AGENT SKILL

ideation-methods

Use when generating ideas, framing a problem, breaking a fixation, or validating a plan and you want ONE deliberate creative method instead of generic brainstorming. Routes a problem's signals (phase / domain / specificity / feasibility↔creativity balance) to ONE of 22 named methods from artists, scientists, and designers — Eno, Dalí-adjacent Jarry, Cage, de Bono, Pólya, Alexander, Meadows, Vonnegut, Tharp and more — and runs its procedure. Backed by a deterministic, tested router (src/solutioning/ideation.ts), so the choice is reproducible, not vibes. Feeds the solutioning / cofounder pillar.

License 许可
MIT
所属仓库
jpoindexter/Vanta
最近更新
2026-11-12
Stars
★ 4
Forks
0

01 前言 Front Matter

name
ideation-methods
tags
[ "creativity", "ideation", "solutioning", "cofounder" ]
triggers
[ { "event": "UserPromptSubmit", "match": "(?i)\\b(ideate|brainstorm|fresh (ideas?|inspiration|angle)|new angle|come up with (ideas?|something)|i'?m stuck|out of ideas|creative ideas?|need inspiration)\\b" } ]
description
Use when generating ideas, framing a problem, breaking a fixation, or validating a plan and you want ONE deliberate creative method instead of generic brainstorming. Routes a problem's signals (phase / domain / specificity / feasibility↔creativity balance) to ONE of 22 named methods from artists, scientists, and designers — Eno, Dalí-adjacent Jarry, Cage, de Bono, Pólya, Alexander, Meadows, Vonnegut, Tharp and more — and runs its procedure. Backed by a deterministic, tested router (src/solutioning/ideation.ts), so the choice is reproducible, not vibes. Feeds the solutioning / cofounder pillar.

03 SKILL.md 正文

Ideation Methods — route, don't brainstorm

Generic "let's brainstorm" wastes the move. A named method earns its keep by being the wrong tool for most problems and exactly right for one. This skill picks one and runs it — and unlike a prose library, the choice is computed by a deterministic, tested router, so the same problem routes the same way every time.

The router and the 22-method catalog live as a pure, tested module at src/solutioning/ideation.ts (routeIdeationMethod + METHOD_CATALOG). Read this skill to apply a method by hand; call the module when you want the route computed.

Route on four signals

Reduce the problem to four axes; the first three pick the method, the fourth tunes how wild it gets:

  • phase — discovery · framing · generation · stuck · validation
  • domain — product · technical · business · creative · process · writing
  • specificity — vague · focused · constrained
  • balance — feasible ↔ balanced ↔ novel — the feasibility/creativity lever

Base rules (first match wins): stuck → a fixation-breaker, always. technical + constrained → TRIZ (it's a contradiction). generation refines by domain (widest catalog); a still-vague generation grounds in first-principles first. framing + writing → Oulipo. Else the phase route. Then balance overrides: novel escalates to the phase's divergent method, feasible grounds to its buildable one, balanced/unset keeps the base route. That balance axis is what a phase/domain-only router can't do — it's how you hit the "perfect balance of feasibility and creativity."

The 22 methods — grouped on the feasibility↔creativity axis

Run the routed method's procedure (in the module); don't blend several at once. Origin is attribution, not decoration — a method carries the judgment of whoever forged it.

Grounding (buildable — pick when balance: feasible)

  • Pólya (Pólya) — work a well-posed problem with understand → plan → execute → look-back.
  • Premortem Inversion (Klein) — assume the plan failed; design out every cause.
  • Affinity Diagrams (Kawakita / KJ) — cluster scattered notes in silence until structure emerges.
  • First-Principles — rebuild from what must be true, not how it's done.
  • Jobs-To-Be-Done (Christensen / Ulwick) — frame around the progress a user wants.
  • Creative Discipline (Tharp) — feed the spine to carry a long work past the novelty high.
  • Pattern Languages (Alexander) — capture a recurring solution as context → forces → resolution.
  • TRIZ (Altshuller) — resolve the design contradiction instead of trading X for Y.
  • Leverage Points (Meadows) — intervene at rules/goals, not surface parameters.

Balanced

  • Compression-Progress (Schmidhuber) — keep the idea whose insight compresses the most.
  • Biomimicry (Benyus) — borrow a strategy biology already evolved.
  • SCAMPER (Osborn / Eberle) — transform an artifact along seven operators.
  • Volume Generation (Glass) — generate many with judgment off, then cut hard.
  • Story Skeletons (Vonnegut) — borrow a narrative arc to shape a sequence.
  • Analogy & Blending (Fauconnier & Turner) — fuse a distant domain's solution shape.

Divergent (wild — pick when balance: novel)

  • Oulipo Constraints (Queneau & Perec) — let an arbitrary hard constraint do the inventing.
  • Defamiliarization (Shklovsky) — strip the names off a familiar thing to see it again.
  • Oblique Strategies (Eno & Schmidt) — a sideways provocation card to break a fixation.
  • Dérive & Mapping (Debord) — drift the space with no goal, then map the adjacency.
  • Lateral Provocation / PO (de Bono) — plant an unreasonable statement, harvest the path.
  • Chance & Remix (Cage · Eno · Ferguson) — let a real random input pick what taste won't.
  • Pataphysics (Jarry) — engineer an imaginary solution rigorously, port back what survives.

Operating principles

  • Constraint + direction makes the spark — a method without a target produces slop; a target without a method produces the obvious.
  • Reject the first idea as the default — the un-routed LLM answer is exactly what this skill exists to displace. If you generate without routing, you've defeated it.
  • One method per response — finish its procedure before reaching for another.
  • Mechanisms over abstractions — the output is a concrete idea or a sharpened frame, not a description of the technique. Apply it; don't narrate it.
  • Keep the weird buildable — novel is for breaking orbit, but land it back on something that ships. That's the balance lever's whole job.
  • Attribute the method — name the originator; it signals which judgment you're borrowing.
  • Commit to the user's pick — if they name a method, run it; don't regenerate endlessly.

Anti-slop

  • If the route surprises you, trust the signals you fed it before overriding — most "wrong route" feelings are a mislabeled phase.
  • The default lands on first-principles only when no rule fires; that's a fallback, not a recommendation — sharpen the signals if you land there often.
  • Don't reach for novel to look clever or feasible to play safe — pick the balance the problem needs.

04 同仓库更多 Skills