If we had 2 agents that could take in context, the changes pushed to the main branch and the corresponding commit message to generate the media (videos and images) using vhs and freeze, it would be a huge boost in content generation and product visibility.
I'm thinking of 2 different agents: vhs_agent and freeze_agent for generating terminal GIF and SVG, respectively.
The agent has to come up with a script that will showcase the changes pushed to the main branch and will be available in the next release of tandem.
if the author pushed a commit with message
<commit_message>
implemented edit, list, read, grep and patch tool for the agent.
</commit_message>
- vhs_agent has to write a
.tape file that would showcase the agent using those tools during the inference and provide a rich anecdote of the feature.
- freeze_agent would have to come up with commands or could simply use a predefined config file for freeze tool to get svg screenshots.
we can make use of github copilot coding agent and github action runners to setup the dev env for the coding agent here.
If we had 2 agents that could take in context, the changes pushed to the main branch and the corresponding commit message to generate the media (videos and images) using vhs and freeze, it would be a huge boost in content generation and product visibility.
I'm thinking of 2 different agents: vhs_agent and freeze_agent for generating terminal GIF and SVG, respectively.
The agent has to come up with a script that will showcase the changes pushed to the main branch and will be available in the next release of tandem.
if the author pushed a commit with message
<commit_message>
implemented edit, list, read, grep and patch tool for the agent.
</commit_message>
.tapefile that would showcase the agent using those tools during the inference and provide a rich anecdote of the feature.we can make use of github copilot coding agent and github action runners to setup the dev env for the coding agent here.