Step 1 of 5
Audit your current tool surface
You cannot design an interface you have not honestly measured.
Audit my agent's current tools against ACI principles. For each tool record: typical output size in tokens, whether output is bounded, whether errors state a cause and suggested next action, whether misuse is guarded (e.g. editing a file never read), and 2-3 transcript examples where the tool confused the model. Score each tool 1-5 on model-friendliness. Research context: SWE-agent found that command simplicity, output windowing, and informative feedback measurably raise task success; raw shells underperform purpose-built commands. Deliver the audit as a table with a fix priority order.
Expected after this step
A scored tool audit with prioritized fixes.
Should not happen
- ✕Giving the model a bare shell and calling it an interface
- ✕Tool outputs that dump thousands of lines and drown the signal
- ✕Edits that silently corrupt files because nothing validated them
- ✕Error messages written for humans (stack spew) instead of models (cause + next step)
Verify before continuing
Do not move on until every check is true. The complete button stays locked until then.
Do not continue if…
- !Giving the model a bare shell and calling it an interface
- !Tool outputs that dump thousands of lines and drown the signal
- !Edits that silently corrupt files because nothing validated them
- !Error messages written for humans (stack spew) instead of models (cause + next step)
If the AI messes this up
Use this when the AI fakes progress or breaks the feature. It forces a real fix.
The audit lacks evidence. Pull five recent failed transcripts and trace each failure to the tool interaction that caused it - then rescore.