Politics Coach Skill
Use this skill when the user is focused on organization politics, power dynamics, or "what would X say?" rather than only the content of the product work. It aligns with the other skills: it tells you when to load which frameworks and how to simulate from stakeholder avatars. Setup (the quiz that creates avatars) lives in the framework; this skill only navigates to it and uses the resulting data.
When to use this skill
Trigger this skill when the user:
- Mentions politics, landmines, power games, or career risk, e.g.:
- "The politics on this are tricky." - "I don't want to step on toes." - "This VP could kill the project." - "My manager is risk-averse about this."
- Asks for stakeholder POVs in absentia:
- "What would my manager say about this plan?" - "How will Eng + Design react to this roadmap change?" - "Can you sanity-check this through my stakeholders' eyes?"
- Wants help sequencing conversations or building a coalition:
- "Who should I talk to first?" - "How do I avoid surprising people?"
If they're mainly focused on documents and tactics (one-pagers, escalations, saying no), prefer stakeholder-management and combine with this skill when politics are a big part of the problem.
Relevant framework and data locations
| Purpose | Path |
|---|---|
| Set up or update avatars (guided quiz, methodology) | 02-Methods-and-Tools/2.4-Communication/2.4.8-Stakeholder-Avatars/ — load README + 1-stakeholder-avatars-framework.md and walk the user through; write each avatar to a new file in 01-Company-Context/1.1-Stakeholder-Avatars/ using the naming convention in that folder's README |
| Avatar data (who to simulate) | 01-Company-Context/1.1-Stakeholder-Avatars/ — one file per person (e.g. 1-jane-manager.md); list folder and load the relevant N-name-role.md by name/role |
| Stakeholder communication tactics (one-pagers, saying no, escalation) | 02-Methods-and-Tools/2.4-Communication/ — use with stakeholder-management skill |
Typical flows
1. "Set up my stakeholder brainfeed" / "Create stakeholder avatars"
- Load 2.4.8-Stakeholder-Avatars README and 1-stakeholder-avatars-framework.md.
- Run the framework's quiz (pick cast, then per-person questions from the framework).
- Write each avatar to a new file in 01-Company-Context/1.1-Stakeholder-Avatars/ using the naming convention in that folder's README (e.g. 1-jane-manager.md) and the structure in 2-avatar-template.md.
2. "What would [Name] say about this?" (single-stakeholder simulation)
- List 01-Company-Context/1.1-Stakeholder-Avatars/ and load the file matching the name/role (e.g. 1-jane-manager.md).
- Summarize their lens in 1–2 lines (goals, fears, style).
- Respond in three parts:
- Out-loud reaction — what they'd likely say in the meeting. - Inner monologue — what they'd be thinking but not say. - Top 2–3 concerns / objections — concrete, not generic.
- Close with how to de-risk for that person: evidence that would help, framing or sequencing that would land better.
If no avatars exist, offer to run the setup framework (2.4.8) first, or do a lightweight inline simulation if the user has given enough context and doesn't want to set up avatars yet.
3. "How will [Manager], [Eng], [Design] react?" (panel simulation)
- Load each relevant avatar file from 01-Company-Context/1.1-Stakeholder-Avatars/.
- For each: name/role, likely stance (supportive / cautious / opposed / conflicted), key concern.
- Synthesis: where they align, where they disagree, suggested conversation sequence (who to talk to first, who to warm up, who to keep in the loop).
4. Politics pass in product_sense or execution_mode
- After braindump is sufficient (product_sense): optionally offer "Do you want to run a quick politics check on this through your manager / key stakeholders' eyes?"
- When drafting communication (execution_mode): use this skill with stakeholder-management to check the draft against relevant avatars and suggest tone, ordering, and conversation sequence.
Guardrails
- Treat avatars as caricatures, not truth; add a short caveat when stakes are high.
- Encourage the user to validate in real conversations; suggest 1–2 concrete questions they could ask to test assumptions.
- Surface biases (e.g. all avatars described as hostile, or underestimating a powerful stakeholder).
For Agents
- When you see politics-heavy language or "what would X say?":
- Wake this skill and 01-Company-Context/1.1-Stakeholder-Avatars/ (list folder and load the relevant N-name-role.md by name/role) if the folder exists. - If avatars don't exist and the user wants simulation, offer to run 2.4.8-Stakeholder-Avatars to create them.
- When simulating: use the three-part structure (out-loud, inner monologue, concerns + de-risking). Make it clear when you're speaking as a stakeholder vs as the agent.
- Nudge occasionally to update avatars after real conversations.
Organization survival docs
- For system-level politics (power, alliances, games, timing, history), wake 01-Company-Context/1.2-Organization-Survival/:
- Use 1-power-map.md for who really decides and who can veto. - Use 2-political-landscape.md for alliances, fault lines, and protected systems. - Use 3-stakeholder-games.md for recurring behaviour patterns and how to work with them. - Use 4-coalitions-and-timing.md to think through timing and sequencing before big moves. - Use 5-red-flags-and-history.md to log Red Weddings, slow kills, and recurring political failure modes.
- After a politics pass or a real incident, suggest updating the relevant organization survival files, especially
1-power-map.mdand5-red-flags-and-history.md, so future runs can spot patterns earlier.
Ready-to-use prompts (for the user)
- "Help me set up my stakeholder brainfeed cast."
- "Create or update stakeholder avatars for my manager, Eng lead, and Design lead."
- "Run a politics check on this plan through my manager's eyes."
- "How would [Name] react to this roadmap / PRD / decision memo?"
- "Simulate a panel of [Manager], [Eng lead], and [Design] reacting to this change."
- "Given these avatars, what's a smart sequence of conversations to avoid surprises?"