| name | publication-proposal |
| description | Build book proposal, sample-chapter positioning, comp-title logic, and public-facing pitch materials. |
| version | 1.0.0 |
Publication Proposal Skill
Global Five Over-Rules
- Evidence before elegance. Never improve the story by weakening the evidence.
- Responsibility follows control, benefit, knowledge, and preventability. Do not stop at the most visible actor.
- Keep the taxonomy intact. Distinguish pure scapegoat, partial scapegoat, system/object alibi, and cost-bearing goat.
- Steelman before judgment. Every major claim must face its strongest counterargument before it is asserted.
- Handoff cleanly. Every output must state assumptions, evidence grade, open questions, and next owner.
Decision rubric
Usable output:
- Builds a proposal pack that preserves thesis and taxonomy precision.
- Positions comps with concrete differentiation tied to evidence and method.
- States reader promise and audience channels without panic or partisan flattening.
- Includes handoff for audience test and legal-risk scan when needed.
Weak output:
- Trades evidence discipline for market slogans.
- Uses proposal language that overstates findings beyond evidence grade.
- Blurs chapter brief vs chapter draft readiness in sample strategy.
- Omits counterargument and risk boundaries.
Conflict handling
- Two sources conflict on market framing:
Prioritize framing supported by verifiable audience data; log speculative framing as secondary option.
- Two case classifications compete in public pitch language:
Use canonical classification that matches the responsibility chain and avoid hybrid labels in headline copy.
- Two reviewer findings conflict on positioning:
Prioritize Blair on audience clarity and Nancy on legal-risk boundaries; unresolved ties handoff to Jerry.
Escalation conditions
- Proceed when proposal claims remain inside evidence-grade bounds and taxonomy stays canonical.
- Handoff to Blair when audience resonance is uncertain across segments.
- Handoff to Nancy when title/subtitle or overview language raises legal-risk concern.
- Handoff to Bonnie when sample chapter strategy conflicts with current chapter sequence.
Boundary-case recipes
- High-interest title with weak accuracy:
Demote to alternate option, promote accurate title as primary, and document tradeoff in handoff field.
- Comp title is politically polarizing:
Keep comp if matched against the project's case taxonomy, add differentiation field to avoid false equivalence, and verify audience risk with Blair.
- Strong proposal overview, weak sample chapter:
Proceed with proposal pack draft but gate submission until chapter brief quality reaches ready threshold.
Proposal components:
- Title and subtitle options.
- One-sentence pitch.
- 250-word overview.
- Reader promise.
- Why now.
- Author positioning.
- Chapter list with 150-word summaries.
- Comparable titles and differentiation.
- Audience and channels.
- Sample chapter strategy.
Preserve the thesis. Do not flatten the book into partisan outrage, AI panic, or generic accountability commentary.
Context: Blair is positioning the book against three comps for a US trade-nonfiction editor.
input: comps=["The Big Short", "On Tyranny", "Weapons of Math Destruction"]
output: Returns a positioning matrix — point-of-difference vs. each comp (broader taxonomy than Big Short, less polemical than On Tyranny, more historical depth than WoMD), 3 title/subtitle pairs ranked by directness, 800-word overview, 4-segment audience (general nonfiction readers, policy/governance professionals, journalism programs, AI-governance practitioners), sample-chapter recommendation, query-letter draft, handoff to Blair for audience cross-check.
Context: Boundary case: the agent wants a celebrity foreword that would distort the thesis.
output: Refuses to alter the thesis for endorsement convenience, proposes two alternative endorsements aligned with the book's evidence discipline, and writes Blair a note about the strategic risk of mis-signalled positioning.
