claudehut-workflowlisted

# ClaudeHut Workflow You are operating under ClaudeHut. The codebase is **pre-indexed** (see `claudehut:claudehut-init`). The full workflow is **7 phases** — Discover → Brainstorm → Spec → Plan → Implement → Review → Learn — and you are always in exactly one. ## Flow ```mermaid flowchart TB start(["session start / task in"]) --> tri["TRIAGE — pick tier by the HARDEST QUESTION<br/>(file-count is a floor, not the test)"] tri --> mini{"≥2 genuinely different approaches<br/>OR sensitive/multi-file surface?"} mini -- "yes" --> full["set-complexity full"] mini -- "no" --> fast["set-complexity small/trivial<br/>+ name ≥2 approaches inline (mini-brainstorm)"] fast --> smell{"only ONE option could be rationalized?<br/>(the 'really full' smell)"} smell -- "yes" --> full smell -- "no" --> prof full --> prof["classify PROFILE by DELIVERABLE<br/>set-profile feature|bugfix|audit|investigation|migration"] prof --> pset{"profile set?"} pset -- "no" --> blocked(["BLOCKED: set-phase implement denied<br/>until set-profile"]) blocked -.-> prof pset -- "yes" --> phase["set-phase → walk profile's phase chain"] phase --> wgate{"PreToolUse write gate:<br/>reuse-scan+ (full: spec+plan) + implement-skill rail?"} wgate -- "no (fast-lane bound exceeded)" --> esc(["escalate: set-complexity full → Spec+Plan"]) esc -.-> full wgate -- "yes" --> done{"gate-done predicate met for THIS profile?<br/>feature/bugfix/migration: review==pass ·<br/>audit/investigation: set-findings