fix(acp,agent): support reasoning_content fallback and turn text publishing for local reasoning models - #3403
Open
mainza-ai wants to merge 16 commits into
Conversation
…ing, and wiki troubleshooting Signed-off-by: mainza-ai <mainza@gmail.com>
Signed-off-by: mainza-ai <mainza@gmail.com>
Signed-off-by: mainza-ai <mainza@gmail.com>
… AgentDefinition Signed-off-by: mainza-ai <mainza@gmail.com>
… AgentDefinition Signed-off-by: mainza-ai <mainza@gmail.com>
…team Signed-off-by: mainza-ai <mainza@gmail.com>
…team Signed-off-by: mainza-ai <mainza@gmail.com>
…urn text Signed-off-by: mainza-ai <mainza@gmail.com>
…ishing Signed-off-by: mainza-ai <mainza@gmail.com>
…do-code cleaning and duplicate card fixes Signed-off-by: mainza-ai <mainza@gmail.com>
…annel discovery Signed-off-by: mainza-ai <mainza@gmail.com>
…el discovery Signed-off-by: mainza-ai <mainza@gmail.com>
Author
Update: Multi-Tenant Loopback Host Normalization & Realtime NIP-29 Channel DiscoveryThis PR has been updated with surgical fixes resolving multi-tenant loopback isolation and realtime channel discovery for mentioned agents:
|
Signed-off-by: mainza-ai <mainza@gmail.com>
Signed-off-by: mainza-ai <mainza@gmail.com>
Signed-off-by: mainza-ai <mainza@gmail.com>
Signed-off-by: mainza-ai <mainza@gmail.com>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
This PR fixes core issues when running AI agents via
buzz-acpandbuzz-agentwith local OpenAI-compatible reasoning models (e.g. 4-bit Gemma models served on OMLX/Ollama endpoints):reasoning_contentorreasoningwhile returningcontent: ""(withouttool_calls). Updatesparse_openaiandparse_responsesinbuzz-agent/src/llm.rssotextfalls back toreasoningwhencontentis empty.buzz messages send.buzz-acpnow accumulatesagent_message_chunkupdates during a turn and automatically posts the cleaned turn text as a signedkind: 9message event to the channel thread onStopReason::EndTurn.clean_agent_text_responseinpool.rsto strip pseudo-tool JSON blocks (json {"reply_to": ...}) and special model tokens (<|tool_call>,<|im_end|>) before channel publishing.Testing & Verification
http://127.0.0.1:8787/v1) using Gemma-4 reasoning models.cargo test -p buzz-acp -p buzz-agent).