| name | chilla-post-edit-launch |
| description | Use when code changes in this repository should be validated by actually launching the local chilla app. After modifying runtime-affecting code, rebuild the debug app when needed and launch `target/debug/chilla` yourself instead of telling the user to run it. |
| allowed-tools | Read, Grep, Bash |
Chilla Post-Edit Launch
Use this skill after code changes that affect the chilla app runtime, especially UI, startup, file browsing, Tauri wiring, or preview behavior.
Core Rule
Do not ask the user to run chilla manually when you can do it locally.
After relevant code changes:
- Rebuild the local debug app when the binary may be stale.
- Launch
target/debug/chilla yourself.
- Reuse the user's repro arguments when they provided them.
- Report the exact command and log path back to the user.
When To Launch
Launch the app yourself when:
- frontend UI or interaction code changed under
src/
- Tauri/backend runtime code changed under
src-tauri/
- startup, navigation, preview, dialog, or window behavior changed
- the user asks you to verify runtime behavior
You can skip launching when:
- only docs, plans, or comments changed
- the user explicitly says not to run the app
Build And Launch Workflow
1. Rebuild when needed
Prefer the repository debug build:
bun run tauri build --debug --no-bundle
Use this when Rust, Tauri config, or bundled frontend assets changed, or when you are unsure whether target/debug/chilla is current.
2. Launch the local binary yourself
No args:
nohup /absolute/path/to/repo/target/debug/chilla >/tmp/chilla-launch.log 2>&1 &
With repro args:
nohup /absolute/path/to/repo/target/debug/chilla <args...> >/tmp/chilla-launch.log 2>&1 &
Use an explicit log file under /tmp/ and tell the user where it is.
3. Prefer the user’s reproduction command
If the user gave a concrete launch command or sample-file path, reuse it with the rebuilt binary rather than inventing a different startup path.
Response Contract
After launching, tell the user:
- whether you rebuilt
- the exact launch command you executed
- the log file path
Do not push the launch step back to the user unless local execution is genuinely blocked.