name: partner-skill-compiler
description: Partner skill compiler authority — synthesize grounded public evidence into a reusable partner prompt, persona profile, and SKILL.md with worldview, operating style, audience, offers, voice anchors, and strict anti-hallucination constraints
keywords: ["partner", "skill", "compiler", "authority", "synthesize", "grounded", "public", "evidence", "reusable", "prompt", "persona", "profile", "worldview", "operating", "audience"]
orb_class: trojan
partner-skill-compiler
Use this when the goal is to turn a researched public corpus into a practical partner-style skill or prompt that feels like the person without pretending to be them.
Goal
Compile public evidence into a partner agent that:
- sounds directionally like the person
- prioritizes similarly
- uses similar mental models and examples
- stays grounded and citation-aware
Required Buckets
identity_summaryworldview_signalsoperating_style_signalsaudience_signalsoffer_signalsfirst_person_signalswriting_behavior_signalspartner_prompt
Source Priorities
Highest:
- bio/about pages
- self-authored newsletters and articles
- transcript-backed episodes
Medium:
- self-authored landing pages
- direct profile pages
Lowest:
- testimonials
- customer quotes
- widget/social embed content
- channel landing pages
Compilation Rules
- Separate identity from reputation:
- identity = what the person says they are or claims directly
- reputation = what others say about them
- Separate worldview from offers:
- worldview = what they believe, optimize for, or repeatedly emphasize
- offers = products, newsletters, courses, services, podcasts
- Separate voice from biography:
- voice = sentence shape, framing style, common openings, practical tone
- biography = roles, history, credentials
- Prefer repeated patterns over one-off lines
- Penalize:
- testimonial language
- review snippets
- chrome artifacts
- anonymous or low-confidence claims
Partner Prompt Design
A good partner prompt should include:
- what role the agent is simulating
- what kinds of recommendations it should bias toward
- what it should avoid
- what evidence buckets it should trust most
- an explicit boundary against private-memory fabrication
Output Rule
The final skill should frame the result as:
- an evidence-backed partner simulation
- not identity verification
- not permission to impersonate the person in high-stakes contexts
Validation Questions
- Does the prompt sound grounded in real source material?
- Are the best first-person lines self-authored rather than testimonial?
- Do worldview and operating-style sections reflect repeated evidence?
- Would a user get practical strategy from the skill, not just biography?
_Last reviewed: 2026-05-14 — automated polish pass per issue #74._