feat: Mod - Adding test harness to Mod Api to make it easier to write tests and have agent verify mod works

Commencez par examiner la Mod API et la skill th-mod-builder, puis comparez les commandes Node d’exemple de l’issue avec le workflow actuel du mod. Définissez ce que le harness doit couvrir, comment il doit être exécuté manuellement ou via l’agent, et à quoi doivent ressembler le chargement réussi du module et les vérifications de la sortie ; aucun fichier d’implémentation ni test spécifique n’est identifié.

Summary

When writing and iterating on a mod, I find the the model will write a temp script which imports some of the mod functions to test the output and if it works properly, for example:

xecute Shell Command
Command Code needs to execute node --experimental-strip-types --no-warnings -e "
Promise.all([
  import('./.commandcode/mods/erd.ts'),
  import('./.commandcode/mods/erd/markdown.ts'),
  import('./.commandcode/mods/erd/renderer.ts'),
  import('./.commandcode/mods/erd/commands.ts'),
  import('./.commandcode/mods/erd/tools.ts'),
]).then(() => { console.log('ALL MODULES LOAD OK'); }).catch(e => { console.error('FAIL', e); process.exit(1); });
" 2>&1 | head -20.

Press [ctrl+e] to explain this command

❯ 1. Yes

Some are more involved and check output not just if load.

Execute Shell Command
Command Code needs to execute node --experimental-strip-types --no-warnings -e "
const fs = require('fs');
Promise.all([
  import('./.commandcode/mods/erd/parser.ts'),
  import('./.commandcode/mods/erd/queries.ts'),
  import('./.commandcode/mods/erd/markdown.ts'),
]).then(([p, q, md]) => {
  const sql = fs.readFileSync('schema/schema.sql', 'utf8');
  const schema = p.parseSchema(sql, 'sqlite');
  console.log('listTables →', md.renderMarkdown(q.listTables(schema)).length, 'lines');
  console.log('listIndexes →', md.renderMarkdown(q.listIndexes(schema)).length, 'lines');
  console.log('listForeignKeys →', md.renderMarkdown(q.listForeignKeys(schema)).length, 'lines');
  console.log('findColumn →', md.renderMarkdown(q.findColumn(schema, 'id')).length, 'lines');
  console.log('schemaSummary →', md.renderMarkdown(q.schemaSummary(schema, 'sqlite')).length, 'lines');
  console.log('describeTable →', md.renderMarkdown(q.describeTable(schema, 'users', 'sqlite')).length, 'lines');
  console.log('ALL QUERY FUNCTIONS RENDER CLEANLY');
}).catch(e => { console.error('FAIL', e); process.exit(1); });
" 2>&1 | head -20.

Press [ctrl+e] to explain this command

❯ 1. Yes
  2. Yes, don't ask again for this exact command in this project
  3. No, tell Command Code what to do differently

Expected Behavior

Would be useful as mods get more complex an refactor to have a proper test harness or suite like vitest that can be run easily manually or within agent as part of th mod-builder skill.

Actual Behavior

Mode has to cobble together one off scripts and runs that have to be permitted each Tim since the command is unique often per run so hard to write permissions to allow.

Can see from example above the shell command on-line a script and not sure I'd want to permit blanket node --experimental-strip-types --no-warnings -e allow permissions here.

Steps to reproduce the issue

1. build mod
2. try to test it

Command Code Version

1.7.0

Operating System

macOS

Terminal/IDE

ghostty

Shell

zsh

Additional context

No response