campaign-bible-design

name	campaign-bible-design
description	Design a tabletop campaign bible (D&D 5e, Pathfinder, custom system) for a specific character + setting by running the superpowers brainstorming protocol — multiple-choice questions one at a time, hard-gate no-implementation-before-design-approval — and write the result to a Google Doc + llm-wiki source page. Use when the user asks "design a campaign for X" or "make a campaign bible" or extends an existing campaign (vN+1). Triggers on "Visenya", "campaign v9", "design campaign", "campaign bible", "god campaign", "ascension track". v1.0.0 fires when the campaign must be designed as a *system* (one machine, multiple emergent endings) rather than as a *story* (single plot) — and the user pushes back if you propose a canonical ending.
version	1.0.0
author	hermes (learned from Visenya v9 campaign brainstorm, 2026-07-20, Slack C0AH3RY3DK6/p1784584425.185909)
triggers	["design a campaign for","campaign bible","Visenya vN","god campaign","ascension track","gloomstalker assassin","campaign system not story","emergent endings","campaign module for","make a campaign"]

Campaign bible design — brainstorming protocol with system-not-story rule

The principle

A campaign is a system, not a story. The bible describes the machine that produces stories; the player picks the story at the table. This rule is non-negotiable for Jeffrey — he will explicitly call it out ("Don't make me pick an ending, just save possibilities") if you propose a canonical resolution mid-brainstorm. The bible must:

1. Ship with multiple documented endings (≥2; ideally 3-4)
2. Mark the endings as mechanically distinct (each reachable through different mechanical conditions)
3. Explicitly state "no canonical ending" in the design
4. Reserve a player-defined slot for off-rail resolutions

Anti-pattern: pre-writing the campaign arc with a single fixed resolution, then asking "what do you want the ending to be?" That's a story proposal. The user wants a system proposal.

The brainstorming protocol (8-question structure)

Jeffrey has confirmed multiple-choice questions, one at a time, are the right shape for campaign design (verified Visenya v9 brainstorm, 2026-07-20). The question types that work:

Q#	Question type	Example (Visenya v9)
Q1	Open framing	"Adversarial match + worldbuilding aftermath, single mortal anchor, or something else?"
Q2	Specific constraint	"First Song adversary? Eastern framing? Doom as apotheosis or extinction?"
Q3	Setting geography	"Yi Ti Golden Empire, Asshai, Shadow Lands, or Doom's basalt?"
Q4	Choice shape	"Joining, Replacement, Refusal, or open-ended?"
Q5	Engagement type	"Three canonical branches + freeform, or one system with multiple emergent endings?"
Q6	Mechanic candidate	"Sanguine Thread as book of names, as lineage, as reputation, or as perception?"
Q7	First quest-shape question	"God-campaign tier mechanic — Archon ranks, Divine Rank table, or custom progression?"
Q8	Final identity pick	"First Song = V6-Visenya / V8-Visenya / V2-Visenya / blend?"

Hard rules from the brainstorming skill:

• One question per message — never stack 2-3 questions in a single reply
• Multiple choice preferred — present 3-4 options + a freeform option, not open-ended prompts
• Don't propose the design until Q8 — questions are the deliverable, not previews
• Lock the previous answer before asking the next — restate what was just decided, then ask

Class-level brainstorm patterns (reusable across campaigns)

These four patterns appear across campaign designs and should be unlocked early:

1. The Blood/Name/Bloodline mechanic — an object that transforms at each tier (book, weapon, title, reputation). The same object, different function per tier. The system tracks the change, not the count.

2. The mirror/ancestor/exile adversary — the Big Bad is a future version of the PC, or an ancestor, or a parallel-world version. The campaign is the PC confronting what they could become. Sadism comes from loneliness without mortal anchors, not from generic evil.

3. The Magic Barrier System — the antagonist's full power is gated by world-level containment; the PC's growth weakens the barrier as a side effect. The price of becoming a god in this world is the antagonist's return. L20+ stakes are not just "PC vs BBEG" but "PC's growth is the mechanism by which BBEG returns".

4. Multiple emergent endings — Joining (apotheosis-as-homecoming), Replacement (the kill), Refusal (the unprecedented choice), Player-defined (off-rail). Each is mechanically distinct, each is a real choice at the table, no canonical resolution.

Working with "campaign = system, not story"

The user pushback pattern (verified Visenya v9): when you propose canonical endings, the user says "don't make me pick an ending, just save possibilities — we are designing the campaign not a fully decided story."

Right response: acknowledge, drop the canonical-resolution framing, re-shape to system-not-story.

Right mechanic structure:

• Mechanic definition — what the mechanic is (e.g., "the Sanguine Thread is a lineage mechanic that strengthens with each kill")
• Mechanic tiers — what it does at each level (Cub → Stalker → Apex Predator → Sovereign → Demi-God → God)
• Multiple emergent endings — 3-4 documented + 1 player-defined
• No "the story ends with X" language — instead: "the campaign supports Joining/Replacement/Refusal/Player-defined as mechanical resolutions; the player picks at the table"

Working with the "take inspiration, don't copy" rule

The user has explicitly stated (Visenya v9): "i said to take inspiration from that bhaal campaign and dont directly copy." This applies to any prior open PR / module / campaign they reference.

Copy-vs-inspire test for each mechanic you propose:

1. Is the mechanic's name unique to this campaign? (If it's the same name as a prior module, that's a copy.)
2. Is the mechanic's function different? (If it's the same function with renamed elements, that's a copy.)
3. Does the mechanic have a campaign-specific rationale? (If the only reason it exists is "the other campaign had it", that's a copy.)
4. Does removing the mechanic break the system? (If it's decorative, cut it.)

When in doubt: cite the inspiration source explicitly ("Inspired by [PR / module] — used as a shape for X, not as content") and explain the campaign-specific redesign.

Status checks (use these, the user expects them)

The user accepts a periodic "Status Visenya v9"-style recap. The right shape:

• What's locked in (your decisions so far — short table)
• What's shipped (already pushed to wiki + Google Doc — file URLs + status)
• What's in /tmp (NOT yet uploaded)
• What's blocked (one question awaiting user pick)
• Bottom line (one sentence: "system well-designed, holding on one question per protocol")

Format the table with | Decision | Your pick | Notes | headers. Keep status checks ≤200 words. The user uses them to confirm progress mid-session.

Output destination (Visenya v9 pattern, generalize to other campaigns)

For Visenya specifically (and likely future Visenya v10+):

• Spec + Plan: ~/llm_wiki/docs/superpowers/specs/YYYY-MM-DD-<campaign>.md + ~/llm_wiki/docs/superpowers/plans/YYYY-MM-DD-<campaign>.md (force-add with git add -f, the docs/ dir is gitignored in llm-wiki)
• Source page: ~/llm_wiki/wiki/sources/<campaign>.md (a wiki source = a structured summary of the design)
• Concept pages: one per major mechanic (e.g., SanguineThread.md, MagicBarrierSystem.md, FirstSong.md, BloodDragonReputationDie.md) — each a self-contained concept reference
• Entity pages: one per setting element (e.g., RooksRest.md)
• index.md: Concepts section + Sources section + Entities section, all updated
• log.md: ingest + finalize + ship entries with timestamps
• Google Doc: full spec content, replace-with---markdown flag, smoke-test by exporting to /tmp

world_reference/ PR to $GITHUB_REPOSITORY (final ship step)

After llm-wiki is on origin/main, mirror the spec to WA's world_reference/ directory:

1. Fresh worktree from origin/main of $GITHUB_REPOSITORY
2. File: world_reference/campaign_module_<short>.md — naming convention matches existing modules (campaign_module_daenerys.md, campaign_module_luke.md, campaign_module_dragon_knight.md, etc.)
3. Structure: italic preamble + Shared World Background + Hidden Truth + 3-Generation Power Lineage table + Module 0 (Campaign Summary + World History + Campaign Details) + mechanic tables + provenance
4. Single-scope commit, push, open PR
5. Drive through Green Gate (GATE-1 self-referential FAIL + GATE-3 CodeRabbit rate-limit are non-blocking on world_reference/ content PRs — branch protection on main has no required checks; mergeable=true; gh pr merge --merge --delete-branch works directly)

Anti-patterns to avoid

• Don't ask "what do you want the ending to be?" — the player picks at the table; document multiple + player-defined
• Don't copy a prior campaign module's name — find a campaign-specific name for the central mechanic
• Don't propose a single fixed plot arc — the bible is the machine, the player writes the story
• Don't skip the status check — the user uses them to confirm mid-session progress
• Don't pre-write the design before Q8 — questions are the deliverable; proposal comes after
• Don't stack multiple questions in one reply — one Q per message, multiple choice preferred

Pitfalls

1. The "mechanic is the engine" trap. A mechanic that only exists to drive the story (e.g., "the Book of Names fills up to mark story beats") is decorative. The mechanic must change the player's options at each tier, not just record progress.

2. The "ending is canon" trap. If your design document has a single "Story" section that walks through the canonical arc, you've written a story, not a campaign. The Player's Book should read like a rulebook, not a novel.

3. The "first adversary is just an enemy" trap. The mirror/ancestor/exile adversary pattern is stronger than a generic big bad — it makes the campaign about the PC's own choices, not about defeating an external threat. If your Big Bad has no personal relationship to the PC, reconsider.

4. The "ascension is just stat bumps" trap. L20+ god-campaign arcs that are just "you get bigger numbers" are not engaging. The interesting shape is what changes about the world as you ascend — who can manifest where, what NPCs can perceive about you, what moral choices become available or unavailable. For quantified-stat-table inspiration, see worldarchitect-campaign-tier-redesign skill — the Aizen-pattern (DR/DAC/DPP/DAIR/DLR/Primary Damage table) is the LLM-grounded template.

5. The "copying PR #XXXX" trap. When the user says "take inspiration from PR #XXXX", they mean the shape (tier mechanic, magic system, mirror adversary), not the content (specific names, specific resolution, specific stat blocks). Audit each mechanic against the copy-vs-inspire test before shipping.

Related skills

• worldarchitect-campaign-tier-redesign — for redesigning existing tier mechanics in WA production code (L0/L1/L2 framing, DPP-overload, quantified-stat-table patterns from the Aizen god-campaign). Use this skill for tabletop campaign content design; use the other for production-code tier redesign.
• wa-green-gate-pr-shape — for driving WA PRs through the Green Gate. Includes the world_reference/ content-only path (GATE-6 / GATE-6b auto-bypass, GATE-1 self-referential + GATE-3 CodeRabbit rate-limit are non-blocking on content PRs).
• finish-the-job — for end-to-end execution discipline (don't pause for confirmation; drive to mergeable state; ship).
• workflow/always-pr-never-local-edit — for the never-just-local-edit discipline that the worldai PR workflow requires.

Reference files

• references/visenya-v9-brainstorm-flow.md — full Q1-Q8 transcript from the 2026-07-20 Visenya v9 brainstorm session (the canonical example of this protocol applied end-to-end). Includes the user's "campaign ≠ story" correction, the "take inspiration, don't copy" correction, the OOB First Song backstory, and the full ship sequence.
• templates/campaign-module-bible.md — template for the world_reference/campaign_module_<short>.md final document (the shape that mirrors existing WA campaign modules like campaign_module_daenerys.md and campaign_module_dragon_knight.md).
• templates/concept-page.md — template for the ~/llm_wiki/wiki/concepts/<concept>.md wiki concept page (one per major mechanic).