菜单
Context-aware story extraction. Loads the maker's full context — identity, expertise, voice, published work — then reads the narrative architecture to understand the story being told and what's missing. Interviews the maker to fill gaps with real moments. Modular: works with any project.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Evaluate the IP constellation — dispatches baseline checks, then 9 lenses in parallel (Millman, Bierut, Appleton, Shaw, Peers, Victore, The Walk, The Walls). Coordinator skill.
Cross-page coherence. Do the three tiers form a coherent argument when read as a sequence? Read-only diagnostic.
Baseline checks — mechanical health of the constellation. Links, images, cross-refs, attribution, voice violations, SEO, tooltips. Read-only diagnostic.
Where lenses agree and disagree. Maps consensus vs. contradiction across audit results. Read-only diagnostic.
Copy verification checklist — 13-item pass/fail check on draft or published copy. Voice rules, grounding, Shaw check, steward voice. Read-only diagnostic.
Would someone send this? Per-page forwardability — is the work remarkable enough to share? Read-only diagnostic.
| name | interview |
| description | Context-aware story extraction. Loads the maker's full context — identity, expertise, voice, published work — then reads the narrative architecture to understand the story being told and what's missing. Interviews the maker to fill gaps with real moments. Modular: works with any project. |
| user_invocable | true |
Extract real moments from a maker's lived experience to fill specific gaps in a narrative architecture. Before asking a single question, the skill loads full context — who the maker is, what they've built, how they talk, what the story needs — so every question is informed by the depth of the work and aimed at a specific gap in the narrative.
This is mandatory. The skill does not interview without completing this sequence first.
Before any interview begins, the skill loads context in this order:
The project's narrative architecture defines the story, its structure, and its gaps. The architecture file's Context Sources section points to everything else the skill needs to load.
The architecture file lists the sources that give the skill depth. Load ALL of them before proceeding. Typical sources:
| Source | What it gives the skill |
|---|---|
| Identity document | Who the maker is — aesthetic DNA, cultural roots, values, the genre. The skill interviews a PERSON, not a role. |
| Core insight / thesis | The governing ideas — what the work is actually about at the deepest level. Without this, the skill asks surface questions. |
| Voice samples | How the maker actually talks — rhythm, openings, vocabulary, humor. The skill recognizes the maker's real language when it hears it. |
| Published work | What's already been said publicly. The skill doesn't ask for material that already exists — it reads the work and builds from it. |
| Domains of expertise | The maker's full skill set, career history, the rooms they've been in. The skill needs to understand the RANGE to ask questions that surface connections across domains. |
| Conversation history | Prior sessions, ideation history. The skill knows what the maker has already articulated so it doesn't re-extract — it goes deeper. |
The exact files and paths are listed in the architecture file's Context Sources section. Different projects point to different sources. The skill reads whatever the manifest tells it to read.
After loading, the skill should be able to answer:
If the skill can't answer all of these, it hasn't loaded enough context. Go back to Step 2.
Now — and only now — the skill presents:
The maker chooses where to begin. The skill follows.
Every project that uses this skill needs an architecture file. The format:
# [Project Name] — Narrative Architecture
## Context Sources
- Identity: [path to identity/persona document]
- Core insight: [path to thesis/insight document]
- Voice: [path to voice sample accumulation]
- Published work: [paths to published pages/posts]
- Domains: [path to career/expertise summary, or inline]
- History: [path to conversation exports or session logs]
## The Story
[One paragraph: what does the finished work communicate?]
## The Governing Line
[The single sentence that anchors every piece]
## The Core Operation
[What the maker actually DOES — the skill or method underneath the work]
## The Maker's Domains
[List of domains with depth notes — not just names but what the maker
actually did in each domain and how long. This is what lets the skill
ask questions that surface cross-domain connections.]
## Voice
[Key voice principles — what language to use, what to avoid,
what imagery to reach for]
## Structure
### [Movement/Chapter/Section Name]
**Purpose:** [What this part teaches or communicates]
**Needs:** [What kind of material fills this part]
**Site pages / deliverables serving this movement:** [list]
#### [Specific gap]
- **Status:** OPEN | CAPTURED | DEEP
- **What's needed:** [Specific description]
- **Why this matters to the story:** [How this gap connects to the thesis]
- **Seed:** [When captured: room, moment, body, read, transfer, maker's words]
- **Placement:** [Which deliverable this feeds]
### [Next section]
...
## Cross-Structure Evidence
[How the same projects/stories appear across multiple movements —
the skill uses this to recognize when a tangent fills a gap in
a different movement than the one being discussed]
## Captured Material
[Running ledger of what's been extracted, with dates and references]
A good seed has a room you can see, a moment you can feel, and a read that only this maker would have made.
When the maker gives you a moment:
## Seed: [gap reference from architecture]
**The room:** [physical/environmental detail]
**The moment:** [what happened]
**The body:** [somatic detail — what the maker or the person on the receiving end physically felt]
**The read:** [what the maker saw that others missed]
**The transfer:** [connections to other domains/movements the maker surfaced]
**Maker's words:** "[direct quotes from the conversation]"
**Movement/Section:** [where this lives in the architecture]
**Status:** CAPTURED
/interviewLoad architecture, load all context sources, assess gaps, recommend where to start. The full startup sequence.
/interview [movement/section]Target a specific part of the architecture. Still loads full context, but focuses the gap assessment on the named section.
/interview batchRapid-fire seeding. Present each OPEN gap as a single line, capture one-sentence seeds. Fast. For when the maker has limited time or energy and you need maximum extraction per minute.
/copy for drafting/grip-test and /copy-verify for evaluation/knowledge finds material in conversation history. Interview elicits NEW material./voice-sample captures how the maker talks. Interview creates the occasion for talking./copy drafts from seeds. Interview produces the seeds./grip-test evaluates finished copy. Interview precedes writing./formwork evaluates across independent layers. The architecture file is a parallel structure — independent movements that register into a whole.The interview is the first move in the production chain. The architecture file is the map that keeps all of it coherent.