Senior + Principal + Distinguished + Executive Engineer Mindset (router)

The gap between junior and senior isn't "how well you write code" — it's what you think about before you write it. This skill bundles that thinking process — four stacked lenses (Senior/Principal/Distinguished/Executive), a workflow of sub-skills, and a three-track classification (Spike/Bounded/Structural) — and is tuned per model, because how much hand-holding, effort, and narration each part needs differs by tier.

Read the file matching the model actually running before applying this skill. Each file is a complete, standalone rewrite covering the four-lens framing, the sub-skill and situational-pick tables, track classification, the "too simple to need a check" trap, output format, and the skip list — calibrated for that model. The hook below is not model-dependent and applies the same way regardless of which file you read.

Running a model not listed above (an older generation, a different model family entirely)? sonnet-5.md carries the fullest general-purpose version of the underlying procedure — read that one rather than skipping this skill.

Optional: enforce with a hook

The rule that a hard-to-reverse decision gets a check-in first is mechanically catchable to a degree: scripts/check_ask_before_hard_change.py watches for an Edit/Write/MultiEdit targeting a path that looks like a hard-to-reverse surface (schema, migration, public API spec, dependency manifest) with no AskUserQuestion call in the recent window (ASK_HOOK_WINDOW_SECONDS, default 1800s). By default it only warns, it doesn't block — set ASK_HOOK_STRICT=1 to make a hit actually block the edit.

Add to .claude/settings.json:

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Edit|Write|MultiEdit",
        "hooks": [
          { "type": "command", "command": "python3 .claude/skills/senior-engineer-mindset/scripts/check_ask_before_hard_change.py" }
        ]
      }
    ],
    "PostToolUse": [
      {
        "matcher": "AskUserQuestion",
        "hooks": [
          { "type": "command", "command": "python3 .claude/skills/senior-engineer-mindset/scripts/check_ask_before_hard_change.py" }
        ]
      }
    ]
  }
}

If your skill is installed at a different path, update the command path to match. This is a path-based heuristic, not a real understanding of what's actually hard to reverse — it can't tell a real schema change from an unrelated edit to a file that happens to live in migrations/, and it can't tell a dependency version bump from a comment-only edit to a manifest file. A false positive just costs a wasted nudge (advisory by default); it isn't a substitute for actually judging whether a decision is hard to reverse. Same example-level-enforcement framing as delegate-to-subagents's bundled hook — not a red-team-tested CRITICAL-tier block. See delegate-to-subagents's hook section for the plugin that does provide that tier (songjiun10-collab/hook).