ape

Plate III

Simia studiosa

Habitat

A research workspace of cloned repositories, outside your project.

Diet

A named gap and a GitHub search, or one repository URL.

Field marks

Read-only in the current project. Every recommendation cites source lines and proposes a small experiment. Never runs third-party code.

Find ideas that transfer to your project.

When to use it

You have a specific local gap and want to compare an approach with working source. Use study for one named GitHub repository; use forage to discover candidates.

Try it

After installation, run in your terminal:

t3-plugins prompt ape doctor

Paste the printed line into a T3 thread attached to your project. Once readiness is clear, generate a task prompt:

t3-plugins prompt ape study https://github.com/python/cpython 'How are parser error cases tested? Compare with this project and propose a small experiment.'

Paste that output into T3. Replace the sample task or path with your own. If you installed skill adapters, select t3-ape from / and append the same request instead.

Operations

doctor, forage, study, clean. The entry file names the file to load for each operation.

What you get

A report pinned to the source commit, with Overlap, Adopt, Adapt, Reject and Command case sections. Each recommendation names local target files, effort, tradeoffs and a bounded experiment. An empty Adopt section is a valid result. The agent returns the source revision and report path; proposed experiments are not measured improvements.

Limits

Cleanup accepts only workspaces carrying a matching .ape-workspace.json marker from the initializer. Unmarked legacy folders are left untouched; inspect and archive them manually before creating a new workspace. Initialization does not adopt existing data.

Research stays read-only in the current project. It never executes third-party code or silently starts implementing recommendations. See compatibility for the distinction between deterministic tests and live provider results.