https://benhouston3d.com/blog/building-an-agentic-code-from-...
My chronology can be summarized as:
Phase 1:
- Add async shell commands with input/output/status checking/kill - combines all of the required functionality.
- text editing that can read/write sub-blocks, rather than whole files. But not multiple files yet (but I can look into that now.)
- Add fetch
- Add browser tools with async input/output
- Add sub-agents (to reduce context, send them on research tasks or other sub-tasks to report back.)
Phase 2:
- Github mode (basically prompt it to use git and gh cli tools in its workflows)
- Github action integration (run automatically on the cloud in response to triggers)
- Token caching (reduce costs by 80%)
I will check out your planning stage. I've been having my tool run off GitHub issues so I get it to write a plan there and then implement, which is sort of the same thing in a way.
Chronology that followed:
1. Some processes take long time: add status checking.
2. Need to kill long running process or interact with tools add option to send control characters; BashInteraction tool.
3. File editing using sed and echo is not reliable, add FileEdit tool.
4. Large output; add truncation based on tokens.
5. But wait I need command output to get truncated at the beginning but file reading `cat $file` to truncate at the end. Add special ReadFile tool.
6. File writing using 'echo "content" >file' needs unncessary quotes escaping leading to degraded quality, add special WriteFile tool.
7. Reading one file at a time is slow, change ReadFile to ReadFiles tool.
8. Directly writing files is not as good as first planning things. Prevent writing using architect mode.
And so on.
[1] https://github.com/rusiaaman/wcgw