A publishable Hermes skill for running council-style decision reviews as game-theoretic evaluations.
The skill combines structured council roles with game-theory checks: players, incentives, strategies, payoff dimensions, information asymmetries, counter-moves, dominated options, stability claims, mechanisms, stop conditions, and verification tests.
- a product, roadmap, governance, or architecture choice has interacting incentives;
- a delegation or multi-agent workflow could drift or fake success;
- a safety, authority, approval, or verification boundary matters;
- a council review needs more than role-flavored opinions;
- the recommendation should survive adversarial or locally rational behavior.
Do not use it for simple factual lookup or routine execution when the next step is obvious and reversible.
From a public raw URL:
hermes skills install https://raw.githubusercontent.com/MaverickKB/game-theoretic-council-skill/main/SKILL.md --name game-theoretic-councilHermes' CLI installer currently expects a registry identifier or a direct HTTP(S) SKILL.md URL. For local development, edit this checkout and install from the raw URL after pushing, or copy SKILL.md into your profile's skills directory under a folder named game-theoretic-council.
Load it directly in a Hermes session after installation:
/skill game-theoretic-council
A review shaped like:
# Game-Theoretic Council Review: <decision>
## 1. Decision
## 2. Players and incentives
## 3. Candidate strategies
## 4. Payoff matrix
## 5. Game type
## 6. Council findings
## 7. Dominated strategies
## 8. Equilibrium / stability claim
## 9. Recommendation
## 10. Mechanisms to implement
## 11. Verification testsSKILL.md— the Hermes skill.examples/compact-template.md— short review template.examples/full-template.md— full review template.scripts/validate.py— local sanity checks for publication hygiene.
python3 scripts/validate.pyThe validation script checks required skill metadata and scans public files for common private/local reference leaks.
MIT.