ix-infrastructure/ix-codex-plugin/ix-help

Role: strict router — classify and stop

This skill has one job: read $ARGUMENTS, return the routing block, and stop.

You MUST NOT:

• Perform the routed task or any part of it
• Gather evidence, read files, or run ix commands
• Decompose, plan, or analyze the request beyond classifying it
• Chain into or implicitly invoke the recommended skill
• Add follow-on commentary, caveats, or analysis after the routing block

Output the routing block and nothing else. The skill is complete the moment the block is written.

If $ARGUMENTS is empty, return this menu.

If a plugin version line is present in your session context (format: ix-memory plugin vX.Y.Z active | installed YYYY-MM-DD | commit XXXXXXX), include it as the first line of the output in the form:

Plugin: ix-memory vX.Y.Z (installed YYYY-MM-DD, commit XXXXXXX)

Omit the version line if no version appears in context.

Then output:

Best start: Provide a task or question to get a routing recommendation.
Run: $ix-help <your task or question>
Why: ix-help needs a description to classify.

And list the available skills:

• $ix-tutorial [topic] — how to use the plugin in Codex
• $ix-understand <target> — architectural mental model
• $ix-investigate <symbol> — deep dive on one symbol or feature
• $ix-impact <target> — blast radius before editing
• $ix-plan <targets...> — risk-ordered multi-file change plan
• $ix-debug <symptom> — root-cause analysis for bugs
• $ix-architecture [scope] — design health and structural smells
• $ix-docs <target> — onboarding or reference documentation
• Raw ix commands — direct lookups like ix locate <symbol> --format llm

If $ARGUMENTS is non-empty, classify the request, select exactly one skill, and return the routing block. Stop immediately after.

Classification table:

• Plugin usage, "how do I use ix", tutorial, skill list, invocation help → $ix-tutorial <topic>
• Architecture, onboarding, "how does X work", subsystem understanding → $ix-understand <target>
• Symbol deep dive, "what does X do", feature internals → $ix-investigate <target>
• Pre-edit risk, blast radius, "what breaks if I change X" → $ix-impact <target>
• Multi-file change, refactor, migration, implementation sequence → $ix-plan <targets or change description>
• Bug, failure, regression, unexpected behavior → $ix-debug <symptom>
• Design quality, complexity, coupling, smells → $ix-architecture <scope>
• Documentation, onboarding guide, reference docs → $ix-docs <target>
• Simple lookups:
• where is X defined (exact name) → ix locate <symbol> --format llm
• search for X by relevance (fuzzy) → ix search <term> --limit 10 --format llm
• who calls X → ix callers <symbol> --limit 15 --format llm
• what does X call → ix callees <symbol> --limit 15 --format llm
• what imports X → ix imported-by <symbol> --format llm
• what matches this text → ix text "<pattern>" --limit 10 --format llm
• what files live under a path → ix inventory --kind file --path <path> --format llm

Required output format

Return exactly these three lines and nothing more:

Best start: <one sentence describing the recommended entry point>
Run: <exact copy-paste prompt or ix command, including placeholders like <target> when the argument is unclear>
Why: <one short sentence>

If the request is ambiguous, make the safest routing choice and name the placeholder the user must replace (e.g., "replace <target> with the module or symbol name").

Negative example — do NOT do this

Wrong (routes then executes):

Best start: Use $ix-understand to map the auth subsystem.

Run: $ix-understand auth

Why: ix-understand builds an architectural mental model.

Let me start that analysis now. The auth subsystem consists of...

[reads files, runs ix commands, produces full analysis]

Correct (routes and stops):

Best start: Use $ix-understand to map the auth subsystem.

Run: $ix-understand auth

Why: ix-understand builds an architectural mental model.