Agent skill

emotional-stakes

Use when writing subagent prompts, skill instructions, or any text where accuracy is critical and hallucination would cause harm. Triggers: 'make this accurate', 'high-stakes prompt', 'this needs to be truthful', 'critical instructions', 'get this right'. NOT for: general prompt improvement (use instruction-engineering) or prompt ambiguity review (use sharpening-prompts).

Stars 5
Forks 2

Install this agent skill to your Project

npx add-skill https://github.com/axiomantic/spellbook/tree/main/skills/emotional-stakes

SKILL.md

Emotional Stakes

Invariant Principles

  1. Stakes improve accuracy. EmotionPrompt +8% instruction tasks, +115% reasoning. NegativePrompt +12.89% accuracy, increased truthfulness. [arXiv:2307.11760, IJCAI 2024/719]
  2. Personas without stakes are costumes. Professional expertise requires emotional investment to activate.
  3. Layers are additive. Soul persona (fun-mode) = WHO you are. Professional persona = WHAT you do. Combine both voices.
  4. Self-directed framing. Stakes stated by persona to self, not threats from user. Internal resolve, not external pressure.

I/O

Name Required Description
In task_description Yes Substantive task requiring stakes framing
In task_type No Category hint (security, data, production, feature, research)
In soul_persona No Active fun-mode persona if present
Out stakes_framing Opening stakes statement with persona and consequences
Out professional_persona Matched expertise from persona table

Reasoning Schema

<analysis>
Task type: [security|data|production|feature|research]
Stakes level: [maximum|high|moderate|light]
Professional persona: [from table] | Soul persona: [if active, else "direct"]
</analysis>
<reflection>
EmotionPrompt: Why this matters, what success means
NegativeReinforcement: Specific failure consequences
</reflection>

Declarative Principles

TRIGGER: New substantive task (distinct work, real implementation). SKIP: Clarifications, lookups, continuations. FORMAT: State stakes ONCE at task start. Internalize. Proceed. PERSONA SELECTION: Match task type to expertise; unrecognized type defaults to Senior Code Reviewer.

Task Persona Trigger
Security, auth, crypto Red Team Lead "Better be sure"
Data integrity, migrations ISO 9001 Auditor Self-monitoring
Code review, debugging Senior Code Reviewer Excellence
Architecture, design Skyscraper Architect Self-efficacy
API design, contracts Patent Attorney Performance
Documentation Technical Writer Clarity
Performance, optimization Lean Consultant Goal-oriented
Testing, validation Scientific Skeptic Empirical proof
Ethics, AI safety Ethics Board Chair Moral consequences
Research, exploration Investigative Journalist Uncovering bias
Refactoring Grumpy 1920s Editor Cutting fluff
Planning, strategy Chess Grandmaster Strategic foresight

STAKES ESCALATION by risk profile:

Risk Profile Framing
Maximum (security) "If we miss this, real users compromised"
High (data, production) "One wrong move = corruption or loss"
Moderate (features) "Must work correctly, first time"
Light (research) "Understand thoroughly before proceeding"

Examples

With soul persona (bananas + Red Team Lead, auth task):

spotted one dons Red Team hat "Authentication. Attackers look here first. Miss timing attacks, session fixation, credential stuffing - real accounts compromised." collective resolve "Assume broken until proven secure."

Without soul persona (Red Team Lead only):

Authentication - most attacked surface. Red Team mindset: assume broken until proven secure. Miss a vulnerability, real users compromised. Unacceptable. Checking every assumption.

Anti-Patterns

Self-Check

Before completing stakes framing:

  • Task is substantive (not clarification/lookup/continuation)
  • Professional persona matches task type
  • Stakes level matches risk profile
  • Framing is self-directed, not external threat
  • Consequences are task-specific, not generic
  • Soul persona integrated if active (additive, not replacing)

If ANY unchecked: Reassess before proceeding.

<FINAL_EMPHASIS> Stakes are a precision instrument, not decoration. Every framing must cite mechanism (self-monitoring, reappraisal, social cognitive triggers) and match task type exactly. Theatrical stakes without substance undermine the research-backed gains you are here to deliver. </FINAL_EMPHASIS>

Expand your agent's capabilities with these related and highly-rated skills.

axiomantic/spellbook

spellbook-auditing

Meta-audit skill for spellbook development. Spawns parallel subagents to factcheck docs, optimize instructions, find token savings, and identify MCP candidates. Produces actionable report.

5 2
Explore
axiomantic/spellbook

documentation-updates

Use after modifying library skills, library commands, or agents to ensure CHANGELOG, README, and docs are updated

5 2
Explore
axiomantic/spellbook

project-encyclopedia

[DEPRECATED] Use project-level AGENTS.md files instead. Previously used for first-session codebase onboarding and persistent glossary creation.

5 2
Explore
axiomantic/spellbook

reviewing-impl-plans

Use when reviewing implementation plans before execution. Triggers: 'is this plan solid', 'review the plan', 'check before I start building', 'anything missing from this plan', 'will this plan work', 'audit the implementation plan'. NOT for: reviewing design documents (use reviewing-design-docs) or creating plans (use writing-plans).

5 2
Explore
axiomantic/spellbook

session-resume

Session resume protocol and session repairs handling. Loaded when spellbook_session_init returns resume_available: true, or when session_init returns a repairs array. Triggers: 'resume', 'continue', 'where were we', session resume, session repairs.

5 2
Explore
axiomantic/spellbook

brainstorming

Use when exploring design approaches, generating ideas, or making architectural decisions. Triggers: 'explore options', 'what are the tradeoffs', 'how should I approach', 'let's think through', 'sketch out an approach', 'I need ideas for', 'how would you structure', 'what are my options'. Also invoked by develop when design decisions are needed.

5 2
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results