#9893/NousResearch#10063 auto-reset fix. Fixes NousResearch#35809
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Fixes #9893 — gateway becomes unresponsive when context compression is exhausted in long-running group chat sessions.
Root Cause (two-layer bug)
Layer 1: When compression fails after max attempts,
run_conversation()returns{completed: False, partial: True}but was missing the"failed"flag. The gateway'sagent_failed_earlyguard only checks for"failed".Layer 2: Even if
"failed"were set, the guard was dead code._run_agent_blockingalways converts error results to afinal_response(line 8622:"⚠️ {error}"), so the guard'sand not agent_result.get("final_response")clause was always False.Result: The oversized session gets persisted to the transcript after every failed compression attempt. The next message loads the same oversized history, hits the same compression failure, and loops. The user sees repeated "compacting context..." messages until the gateway dies.
Changes
run_agent.py
"failed": Trueand"compression_exhausted": Trueto all 5 compression-exhaustion return paths (413 payload too large, context length exceeded, cannot compress further)gateway/run.py
_run_agent_blocking: Forward"failed"and"compression_exhausted"flags from the agent result to the caller_handle_message_with_agent: Fixagent_failed_earlyto usebool(agent_result.get("failed"))— removing the brokenand not final_responseclause that made it dead codecompression_exhaustedis True, automatically reset the session (reset_session+ cache eviction + model override clear) so the next message starts fresh. Appends a user-visible notice explaining the auto-reset.Tests
TestGatewaySkipsPersistenceOnFailureto match the new guard logicTestCompressionExhaustedFlagclass with 2 testsTest Results
tests/run_agent/test_1630_context_overflow_loop.py— 16/16 passedtests/run_agent/test_compression_feasibility.py— 12/12 passedtests/gateway/— 2879 passed (10 failures + 10 errors are pre-existing, unrelated)