Repository navigation

fix(harness): keep tool-call markup out of summaries and tool-less answers - #277

Merged

senamakel merged 27 commits into

tinyhumansai:mainfrom

senamakel:deepseek-markup-leak

Oct 3, 2026

Commits on Oct 3, 2026

fix(harness): handle empty model output in summarization

senamakel
and
medullabot
committed
5c7157c
fix(harness): handle empty model output in summarization

senamakel
and
medullabot
committed
54c766c
test(model_summarizer): add tests for model summarization

senamakel
and
medullabot
committed
ed09cc0
fix(test): update test to use fully qualified function path

senamakel
and
medullabot
committed
a410d19
fix(agent_loop): correct dialect parsing for multi-word agent names

senamakel
and
medullabot
committed
cf71439
fix(harness): handle empty dialect in agent loop

senamakel
and
medullabot
committed
db7f1fb
fix(harness): handle empty dialect list in agent loop

senamakel
and
medullabot
committed
82417a7
fix(harness): handle missing dialect in agent loop

senamakel
and
medullabot
committed
06e8285
fix(harness): handle missing dialect in agent loop

senamakel
and
medullabot
committed
1cc81e3
fix(harness): handle missing dialect in agent loop

senamakel
and
medullabot
committed
b8c4b80
fix(harness): handle empty dialect list in agent loop

senamakel
and
medullabot
committed
5a93dea
fix(agent_loop): handle model call errors without panicking

senamakel
and
medullabot
committed
cbc19d7
fix(agent_loop): handle empty tool call arguments in run loop

senamakel
and
medullabot
committed
e0c59d1
fix(agent_loop): handle empty tool call arguments gracefully

senamakel
and
medullabot
committed
97b476d
fix(agent_loop): handle empty tool call arguments gracefully

senamakel
and
medullabot
committed
a6c2d5f
fix(harness): handle agent loop termination on empty action list

senamakel
and
medullabot
committed
e564a22
fix(agent_loop): handle empty tool call arguments gracefully

senamakel
and
medullabot
committed
dd18b3f
fix(agent_loop): handle empty tool call arguments

senamakel
and
medullabot
committed
b6933a2
fix(harness): handle empty model output in summarization

senamakel
and
medullabot
committed
4edb362
fix(agent_loop): handle empty tool call arguments gracefully

senamakel
and
medullabot
committed
e7b84c2
fix(test): add missing withhold field in recovery test helper

senamakel
and
medullabot
committed
d8920bf
fix(test): correct test assertions for withheld agent loop

senamakel
and
medullabot
committed
eb3823d
fix(test): change max_calls parameter type from u32 to usize

senamakel
and
medullabot
committed
a49a5e9
chore: reformat long lines and function signatures

senamakel
and
medullabot
committed
9e05fba
fix(agent_loop): handle empty tool call arguments in run loop

senamakel
and
medullabot
committed
b7b5cb5
docs(summarization): add README for the summarization module

senamakel
and
medullabot
committed
314e257
docs(agent_loop): add README for the agent loop module

senamakel
and
medullabot
committed
29ecb4e

Navigation Menu

Repository navigation

fix(harness): keep tool-call markup out of summaries and tool-less answers - #277

fix(harness): keep tool-call markup out of summaries and tool-less answers#277
senamakel merged 27 commits into
tinyhumansai:mainfrom
senamakel:deepseek-markup-leak

Commits on Oct 3, 2026

fix(harness): handle empty model output in summarization

fix(harness): handle empty model output in summarization

test(model_summarizer): add tests for model summarization

fix(test): update test to use fully qualified function path

fix(agent_loop): correct dialect parsing for multi-word agent names

fix(harness): handle empty dialect in agent loop

fix(harness): handle empty dialect list in agent loop

fix(harness): handle missing dialect in agent loop

fix(harness): handle missing dialect in agent loop

fix(harness): handle missing dialect in agent loop

fix(harness): handle empty dialect list in agent loop

fix(agent_loop): handle model call errors without panicking

fix(agent_loop): handle empty tool call arguments in run loop

fix(agent_loop): handle empty tool call arguments gracefully

fix(agent_loop): handle empty tool call arguments gracefully

fix(harness): handle agent loop termination on empty action list

fix(agent_loop): handle empty tool call arguments gracefully

fix(agent_loop): handle empty tool call arguments

fix(harness): handle empty model output in summarization

fix(agent_loop): handle empty tool call arguments gracefully

fix(test): add missing withhold field in recovery test helper

fix(test): correct test assertions for withheld agent loop

fix(test): change max_calls parameter type from u32 to usize

chore: reformat long lines and function signatures

fix(agent_loop): handle empty tool call arguments in run loop

docs(summarization): add README for the summarization module

docs(agent_loop): add README for the agent loop module