a more agentic sibling to `btw_app()`
Nobody has claimed this yet.
- Dominant language
- R
- Stars
- 137
- Forks
- 10
- Avg merge
- 10h 48m
- Merged PRs (30d)
- 4
Description
Just jotting some notes down from an internal conversation. tl;dr: btw_app() should have a sibling with a more agentic, "Claude Code Dupe" feel.
As of now, side::kick() is a separate package implementing something like this Claude Code Dupe idea:
- It has a
run_r_code()andbash()tool that are registered automatically and, in most cases, do not require approval. - It has more direct prompting about poise and feel that tell it to "just do stuff and say as little as possible."
- There's no setup flow other than installing the package--the app can discover providers and register them in a persistent way.
That said, side::kick() is quite scrappy and does all sorts of naughty namespacing tricks to make things happen. Because of this, 1) ellmer and shinychat and btw do not benefit when something gets implemented in side::kick(), and 2) side::kick() doesn't benefit when ellmer and shinychat and btw do things the right way.
These issues could be resolved by a sibling to btw_app() that is less configurable and more agentic, e.g. btw::btw_agent() or btw::sidekick() or otherwise. btw_app() can remain a more tailored-to-fit / safe interface to the suite of tools that btw provides.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reading the existing btw_app() interface and the side::kick() behavior described in the issue, then compare how ellmer and shinychat integrate with btw. Done would require a decided sibling API and agreed scope for its more agentic behavior, setup flow, and tool approvals.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- r
- Domain
- ai, backend-api-design
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100