Modify workflows to support easy addition/replacement of agents

[ggallen]

Currently the workflows have hardcoded values. The shim calls specific workflows in the .fullsend repository which run the agents.

This is not very flexible, and makes it much more difficult to swap in a replacement agent or have multiple agents run per stage.

The workflows needs to change to a dispatch model, when the shim sends the stage name to a dispatch router in fullsend, which in turn will use a repository_dispatch to start any agents that match.

[ggallen]

Resolved by #390.

[fullsend-ai-triage]

Meeting update — 2026-05-06

Relevant to this issue: Administrator feedback explicitly requested flexible, customizable agent workflows that allow easier addition and replacement of agents.

• Feedback from potential administrators highlighted significant architectural concerns about the current workflow structure
• Users want easier integration paths and the ability to customize which agents run and how they interact
• This reinforces the need for workflows that support modular agent addition/replacement

Related PRs: PR #660 (namespaced slash commands) addresses a related customization concern.

Next steps: User journey documentation is being written that will inform the required workflow flexibility.

Meeting notes