# St Eval

> Smalltalk code evaluator for Pharo or Squeak via MCP. Use when executing Smalltalk expressions, verifying object state or intermediate values, debugging code incrementally, checking the Smalltalk image connection, or running quick experiments.

- Skill: `mumez/st-eval` (Agent Skill)
- Install (CLI): `npx skillmds@latest add mumez/st-eval`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mumez/st-eval/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: mumez (https://skillmd.com/u/mumez)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/mumez/st-eval

---


# Execute Smalltalk Code

Evaluate arbitrary Smalltalk expressions in the running Smalltalk image via `mcp__smalltalk-interop__eval`.

## Essential Rules

**Always use `printString`** — MCP returns serialized output. Raw objects cause errors or unreadable results:

```smalltalk
✅ MyClass new doSomething printString
✅ collection printString
❌ MyClass new doSomething
```

**Wrap risky code in `on:do:`** — Uncaught errors in eval crash the MCP call with no useful message:

```smalltalk
| result |
result := Array new: 2.
[ result at: 1 put: (riskyOperation) printString ]
  on: Error do: [:ex | result at: 2 put: ex description].
^ result
```

**Use `fork` for blocking operations** — Code that opens a dialog or blocks the image will cause the MCP call to hang indefinitely. Detach it:

```smalltalk
[ <blocking expression> ] fork.
^ 'started'
```

## Common Patterns

### Connection check
```smalltalk
Smalltalk version
```

### Inspect intermediate values
```smalltalk
| step1 step2 |
step1 := objA computeStep1.
step2 := step1 processStep2.
^ { 'step1' -> step1 printString. 'step2' -> step2 printString } asDictionary printString
```

### Inspect object state
```smalltalk
{
  'class' -> obj class name.
  'value' -> obj printString.
  'size'  -> obj size printString
} asDictionary printString
```

### Safe collection access
```smalltalk
collection ifEmpty: ['empty'] ifNotEmpty: [:col | col first printString]
```

### Error-capturing eval
```smalltalk
| result |
result := Array new: 2.
[
  | obj |
  obj := MyClass new name: 'Test'.
  result at: 1 put: obj process printString.
] on: Error do: [:ex |
  result at: 2 put: ex description
].
^ result
```

## Troubleshooting

| Symptom | Cause | Fix |
|---------|-------|-----|
| MCP call hangs / no response | Blocking operation (dialog, modal, infinite loop) | Wrap in `fork` |
| `Error: cannot serialize` | Returning raw object | Add `printString` |
| Debugger window opened in the Smalltalk image | Uncaught error triggered the debugger | Close debugger in the Smalltalk image, then re-eval with `on:do:` |
| `MessageNotUnderstood` | Method doesn't exist or typo | Check with `mcp__smalltalk-interop__search_implementors` |

## Related Skills

- `smalltalk-dev:smalltalk-debugger` — Systematic debugging workflow using eval
- `smalltalk-dev:st-init` — Verify the Smalltalk image connection if eval fails unexpectedly