the-promotion-committee
GitHub模拟晋升委员会校准会议，通过四种典型角色辩论用户晋升材料。输出会议记录、真实判决及可修复差距分析，区分客观不足与政治因素，帮助优化晋升策略。
Trigger Scenarios
Install
npx skills add mohitagw15856/pm-claude-skills --skill the-promotion-committee -g -y
SKILL.md
Frontmatter
{
"name": "the-promotion-committee",
"description": "Simulate the calibration meeting that discusses your promotion after your manager leaves the room — the debate, the packet's holes, the verdict. Use when asked will I get promoted, simulate the promo committee, stress-test my promotion packet, or why did my promo get rejected. Produces the committee transcript (four archetypes on YOUR packet), the internal verdict with the real reason, and a debrief separating fixable gaps from timing politics."
}
The Promotion Committee Skill
Promotions are decided in a room you're not in, by people comparing you to candidates you've never met, reading a packet in four minutes. This skill runs that room early: your actual evidence, debated by the archetypes calibration meetings really contain — then breaks character to tell you which objections are fixable this cycle and which are politics wearing a rubric.
What This Skill Produces
- The transcript — four committee archetypes debating your packet, 12–18 exchanges
- The verdict memo — promote / hold / needs-a-cycle, with the real reason vs the feedback you'd be given
- The debrief — gaps ranked by fixability, with the evidence that would close each
Required Inputs
Ask for these if not provided:
- The packet substance — accomplishments with scope/impact, the level sought, the ladder criteria if available
- The manager's pitch — how strongly and with what evidence they'll advocate (honest read)
- The context — how many slots vs candidates, your tenure at level, any known skeptics
- The known weakness — the thing you hope nobody asks; the committee always asks
Framework: The Four Chairs
| Archetype | What they weigh | Their tell |
|---|---|---|
| The Sponsor (your manager's proxy) | Sells your two strongest artifacts | Overreaches on scope claims — invites the challenge |
| The Bar-Keeper | The ladder text, literally | "That's solid at-level performance" — the sentence that kills promos |
| The Comparator | You vs the other packets THIS cycle | "Both are strong; we have one slot" |
| The Skeptic-of-Evidence | Whether impact claims survive reading | "Led — or participated? The doc says the team shipped it" |
Committee mechanics to honor: packets get minutes, not hours — the first artifact carries disproportionate weight · "next-level work already happening" beats "ready for next-level work" everywhere · scope inflation, once caught, taints the true claims too · absence of a named sponsor beyond the manager is itself discussed.
Output Format
Calibration: [name] → [level] — [cycle]
Simulation — a plausible adversarial reading, not a prediction.
Transcript
[The debate, grounded ONLY in the supplied packet; where evidence is missing, the committee notices the absence — that's the feedback]
Verdict Memo
Outcome: PROMOTE / HOLD (real reason: …) — and the softer version HR will relay What tipped it: … The comparison that mattered: …
Debrief — out of character
| Gap the committee found | Fixable this cycle? | The evidence that closes it |
|---|---|---|
| [Plus: the one artifact to lead the next packet with, and the sponsor-beyond-your-manager problem if present] |
Quality Checks
- Every committee claim traces to the supplied packet or a noticed absence
- The Bar-Keeper quotes level criteria against specific accomplishments
- The verdict's real reason differs from the relayed feedback when it realistically would
- The debrief separates evidence gaps from slot/timing politics honestly
- "At-level vs next-level" framing appears — it's the axis promotions actually turn on
Anti-Patterns
- Do not pull punches — a committee that loves the packet teaches nothing
- Do not invent accomplishments to debate — absences ARE the finding
- Do not let the Sponsor win by enthusiasm — packets win on evidence the skeptic can't dent
- Do not present politics as fixable — naming the unfixable is the kindest output
- Do not stay in character in the debrief
Version History
- 54fad50 Current 2026-07-19 13:11
