wip: refactor of cli interface to include prompt, agent, list.#6
Merged
vsoch merged 2 commits intoadd-manager-agentfrom Feb 27, 2026
Merged
wip: refactor of cli interface to include prompt, agent, list.#6vsoch merged 2 commits intoadd-manager-agentfrom
vsoch merged 2 commits intoadd-manager-agentfrom
Conversation
Insights include: The planner agent should not be tasked with multiple steps, but just one. I am finding the using single steps (each an agent with full access to tools and ability to respond to error) is a more robust approach than a pre-encoded state machine. We can still combine sub agents for more complex orchestration, and probably when this maps to multi-scheduling we will need that separation anyway. Hugely problematic when we do not tell it about data or how to find software. When it has trouble with something, it starts getting messy with other details. We also need a more intelligent mcp server that handles task queues and event delivery. I can build it. Note that the changes are well documented for this WIP but it is not working yet - I need time to work on it. Signed-off-by: vsoch <vsoch@users.noreply.github.com>
Member
Author
|
Listing of available sub-agents. Defaults: 
Adding the optimize on the fly: 
That prints as json too. I can also probably make this a verbose mode (and non-verbose is just the names and descriptions). |
The manager is now a general manager sub-agent. We still use it alongside the state machine, but now other tools can decide to use it too. Signed-off-by: vsoch <vsoch@users.noreply.github.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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Insights include: The planner agent should not be tasked with multiple steps, but just one. I am finding the using single steps (each an agent with full access to tools and ability to respond to error) is a more robust approach than a pre-encoded state machine. We can still combine sub agents for more complex orchestration, and probably when this maps to multi-scheduling we will need that separation anyway. Hugely problematic when we do not tell it about data or how to find software. When it has trouble with something, it starts getting messy with other details. We also need a more intelligent mcp server that handles task queues and event delivery. I can build it. Note that the changes are well documented for this WIP but it is not working yet - I need time to work on it.