The addon already has a way to configure API keys, in this case we should add an API key and an end-point that can be queried using OpenAI v1 API, as well as the model to be used for comment processing, which should work with OpenAI platform and with Ollama self-hosted versions of models.
Such LLM can probably be used with some prompting logic to extract all bug numbers listed in comments, and determine what are the relations with the current bug. We should probably filter out comment which include Bug word in them to reduce the overhead of giving inputs to the LLMs.
With this new tools, we should be able to extract bugs which are dependent on, blocked, regressed by and see also, in order to made recommendation to the persons who are triaging bugs.
Looking at existing regressed by, we are quite good at updating the bug fields as when we comment about it … so I do not think any tool would help us with this specific use case.
The addon already has a way to configure API keys, in this case we should add an
API key
and anend-point
that can be queried using OpenAI v1 API, as well as themodel
to be used for comment processing, which should work with OpenAI platform and with Ollama self-hosted versions of models.Such LLM can probably be used with some prompting logic to extract all bug numbers listed in comments, and determine what are the relations with the current bug. We should probably filter out comment which include
Bug
word in them to reduce the overhead of giving inputs to the LLMs.With this new tools, we should be able to extract bugs which are
dependent on
,blocked
,regressed by
andsee also
, in order to made recommendation to the persons who are triaging bugs.