Errors your AI
can act on.
Give Claude Code the error, the source line, and a way to check the fix. Console exposes AutoHotkey failures as structured diagnostics. ClautoHotkey brings that feedback into your editing workflow.
SOURCE / ERROR-DEMO.AHK · LINE 4
#Requires AutoHotkey v2.1-alpha.30
settings := Map()
settings["retries"] := 3
attempts := settings["retry"]STDERR / JSON ERROR · SELECTED FIELDS
{
"type": "UnsetItemError",
"message": "Item has no value.",
"extra": "[\"retry\"]",
"file": "error-demo.ahk",
"line": 4,
"source": "attempts := settings[\"retry\"]"
}The error reaches the calling tool with its type, source file, line, and failing statement.
Recorded Console + Claude Code example · How it was verified
Structured failures give Claude a specific problem to diagnose and repair.
Line Out shows which statements ran, with source files and line numbers.
Send values and JSON to stdout for tools, pipelines, and assertions.
- Run the scriptExecute your AutoHotkey code
- Capture the errorConsole reports the error, file & line
- AI reads & fixesClaude uses the feedback to edit
- Run checks againCheck syntax, runtime & tests
- Fix verifiedChecks pass with the expected result
Still failing? Send the new error back to the AI.
THE FEEDBACK LOOP
Give Claude the
missing context.
The error class explains what failed. The file, line, and source text show where to look. Claude can use those details to propose a focused edit, then read the next check or runtime result.
Connect edits to checks1. Start with a real failure
error-demo.ahk Download the failing example ↓2. Capture an error the AI can read
ahk /Headless /Diag=json .\error-demo.ahkFirst time here? Build the Console executable and set up the ahk alias.
stderr carries the diagnostic; exit 10 signals this runtime failure. Read the diagnosis, correction, and verified rerun.
CONSOLE + CLAUTOHOTKEY
Build the repair loop.
Connect the interpreter, editor, and verification tools.
Readable failures
Error types, source locations, captured stderr, and exit codes.
Claude Code feedback
ClautoHotkey hooks check edits; local MCP tools expose results to the AI.
Checks & tests
Syntax checks, runtime assertions, and coverage.
Explore the full toolkit
Structured diagnostics
Give AI the error type, source location, failing statement, and exit code it needs to diagnose a script.
Console Read the guideLine Out & trace
Follow the statements that actually ran, with source files and line numbers to explain the path to a failure.
Console Read the guidePrint & JSON
Turn a script into a composable command-line tool, with UTF-8 output and native JSON.
Console Read the guidePersistent REPL
Keep state between expressions. Explore functions and recover from invalid input.
Console Read the guideInspect & Check
Describe values as JSON and ask the interpreter to check source before running it.
Console Read the guideProcessPipe
Exchange UTF-8 lines with a child process and manage its lifetime.
Console Read the guideAutomatic edit checks
A real Claude Code hook checks AHK syntax after Write or Edit, with an optional runtime probe.
ClautoHotkey Read the guideSkills & knowledge
Load task-specific AHK guidance for GUIs, COM, classes, debugging, and more.
ClautoHotkey Read the guideGrading gates
Keep static and dry-run findings, parser confidence, and intercepted effects in a shared format.
ClautoHotkey Read the guideNative MCP tools
Let a local tool client check, run, test, and explore source through the same engine.
Together Read the guideTest → inspect → repair
Combine assertions, structured failures, and source context into a repeatable repair loop.
Together Read the guideCrash forensics
Preserve startup, error, and exit events when stderr alone is not enough.
Console Read the guideWant to inspect a complete workflow? See the real session gallery →
The error demonstration was recorded with alpha.31 +Console and Claude Code. Check your executable’s capabilities before using development features.
Build compatibility →