ix-plan

ix-claude-plugin shared model

Check command -v ix first. If unavailable, use Grep + Read to manually assess blast radius per target.

Argument parsing

Strip --save and any following path token from $ARGUMENTS before resolving targets.

• If --save <path> is present, set SAVE_PATH to that path.
• If --save is present without a path, auto-generate ix-plan-<target-slug>.md in cwd (target slug = the first target or first three words of the description with spaces and slashes replaced by -).
• If --save is absent, SAVE_PATH is empty — do not write a file.

Pro check (optional)

Run once at the start:

ix briefing --format json 2>&1

If it returns JSON with a revision field, Pro is available. Extract activeGoals, activePlans, and openBugs for use in Pro steps below. If it errors, skip all [Pro] labeled steps.

Goal

Answer: in what order should these changes be made, what will break, and what needs testing?

Phase 1 — Scope (always)

If $ARGUMENTS contains symbol names, proceed. If $ARGUMENTS is a description (no identifiable symbols), first run:

ix text "$ARGUMENTS" --limit 10 --format llm
ix locate "$ARGUMENTS" --format llm

Identify the 1–4 most relevant symbols and treat those as targets.

Phase 2 — Impact per target (parallel)

For each identified target, run simultaneously:

ix impact  <target> --format llm
ix callers <target> --limit 10 --format llm

Rank targets by risk level: critical > high > medium > low.

Fast path — all low risk: If every target is low risk AND has < 3 dependents, skip Phases 3–5. Go directly to Output with verdict "SAFE — all targets low risk; no additional data-flow, shared-dependent, or project-context analysis needed."

Delegation gate — high-complexity path: If the fast path did not trigger, check for high complexity:

1. From Phase 2 results: does any target have dependents > 20?
2. If not already known, run ix subsystems --format llm (reads cached data — cheap) and check if any non-low-risk target's region has coupling > 5
3. If either condition is true:
   • Spawn ix-safe-refactor-planner with pre-filled context:
     • TARGETS: the resolved symbol list from Phase 1
     • RISK_TABLE: the ranked table from Phase 2 (agent skips its own Steps 1–3)
     • SUBSYSTEMS: subsystems output from step 2
   • Stop — the agent produces the full sequenced plan