[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6D6XRB1REYYY8CNHZ4ASH7E":3,"skill-same:01M6D6XRB1REYYY8CNHZ4ASH7E":71},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":23,"publishedAt":40,"linkOnly":41,"markdownContent":42,"markdownBody":43,"frontMatter":44,"refName":47,"commitSha":48,"skillDirectory":49,"forksCount":50,"repositoryHtmlUrl":51,"repositoryHomepageUrl":52,"ownerHtmlUrl":53,"ownerBlogUrl":54,"ownerDescription":55,"updatedAt":56,"install":57,"resources":70},"01M6D6XRB1REYYY8CNHZ4ASH7E","customer-discovery",null,"Orchestrator — detect pre-product vs product-exists mode, bootstrap ICP candidates, recommend customer-discovery frameworks, synthesize outputs into unified ICP research","GeorgeQLe\u002Fagentic-skills","GeorgeQLe","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F21134126?v=4","01M6D2JP418S0VGJGF8M93EC3H","George Le","01M6D2JPBAKYD82TA5FS6HKQ3D",1,"MIT","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Fpacks\u002Fbusiness-research\u002Fcodex\u002Fcustomer-discovery\u002Farchive\u002Fv1.10\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000021","资料检索与研究","SCENARIO","使用场景",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39],"customer","discovery","customer discovery","spec","research","validate","bootstrap","product","file","orchestrator","idea","frameworks","synthesize","visual","deep","icp","2026-11-10 10:12:30",false,"---\nname: customer-discovery\ndescription: Orchestrator — detect pre-product vs product-exists mode, bootstrap ICP candidates, recommend customer-discovery frameworks, synthesize outputs into unified ICP research\ntype: research\nversion: v1.10\nargument-hint: \"[optional: \\\"discovery\\\" | \\\"validate\\\" | \\\"--synthesize\\\" | concept\u002Fidea, spec file path]\"\ninvocation: orchestrator\ncontext_intake: deep\nvisual_tier: visual\n---\n\n## Pack Availability Guard\n\nBefore telling the user to run a skill from another project-local pack, check `.agents\u002Fproject.json.enabled_packs`. If the target pack is not enabled, recommend `npx skillpacks install \u003Cpack>` from the project shell, instead of the target skill. After install, tell Codex users to start a fresh Codex CLI session if the `$` skill list remains stale. Global skills are always valid. Skills from this same pack are valid because the current skill is already running from that pack.\n\n# Customer Discovery — Orchestrator\n\nInvoke as `$customer-discovery`.\n\nThis is an **orchestrator skill** using the parent router delegation pattern. It detects context, bootstraps ICP candidates, recommends applicable customer-discovery frameworks, and synthesizes their outputs into unified ICP research. Individual frameworks live as child skills under `frameworks\u002F`.\n\nAvailable frameworks:\n\n| Framework | Slug | Creator | Lens |\n|-----------|------|---------|------|\n| W3 Hypothesis | `w3-hypothesis` | Schwartzfarb | WHO\u002FWHAT\u002FWHY hypothesis generation + disproval |\n| JTBD Needs Analysis | `jtbd-needs` | Ulwick \u002F Christensen | Needs-based segmentation via jobs and outcomes |\n| Four Forces | `four-forces` | Moesta | Push\u002FPull\u002FAnxiety\u002FHabit switching moment analysis |\n| Five Rings of Buying Insight | `five-rings` | Revella | Decision psychology and buyer journey |\n| Seven Dimensions | `seven-dimensions` | Lincoln Murphy | Ready\u002FWilling\u002FAble\u002FSuccess\u002FAcquisition\u002FAscension\u002FAdvocacy scoring |\n| PMF Engine | `pmf-engine` | Vohra \u002F Supan | Empirical ICP from existing user data (Sean Ellis test + HXC) |\n\n## Report-First Approval Gate\n\nDefault to scope-first approval: before synthesized research, inspect only enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions in a `review` alignment page plus a concise conversation summary.\n\nDo not perform synthesized research, rank candidates, make recommendations, or write working packets or canonical deliverables until final compiled YAML approves the research scope. Minimal pre-approval discovery may identify available files, source categories, and open questions; label it as scope evidence, not findings.\n\nAfter approved research-scope YAML, perform the research and write only the non-canonical working packet defined in the staged workflow. Then update the `review` alignment page with findings and stop again for feedback-only YAML or final compiled YAML artifact approval before creating or updating canonical research, spec, or task files.\n\nDo not include downstream or cross-skill command recommendations while a scope, framework findings, or synthesis approval is pending. The approval request itself is the next action. Parent-loop continuation is not downstream routing: after framework completion, hand control back to `$customer-discovery`, never an execution-loop command or a path-shaped child framework invocation. Only emit downstream next-skill routing after the synthesized `icp.md` artifact has been approved and written.\n\n## Staged Research Workflow\n\nUse this staged workflow for synthesized research or report outputs that would create or update canonical research, spec, or task files.\n\n1. **Stage 1 - Scope discovery and approval.** Inspect only enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions. Build the `review` HTML alignment page before synthesized research. The page must render the proposed scope, available source categories, known context, assumptions\u002Fconfidence, proposed working-packet and canonical output paths, and research-scope approval gates. Stop for final compiled YAML approval of the research scope. Do not perform synthesized research, rank candidates, make recommendations, or write working packets, canonical research, spec, or task files in Stage 1.\n2. **Stage 2 - Research and artifact review.** Only after approved research-scope YAML with no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback, perform the synthesized research, run required source\u002Fcode checks, and write only a non-canonical working packet: flat mode uses `research\u002F_working\u002Fpreliminary-\u003Cskill>-research.md`; product-path mode uses `research\u002F{slug}\u002F_working\u002Fpreliminary-\u003Cskill>-research.md`. Replace `\u003Cskill>` with this skill's `name` value. Raw evidence or search logs may remain as supporting evidence where this skill already requires them, but synthesized deliverables stay in the working packet. Update the `review` HTML alignment page so it renders the complete working-packet substance as structured HTML review UI: purpose-built sections, tables, matrices, gates, cards, and tier-appropriate charts or diagrams that preserve every packet section, finding, caveat, and decision detail without summary loss. Raw Markdown packet text may appear only as a supplemental source view after the rendered review UI; do not make a `Full Preliminary Packet` or `Full Working Packet` raw Markdown dump, giant `\u003Cpre>\u003Ccode>` block, link-only view, or source-only view the primary review surface. Include the evidence matrix, assumptions\u002Fconfidence register, source coverage gaps, proposed canonical file changes, and artifact approval gates. Stop for either feedback-only YAML or final compiled YAML. Feedback-only YAML revises the working packet and page, then remains in Stage 2.\n3. **Stage 3 - Finalize approved artifacts.** Consume final compiled YAML for artifact approval only when it has no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Apply approved edits first, archive the working packet to `docs\u002Fhistory\u002Farchive\u002FYYYY-MM-DD\u002FHHMMSS\u002F\u003Coriginal-working-path>`, remove the active working packet, write the approved canonical artifacts to the unchanged output paths below, and convert the alignment page to `confirmed` with the approval record preserved. As part of confirming, reconcile each displayed gate decision against the final compiled YAML and the written canonical artifact, render any `other`\u002Ffreeform choice as the read-only decision and drop superseded options, and run the post-confirmation self-check from the alignment-page contract before handoff.\n\nCanonical output paths remain unchanged. Search logs and other supporting evidence remain allowed only where this skill's output contract already requires them.\n\n**How this maps onto the Research Session Loop.** This orchestrator runs as a self-advancing loop (see Execution Model below and `docs\u002Fresearch-session-loop-convention.md`): each approval stop above is a **session boundary** — the agent stops, the user reviews and compiles YAML, then starts a fresh Codex session and re-invokes `$customer-discovery` to continue. The multi-select framework page is the single scope-and-candidate approval gate (Stage 1) for the whole selected set; synthesis has its own `review` page. When the loop runs a framework subskill inline (state C), that multi-select approval already satisfies the framework's Stage-1 scope gate, so the framework **enters at its research stage (Stage 2)** and produces exactly one findings `review` page — one approval gate per framework. Standalone (non-loop) invocation of a framework subskill still runs its full two-stage workflow.\n\n## Evidence And Feedback Handling\n\nTreat user feedback as input to evaluate, not as automatic ground truth.\n\n- For factual, evidentiary, technical, or source-backed claims: verify against available evidence. If the user appears to misunderstand the evidence or states something factually incorrect, push back clearly and cite the evidence. Do not rewrite findings merely to agree.\n- For taste, brand, positioning preference, risk appetite, prioritization, or other subjective judgment calls: weigh user feedback heavily and adapt the recommendation unless it conflicts with verified evidence.\n- When feedback mixes facts and preference, separate them explicitly: correct the factual part, then incorporate the preference where it is a legitimate judgment call.\n- When uncertain, say what is known, what is inferred, and what would change the conclusion.\n\n## Prerequisites\n\n- **Hard**: None — this is the first research skill in the AFPS chain.\n- **Soft**: Read these if they exist:\n  - `research\u002Fidea-brief.md` (or `research\u002F{slug}\u002Fidea-brief.md`) — concept, hypotheses, ICP readiness notes\n  - Specs, README\u002FCLAUDE, and relevant source files for product context\n  - `research\u002Fcustomer-feedback.md` — real customer evidence (triggers product-exists mode)\n\n## Execution Model — Research Session Loop\n\nThis is a **self-advancing Pattern A research orchestrator** (see `docs\u002Fresearch-session-loop-convention.md`). Each invocation starts cold, resolves its state from **pasted YAML + filesystem**, runs **exactly one heavy phase**, emits the next gate, and stops. The user advances the loop by starting a fresh Codex session and re-invoking `$customer-discovery`. The user never invokes a framework subskill directly — the orchestrator follows each selected framework's subskill inline.\n\nWhen a framework is pending, the only user-facing continuation route is re-invoking `$customer-discovery` with the same product\u002Fresearch path argument when present, for example `$customer-discovery research\u002Fafps-tracker`. Never tell the user to run a path-shaped child framework command; the parent resolves the pending framework from the run manifest and filesystem.\n\n\nState lives in two places only:\n\n- **Run manifest** — `research\u002F_working\u002Fcustomer-discovery-run.yaml` (flat) or `research\u002F{slug}\u002F_working\u002Fcustomer-discovery-run.yaml` (product-path). Records the selected framework set and each framework's intermediate path. Written when the multi-select YAML is approved. Shape:\n\n  ```yaml\n  orchestrator: customer-discovery\n  slug: skills-showcase            # omit in flat mode\n  selected_frameworks:\n    - slug: w3-hypothesis\n      intermediate: research\u002Fskills-showcase\u002Fcustomer-discovery-w3-hypothesis.md\n    - slug: jtbd-needs\n      intermediate: research\u002Fskills-showcase\u002Fcustomer-discovery-jtbd-needs.md\n  ```\n\n- **Canonical-intermediate existence** — a selected framework is *done* when `research\u002Fcustomer-discovery-{framework}.md` (or `research\u002F{slug}\u002Fcustomer-discovery-{framework}.md`) exists, *pending* otherwise. `pending = selected − existing-intermediates`. The manifest stores selection only, not per-framework status.\n\n`research\u002F.progress.yaml` stays coarse — its `pipeline_stage` is a pointer, not per-framework status.\n\n### State resolution (resolve the first match; YAML first, then most-progressed A→F)\n\nOn each invocation, after Product-Path Scope Resolution (step 0), resolve state:\n\n| State | Detected when | Heavy phase this session | Emits \u002F stops with |\n|---|---|---|---|\n| **0 — pasted YAML** | a compiled alignment YAML is pasted | branch on `approval_status`: `ready-for-agent-review` → apply the approval for the gate it answers (light: write manifest and\u002For prior framework intermediate, archive consumed source), then fall through to the next pending state below; `not-approved` → amend the named page (refinement session) and stop | amended page, or proceeds ↓ |\n| **A — done** | canonical `research\u002Ficp.md` (or `research\u002F{slug}\u002Ficp.md`) exists | — | done; emit next-skill route (step 9) |\n| **B — synthesize** | run manifest exists, all selected intermediates exist, no canonical `icp.md` (also forced by `--synthesize`) | **synthesis** (step 6) | synthesis `review` page |\n| **C — run framework** | run manifest exists, ≥1 selected framework pending | **run next pending framework inline at its research stage** (step 5b) | that framework's findings `review` page |\n| **E — build selection** | preliminary interview handoff exists, no run manifest, no multi-select page | mode detect → candidate bootstrap → build multi-select page (steps 1–5a) | multi-select `review` page |\n| **F — interview** | nothing yet (no handoff, no manifest) | deep interview (`context_intake: deep`) | preliminary interview handoff, stop |\n\n**State F handoff.** When the deep interview completes, write the preliminary interview handoff to `research\u002F_working\u002Fpreliminary-customer-discovery-interview.md` (or `research\u002F{slug}\u002F_working\u002Fpreliminary-customer-discovery-interview.md`) and **stop** — do not bootstrap candidates or build the multi-select page in the interview session. The handoff is a complete context transfer for the next cold session (state E): provisional mode signal (pre-product vs product-exists, with evidence), context summary, recommended framework subset with rationale, and all user answers that shape candidate generation and framework selection.\n\n**Light vs heavy.** Recording the approved selection into the run manifest (state 0→C head), writing an already-reviewed framework intermediate, and archiving a consumed source are *light* — they fold into the head of the next heavy session and do not get their own round-trip. The heavy phase (interview, one framework's research, synthesis) is the only thing isolated per session.\n\n**Shortcuts.** `$customer-discovery discovery` and `$customer-discovery validate` short-circuit states E→C: they write a fixed framework set straight into the run manifest after the user approves the shortcut plan, then enter state C (steps 7–8).\n\n---\n\n## Process\n\n### 0. Product-Path Scope Resolution\n\nResolve research scope by product path before using code or app structure as a hint:\n\n1. If `$ARGUMENTS` names a non-archived `research\u002F{slug}\u002F` directory or a product-path ID whose `scope_path` points there, use that path. Treat `{slug}` as the product\u002Fapp name, not the ICP, audience, or segment label.\n2. If `$ARGUMENTS` names only `research\u002F_archive\u002F{slug}\u002F` or a manifest entry with `status: archived` or legacy `status: abandoned`, stop and warn that the path is archived; do not write or update scoped outputs there.\n3. Read `research\u002F.progress.yaml` when present. Normalize legacy `active_path` to `active_paths` on read and write back `active_paths` on manifest updates. Treat legacy `abandoned` as `archived`; exclude `archived`, `abandoned`, `deferred`, `revisit_candidate`, `promoted`, and any `scope_path` under `research\u002F_archive\u002F` from active target selection.\n4. If active product paths exist in the manifest, use those paths. If multiple active paths exist, ask which one to target unless this skill explicitly supports cross-path output.\n5. If no active manifest target exists, list non-archived product directories under `research\u002F`, excluding `research\u002F_archive\u002F` and dot directories. Auto-select only when exactly one exists; ask when multiple exist.\n6. If no product directories exist, use flat `research\u002F` single-product mode.\n7. Detect monorepo\u002Fapp\u002Fpackage structure only as a secondary hint.\n\nWhen product path `{slug}` is active, read and write research under `research\u002F{slug}\u002F`, specs under `specs\u002F{slug}\u002F`.\n\n0b. **Product-path manifest**: Read `research\u002F.progress.yaml` when present. Normalize `active_path` (singular legacy) to `active_paths` (plural list) when reading; treat legacy `abandoned` as `archived` and exclude archived\u002Fdeferred\u002Frevisit\u002Fpromoted paths plus `research\u002F_archive\u002F` scopes from active target selection.\n\n### 1. Mode Detection\n\nRuns in **state E** (the build-selection session, after reading the interview handoff). Detect **pre-product mode** (default) or **product-exists mode**:\n\n**Pre-product mode** activates when:\n- No production codebase with live users detected, AND\n- No `research\u002Fcustomer-feedback.md` with post-launch customer evidence, AND\n- `$ARGUMENTS` does not contain \"product\" or \"validate\"\n\nAvailable frameworks in pre-product mode:\n- `w3-hypothesis` (default) — Schwartzfarb WHO\u002FWHAT\u002FWHY hypothesis generation + disproval\n- `jtbd-needs` (default) — Ulwick\u002FChristensen needs-based segmentation via jobs and outcomes\n- `four-forces` (default) — Moesta Push\u002FPull\u002FAnxiety\u002FHabit switching moment analysis\n- `five-rings` (optional) — Revella decision psychology and buyer journey. Pre-check when idea brief indicates B2B model.\n- `seven-dimensions` (optional) — Lincoln Murphy ICP scoring with post-sale dimensions\n- `pmf-engine` (unavailable) — requires real user data; grayed out with explanation\n\n**Product-exists mode** activates when:\n- Production code or deployment exists, OR\n- `research\u002Fcustomer-feedback.md` exists with real customer data, OR\n- `$ARGUMENTS` contains \"product\" or \"validate\"\n\nAvailable frameworks in product-exists mode:\n- `pmf-engine` (default) — Vohra\u002FSupan empirical ICP from existing user data\n- `seven-dimensions` (default) — Lincoln Murphy structured scoring with post-sale dimensions\n- `four-forces` (optional) — switching analysis grounded in real churn\u002Fadoption data\n- `five-rings` (optional) — buying insight refresh with real buyer evidence\n- `jtbd-needs` (optional) — job validation against actual usage\n- `w3-hypothesis` (unavailable) — superseded by empirical data; grayed out with explanation\n\n### 2. Load Context\n\n- Read `$ARGUMENTS`: if file path, read it; if text, treat as concept; if empty, check for idea brief\n- Read `research\u002F{slug}\u002Fidea-brief.md` or `research\u002Fidea-brief.md` — treat as starting hypotheses, not settled truth\n- Read CLAUDE.md, README for product context\n- Read codebase (if it exists): package config, key source files, routes, data models\n- Read existing `research\u002Ficp.md` if it exists (prior run — ask user if they want to refine or start fresh)\n- Read any existing `research\u002Fcustomer-discovery-*.md` intermediate artifacts (from prior framework runs)\n\n**Provisional product-path evidence.** When a referenced product path is not present in `research\u002F.progress.yaml`, do not treat it as a canonical active path. Require an explicit provisional-path evidence reference before using it as source context. If no such evidence exists, ask the user whether to proceed or to run `$idea-scope-brief` first.\n\n### 3. Marketplace Side Preflight\n\nBefore identifying ICP candidates, resolve market-structure side coverage:\n\n- Read any `Market Structure Handoff` or marketplace\u002Fplatform\u002FB2B2C\u002Fmulti-sided notes from the idea brief, especially `## ICP Readiness`. Treat side names and value exchange as hypotheses, not validated ICPs.\n- If no idea brief exists, infer likely sides from `$ARGUMENTS`, README, specs, codebase context, or product description when the concept suggests a marketplace\u002Fplatform\u002FB2B2C\u002Fmulti-sided model.\n- During broad market research, validate or refute the classification with evidence.\n- Before candidate generation, write a side-coverage note in the working packet.\n- Candidate generation must cover each material side or explicitly explain why a side is excluded.\n\n### 4. Candidate Bootstrapping\n\nThis step runs in **state E** only (skipped when `research\u002Ficp.md` already exists and user chooses to refine).\n\n**Broad market research** using web search with **8–12 diverse query strategies** (direct persona, pain point, market segment, trend, competitor user, forum\u002Fcommunity, job posting, industry report, switching trigger, adjacent market, geographic\u002Fregulatory, business model, WTP signal, named account searches).\n\n**Classify the business model** into one or more of: B2B SaaS (PLG), B2B SaaS (SLG), B2C, B2C subscription, marketplace\u002Fplatform, B2B2C, D2C, open-source\u002Fopen-core, API\u002Fdeveloper-first.\n\n**Identify 2–5 ICP candidates.** For each, note: who they are, pain evidence, accessibility, value delivery potential, WTP signal strength.\n\nWrite candidates to `research\u002F_working\u002Fpreliminary-customer-discovery-research.md`.\n\n**Checkpoint — Present candidates to the user.** Show ICP candidates with evidence. Ask for corrections and missing segments. Incorporate feedback before framework selection.\n\n### 5a. State E — Build Framework Multi-Select Page\n\nBuild the framework multi-select `review` alignment page with mode explanation, ICP candidates summary, available evidence, multi-select framework section (with mode-appropriate defaults pre-checked), a loop explanation (the selected set is the scope-and-candidate approval gate; each selected framework is then run inline, one findings page per framework, and the run advances by re-invoking `$customer-discovery`), and the approval gate.\n\nThis multi-select approval **is** the Stage-1 scope approval for the whole selected set. Stop for compiled YAML. Do **not** write the run manifest or run any framework in this session — that is state C.\n\n### 5b. State C — Run Next Pending Framework (inline)\n\nThis session consumes the approved multi-select YAML (state 0→C) or advances after a prior framework's approval. At the **head** of the session, do the light bookkeeping first:\n\n1. **Write\u002Fappend the run manifest** if it does not yet exist: `research\u002F_working\u002Fcustomer-discovery-run.yaml` (flat) or `research\u002F{slug}\u002F_working\u002Fcustomer-discovery-run.yaml` (product-path), recording `selected_frameworks` with each framework's `slug` and canonical `intermediate` path. Include only frameworks the user selected.\n2. **Archive the preliminary interview handoff** at this selection-commit point (move it under `docs\u002Fhistory\u002Farchive\u002FYYYY-MM-DD\u002FHHMMSS\u002F\u003Coriginal-working-path>`). A rejected multi-select page must still rebuild from the handoff, so do not archive it earlier.\n3. **If a prior framework's reviewed content was just approved** by the pasted YAML, write its canonical intermediate `research\u002Fcustomer-discovery-{fw}.md` (or `research\u002F{slug}\u002Fcustomer-discovery-{fw}.md`) from the already-reviewed working packet, and archive that framework's working packet and superseded review page.\n\nThen run the **one heavy phase**: determine the next pending framework (first selected framework whose canonical intermediate does not yet exist), then **load and follow that framework subskill's `SKILL.md` inline, entering at its research stage (Stage 2)** — the multi-select approval already satisfied the framework's Stage-1 scope gate, so perform the research, write its working packet, and build a single findings `review` page. Stop for that framework's compiled YAML.\n\n**Advance the loop by self-re-invocation.** The confirmed-page handoff and the terminal message name `$customer-discovery` and tell the user to start a fresh Codex session and re-invoke, reporting progress as \"k of N frameworks complete\". Do not emit cross-skill routing here — that happens only after synthesis (step 9).\n\n### 6. State B — Synthesis (auto-detected; also `$customer-discovery --synthesize`)\n\nEnter synthesis when the run manifest exists, **all** selected framework intermediates exist, and no canonical `research\u002Ficp.md` yet exists. An explicit `$customer-discovery --synthesize` also forces this state. Read all intermediate framework outputs (`research\u002Fcustomer-discovery-{slug}.md`). At least one must exist.\n\n**Synthesis mapping** — how framework outputs merge into the canonical 9+1 section format:\n\n| Canonical Section | Primary Source | Secondary Sources |\n|---|---|---|\n| Customer Profile | w3-hypothesis (WHO), pmf-engine (HXC) | seven-dimensions (Ready\u002FWilling\u002FAble) |\n| User Profile(s) | w3-hypothesis (WHO detail), pmf-engine (daily users) | jtbd-needs (job performers) |\n| Trigger Events | four-forces (Push), five-rings (Priority Initiative) | jtbd-needs (unmet outcomes) |\n| Current State Journey | jtbd-needs (current job map) | four-forces (Habit) |\n| Pain Map | jtbd-needs (underserved outcomes) | four-forces (Push intensity) |\n| Current Alternatives | four-forces (Habit, current solution language) | five-rings (Decision Criteria) |\n| Market Sizing | Aggregated across frameworks | seven-dimensions (Acquisition Efficiency) |\n| Stated Value Drivers | w3-hypothesis (WHAT\u002FWHY), jtbd-needs (desired outcomes) | five-rings (Success Factors), pmf-engine |\n| Customer ↔ User Dynamics | seven-dimensions (post-sale dimensions) | five-rings (Buyer's Journey) |\n| Discovery & Evaluation | five-rings (all 5 rings) | four-forces (Active Looking) |\n\n**Cross-ICP handling**: Merge per-candidate evidence across frameworks, select primary ICP using seven-dimensions scoring (or value × accessibility matrix), populate Additional ICPs and Cross-ICP Analysis sections.\n\nBuild alignment page for synthesis approval. After approval: write `research\u002Ficp.md` and `research\u002Ficp-search-log.md`, then on this canonical write **archive the run manifest** (`customer-discovery-run.yaml`) and synthesis working packet under `docs\u002Fhistory\u002Farchive\u002FYYYY-MM-DD\u002FHHMMSS\u002F\u003Coriginal-working-path>`, **update `research\u002F.progress.yaml`** `pipeline_stage` to `customer-discovery`, and emit the downstream next-step routing (step 9).\n\n### 7. State C via Discovery Shortcut (`$customer-discovery discovery`)\n\nPre-product shortcut — short-circuits states E→C with a fixed framework set (`w3-hypothesis`, `jtbd-needs`, `four-forces`). Build a `review` alignment page for the shortcut plan and require final compiled YAML approval. After approval, write the fixed framework set straight into the run manifest (`research\u002F_working\u002Fcustomer-discovery-run.yaml` or the product-path equivalent), then enter **state C** and run framework 1 inline per step 5b. The loop advances by re-invoking `$customer-discovery`.\n\n### 8. State C via Validation Shortcut (`$customer-discovery validate`)\n\nProduct-exists shortcut — short-circuits states E→C with a fixed framework set (`pmf-engine`, `seven-dimensions`). Build a `review` alignment page for the shortcut plan and require final compiled YAML approval. After approval, write the fixed framework set straight into the run manifest, then enter **state C** and run framework 1 inline per step 5b. The loop advances by re-invoking `$customer-discovery`.\n\n### 9. Next Steps (after synthesis only)\n\nRecommend the first matching condition as **Recommended**, remaining as **Other options**:\n\n- ALWAYS: `$competitive-analysis` — Research competitors and market gaps for this ICP\n- IF no `specs\u002F`: `$competitive-analysis` — Map the competitive landscape\n- IF `specs\u002F` exist but no `research\u002Fjourney-map.md`: check pack availability for `customer-lifecycle`, recommend `$journey-map` if enabled, or `npx skillpacks install customer-lifecycle` from the project shell if not enabled\n- IF codebase exists: check pack availability for `business-ops`, recommend `$mvp-gap` if enabled, or `npx skillpacks install business-ops` from the project shell if not enabled\n- IF `research\u002Fcompetitive-analysis.md` exists: check pack availability for `product-design`, recommend `$brainstorm` if enabled, or `npx skillpacks install product-design` from the project shell if not enabled\n\n### 10. Downstream Impact Check\n\nAfter writing, check existing downstream documents for conflicts. Classify as None\u002FMinor\u002FMajor. For Major, recommend `$reconcile-research`.\n\n## Output\n\n### State E output: framework multi-select `review` page + working packet\n\nThe run manifest `research\u002F_working\u002Fcustomer-discovery-run.yaml` is written at the head of the first state-C session (on multi-select approval), not in state E.\n\n### State C output: per-framework findings `review` page + `research\u002Fcustomer-discovery-{framework}.md`\n\n### State B output (synthesis): `research\u002Ficp.md` (or `research\u002F{slug}\u002Ficp.md`)\n\nCanonical 9+1 section format preserved for downstream compatibility. See claude mirror for full template.\n\n### `research\u002Ficp-search-log.md`\n\nRaw research log with all queries, findings, evidence, and data gaps.\n\n### `research\u002F.progress.yaml`\n\nProduct-path manifest updated when secondary ICPs create parked or promotable paths.\n\n## Task Classification\n\n- Immediately actionable → `tasks\u002Ftodo.md`\n- Human-only external actions → `tasks\u002Fmanual-todo.md`\n- Condition-gated records → `tasks\u002Frecord-todo.md`\n- Cadence-based reviews → `tasks\u002Frecurring-todo.md`\n\n## Constraints\n\n- **Parent self-advances one phase per invocation** and follows the next pending framework's subskill inline (entering at its research stage). It bootstraps candidates, builds the multi-select selection, runs each selected framework inline, and synthesizes. The run manifest records the selected framework set; progress is the existence of canonical intermediates. The loop advances by re-invoking `$customer-discovery` (fresh Codex session between sessions).\n- **Synthesis requires at least one framework output.**\n- **Mode detection is evidence-based.**\n- **Stay in problem space.** No features, architecture, UI, or technical solutions.\n- **Evidence-based.** Every claim must trace back to research evidence.\n- **Primary ICP must use the canonical 9 top-level `##` sections** for downstream compatibility.\n- **Section 10 captures behavioural signals only.**\n- **WTP is evidence, not pricing strategy.** Route raw signals to `$monetization`.\n- **Do not overwrite existing `research\u002Ficp.md`** without asking.\n- **Minimum 8 web search queries** before identifying ICP candidates.\n- **Present before writing.**\n\n## Alignment Page\n\nWhen this skill produces durable deliverables (research, specs, plans, reports, prototypes, or any document output), build a full-depth HTML alignment page following `ALIGNMENT-PAGE.md` in this skill's directory. Output: `alignment\u002Fcustomer-discovery-{topic}.html`.\n\n## Default Shipping Contract\n\nFollow the shared shipping contract convention in CLAUDE.md.\n","## Pack Availability Guard\n\nBefore telling the user to run a skill from another project-local pack, check `.agents\u002Fproject.json.enabled_packs`. If the target pack is not enabled, recommend `npx skillpacks install \u003Cpack>` from the project shell, instead of the target skill. After install, tell Codex users to start a fresh Codex CLI session if the `$` skill list remains stale. Global skills are always valid. Skills from this same pack are valid because the current skill is already running from that pack.\n\n# Customer Discovery — Orchestrator\n\nInvoke as `$customer-discovery`.\n\nThis is an **orchestrator skill** using the parent router delegation pattern. It detects context, bootstraps ICP candidates, recommends applicable customer-discovery frameworks, and synthesizes their outputs into unified ICP research. Individual frameworks live as child skills under `frameworks\u002F`.\n\nAvailable frameworks:\n\n| Framework | Slug | Creator | Lens |\n|-----------|------|---------|------|\n| W3 Hypothesis | `w3-hypothesis` | Schwartzfarb | WHO\u002FWHAT\u002FWHY hypothesis generation + disproval |\n| JTBD Needs Analysis | `jtbd-needs` | Ulwick \u002F Christensen | Needs-based segmentation via jobs and outcomes |\n| Four Forces | `four-forces` | Moesta | Push\u002FPull\u002FAnxiety\u002FHabit switching moment analysis |\n| Five Rings of Buying Insight | `five-rings` | Revella | Decision psychology and buyer journey |\n| Seven Dimensions | `seven-dimensions` | Lincoln Murphy | Ready\u002FWilling\u002FAble\u002FSuccess\u002FAcquisition\u002FAscension\u002FAdvocacy scoring |\n| PMF Engine | `pmf-engine` | Vohra \u002F Supan | Empirical ICP from existing user data (Sean Ellis test + HXC) |\n\n## Report-First Approval Gate\n\nDefault to scope-first approval: before synthesized research, inspect only enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions in a `review` alignment page plus a concise conversation summary.\n\nDo not perform synthesized research, rank candidates, make recommendations, or write working packets or canonical deliverables until final compiled YAML approves the research scope. Minimal pre-approval discovery may identify available files, source categories, and open questions; label it as scope evidence, not findings.\n\nAfter approved research-scope YAML, perform the research and write only the non-canonical working packet defined in the staged workflow. Then update the `review` alignment page with findings and stop again for feedback-only YAML or final compiled YAML artifact approval before creating or updating canonical research, spec, or task files.\n\nDo not include downstream or cross-skill command recommendations while a scope, framework findings, or synthesis approval is pending. The approval request itself is the next action. Parent-loop continuation is not downstream routing: after framework completion, hand control back to `$customer-discovery`, never an execution-loop command or a path-shaped child framework invocation. Only emit downstream next-skill routing after the synthesized `icp.md` artifact has been approved and written.\n\n## Staged Research Workflow\n\nUse this staged workflow for synthesized research or report outputs that would create or update canonical research, spec, or task files.\n\n1. **Stage 1 - Scope discovery and approval.** Inspect only enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions. Build the `review` HTML alignment page before synthesized research. The page must render the proposed scope, available source categories, known context, assumptions\u002Fconfidence, proposed working-packet and canonical output paths, and research-scope approval gates. Stop for final compiled YAML approval of the research scope. Do not perform synthesized research, rank candidates, make recommendations, or write working packets, canonical research, spec, or task files in Stage 1.\n2. **Stage 2 - Research and artifact review.** Only after approved research-scope YAML with no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback, perform the synthesized research, run required source\u002Fcode checks, and write only a non-canonical working packet: flat mode uses `research\u002F_working\u002Fpreliminary-\u003Cskill>-research.md`; product-path mode uses `research\u002F{slug}\u002F_working\u002Fpreliminary-\u003Cskill>-research.md`. Replace `\u003Cskill>` with this skill's `name` value. Raw evidence or search logs may remain as supporting evidence where this skill already requires them, but synthesized deliverables stay in the working packet. Update the `review` HTML alignment page so it renders the complete working-packet substance as structured HTML review UI: purpose-built sections, tables, matrices, gates, cards, and tier-appropriate charts or diagrams that preserve every packet section, finding, caveat, and decision detail without summary loss. Raw Markdown packet text may appear only as a supplemental source view after the rendered review UI; do not make a `Full Preliminary Packet` or `Full Working Packet` raw Markdown dump, giant `\u003Cpre>\u003Ccode>` block, link-only view, or source-only view the primary review surface. Include the evidence matrix, assumptions\u002Fconfidence register, source coverage gaps, proposed canonical file changes, and artifact approval gates. Stop for either feedback-only YAML or final compiled YAML. Feedback-only YAML revises the working packet and page, then remains in Stage 2.\n3. **Stage 3 - Finalize approved artifacts.** Consume final compiled YAML for artifact approval only when it has no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Apply approved edits first, archive the working packet to `docs\u002Fhistory\u002Farchive\u002FYYYY-MM-DD\u002FHHMMSS\u002F\u003Coriginal-working-path>`, remove the active working packet, write the approved canonical artifacts to the unchanged output paths below, and convert the alignment page to `confirmed` with the approval record preserved. As part of confirming, reconcile each displayed gate decision against the final compiled YAML and the written canonical artifact, render any `other`\u002Ffreeform choice as the read-only decision and drop superseded options, and run the post-confirmation self-check from the alignment-page contract before handoff.\n\nCanonical output paths remain unchanged. Search logs and other supporting evidence remain allowed only where this skill's output contract already requires them.\n\n**How this maps onto the Research Session Loop.** This orchestrator runs as a self-advancing loop (see Execution Model below and `docs\u002Fresearch-session-loop-convention.md`): each approval stop above is a **session boundary** — the agent stops, the user reviews and compiles YAML, then starts a fresh Codex session and re-invokes `$customer-discovery` to continue. The multi-select framework page is the single scope-and-candidate approval gate (Stage 1) for the whole selected set; synthesis has its own `review` page. When the loop runs a framework subskill inline (state C), that multi-select approval already satisfies the framework's Stage-1 scope gate, so the framework **enters at its research stage (Stage 2)** and produces exactly one findings `review` page — one approval gate per framework. Standalone (non-loop) invocation of a framework subskill still runs its full two-stage workflow.\n\n## Evidence And Feedback Handling\n\nTreat user feedback as input to evaluate, not as automatic ground truth.\n\n- For factual, evidentiary, technical, or source-backed claims: verify against available evidence. If the user appears to misunderstand the evidence or states something factually incorrect, push back clearly and cite the evidence. Do not rewrite findings merely to agree.\n- For taste, brand, positioning preference, risk appetite, prioritization, or other subjective judgment calls: weigh user feedback heavily and adapt the recommendation unless it conflicts with verified evidence.\n- When feedback mixes facts and preference, separate them explicitly: correct the factual part, then incorporate the preference where it is a legitimate judgment call.\n- When uncertain, say what is known, what is inferred, and what would change the conclusion.\n\n## Prerequisites\n\n- **Hard**: None — this is the first research skill in the AFPS chain.\n- **Soft**: Read these if they exist:\n  - `research\u002Fidea-brief.md` (or `research\u002F{slug}\u002Fidea-brief.md`) — concept, hypotheses, ICP readiness notes\n  - Specs, README\u002FCLAUDE, and relevant source files for product context\n  - `research\u002Fcustomer-feedback.md` — real customer evidence (triggers product-exists mode)\n\n## Execution Model — Research Session Loop\n\nThis is a **self-advancing Pattern A research orchestrator** (see `docs\u002Fresearch-session-loop-convention.md`). Each invocation starts cold, resolves its state from **pasted YAML + filesystem**, runs **exactly one heavy phase**, emits the next gate, and stops. The user advances the loop by starting a fresh Codex session and re-invoking `$customer-discovery`. The user never invokes a framework subskill directly — the orchestrator follows each selected framework's subskill inline.\n\nWhen a framework is pending, the only user-facing continuation route is re-invoking `$customer-discovery` with the same product\u002Fresearch path argument when present, for example `$customer-discovery research\u002Fafps-tracker`. Never tell the user to run a path-shaped child framework command; the parent resolves the pending framework from the run manifest and filesystem.\n\n\nState lives in two places only:\n\n- **Run manifest** — `research\u002F_working\u002Fcustomer-discovery-run.yaml` (flat) or `research\u002F{slug}\u002F_working\u002Fcustomer-discovery-run.yaml` (product-path). Records the selected framework set and each framework's intermediate path. Written when the multi-select YAML is approved. Shape:\n\n  ```yaml\n  orchestrator: customer-discovery\n  slug: skills-showcase            # omit in flat mode\n  selected_frameworks:\n    - slug: w3-hypothesis\n      intermediate: research\u002Fskills-showcase\u002Fcustomer-discovery-w3-hypothesis.md\n    - slug: jtbd-needs\n      intermediate: research\u002Fskills-showcase\u002Fcustomer-discovery-jtbd-needs.md\n  ```\n\n- **Canonical-intermediate existence** — a selected framework is *done* when `research\u002Fcustomer-discovery-{framework}.md` (or `research\u002F{slug}\u002Fcustomer-discovery-{framework}.md`) exists, *pending* otherwise. `pending = selected − existing-intermediates`. The manifest stores selection only, not per-framework status.\n\n`research\u002F.progress.yaml` stays coarse — its `pipeline_stage` is a pointer, not per-framework status.\n\n### State resolution (resolve the first match; YAML first, then most-progressed A→F)\n\nOn each invocation, after Product-Path Scope Resolution (step 0), resolve state:\n\n| State | Detected when | Heavy phase this session | Emits \u002F stops with |\n|---|---|---|---|\n| **0 — pasted YAML** | a compiled alignment YAML is pasted | branch on `approval_status`: `ready-for-agent-review` → apply the approval for the gate it answers (light: write manifest and\u002For prior framework intermediate, archive consumed source), then fall through to the next pending state below; `not-approved` → amend the named page (refinement session) and stop | amended page, or proceeds ↓ |\n| **A — done** | canonical `research\u002Ficp.md` (or `research\u002F{slug}\u002Ficp.md`) exists | — | done; emit next-skill route (step 9) |\n| **B — synthesize** | run manifest exists, all selected intermediates exist, no canonical `icp.md` (also forced by `--synthesize`) | **synthesis** (step 6) | synthesis `review` page |\n| **C — run framework** | run manifest exists, ≥1 selected framework pending | **run next pending framework inline at its research stage** (step 5b) | that framework's findings `review` page |\n| **E — build selection** | preliminary interview handoff exists, no run manifest, no multi-select page | mode detect → candidate bootstrap → build multi-select page (steps 1–5a) | multi-select `review` page |\n| **F — interview** | nothing yet (no handoff, no manifest) | deep interview (`context_intake: deep`) | preliminary interview handoff, stop |\n\n**State F handoff.** When the deep interview completes, write the preliminary interview handoff to `research\u002F_working\u002Fpreliminary-customer-discovery-interview.md` (or `research\u002F{slug}\u002F_working\u002Fpreliminary-customer-discovery-interview.md`) and **stop** — do not bootstrap candidates or build the multi-select page in the interview session. The handoff is a complete context transfer for the next cold session (state E): provisional mode signal (pre-product vs product-exists, with evidence), context summary, recommended framework subset with rationale, and all user answers that shape candidate generation and framework selection.\n\n**Light vs heavy.** Recording the approved selection into the run manifest (state 0→C head), writing an already-reviewed framework intermediate, and archiving a consumed source are *light* — they fold into the head of the next heavy session and do not get their own round-trip. The heavy phase (interview, one framework's research, synthesis) is the only thing isolated per session.\n\n**Shortcuts.** `$customer-discovery discovery` and `$customer-discovery validate` short-circuit states E→C: they write a fixed framework set straight into the run manifest after the user approves the shortcut plan, then enter state C (steps 7–8).\n\n---\n\n## Process\n\n### 0. Product-Path Scope Resolution\n\nResolve research scope by product path before using code or app structure as a hint:\n\n1. If `$ARGUMENTS` names a non-archived `research\u002F{slug}\u002F` directory or a product-path ID whose `scope_path` points there, use that path. Treat `{slug}` as the product\u002Fapp name, not the ICP, audience, or segment label.\n2. If `$ARGUMENTS` names only `research\u002F_archive\u002F{slug}\u002F` or a manifest entry with `status: archived` or legacy `status: abandoned`, stop and warn that the path is archived; do not write or update scoped outputs there.\n3. Read `research\u002F.progress.yaml` when present. Normalize legacy `active_path` to `active_paths` on read and write back `active_paths` on manifest updates. Treat legacy `abandoned` as `archived`; exclude `archived`, `abandoned`, `deferred`, `revisit_candidate`, `promoted`, and any `scope_path` under `research\u002F_archive\u002F` from active target selection.\n4. If active product paths exist in the manifest, use those paths. If multiple active paths exist, ask which one to target unless this skill explicitly supports cross-path output.\n5. If no active manifest target exists, list non-archived product directories under `research\u002F`, excluding `research\u002F_archive\u002F` and dot directories. Auto-select only when exactly one exists; ask when multiple exist.\n6. If no product directories exist, use flat `research\u002F` single-product mode.\n7. Detect monorepo\u002Fapp\u002Fpackage structure only as a secondary hint.\n\nWhen product path `{slug}` is active, read and write research under `research\u002F{slug}\u002F`, specs under `specs\u002F{slug}\u002F`.\n\n0b. **Product-path manifest**: Read `research\u002F.progress.yaml` when present. Normalize `active_path` (singular legacy) to `active_paths` (plural list) when reading; treat legacy `abandoned` as `archived` and exclude archived\u002Fdeferred\u002Frevisit\u002Fpromoted paths plus `research\u002F_archive\u002F` scopes from active target selection.\n\n### 1. Mode Detection\n\nRuns in **state E** (the build-selection session, after reading the interview handoff). Detect **pre-product mode** (default) or **product-exists mode**:\n\n**Pre-product mode** activates when:\n- No production codebase with live users detected, AND\n- No `research\u002Fcustomer-feedback.md` with post-launch customer evidence, AND\n- `$ARGUMENTS` does not contain \"product\" or \"validate\"\n\nAvailable frameworks in pre-product mode:\n- `w3-hypothesis` (default) — Schwartzfarb WHO\u002FWHAT\u002FWHY hypothesis generation + disproval\n- `jtbd-needs` (default) — Ulwick\u002FChristensen needs-based segmentation via jobs and outcomes\n- `four-forces` (default) — Moesta Push\u002FPull\u002FAnxiety\u002FHabit switching moment analysis\n- `five-rings` (optional) — Revella decision psychology and buyer journey. Pre-check when idea brief indicates B2B model.\n- `seven-dimensions` (optional) — Lincoln Murphy ICP scoring with post-sale dimensions\n- `pmf-engine` (unavailable) — requires real user data; grayed out with explanation\n\n**Product-exists mode** activates when:\n- Production code or deployment exists, OR\n- `research\u002Fcustomer-feedback.md` exists with real customer data, OR\n- `$ARGUMENTS` contains \"product\" or \"validate\"\n\nAvailable frameworks in product-exists mode:\n- `pmf-engine` (default) — Vohra\u002FSupan empirical ICP from existing user data\n- `seven-dimensions` (default) — Lincoln Murphy structured scoring with post-sale dimensions\n- `four-forces` (optional) — switching analysis grounded in real churn\u002Fadoption data\n- `five-rings` (optional) — buying insight refresh with real buyer evidence\n- `jtbd-needs` (optional) — job validation against actual usage\n- `w3-hypothesis` (unavailable) — superseded by empirical data; grayed out with explanation\n\n### 2. Load Context\n\n- Read `$ARGUMENTS`: if file path, read it; if text, treat as concept; if empty, check for idea brief\n- Read `research\u002F{slug}\u002Fidea-brief.md` or `research\u002Fidea-brief.md` — treat as starting hypotheses, not settled truth\n- Read CLAUDE.md, README for product context\n- Read codebase (if it exists): package config, key source files, routes, data models\n- Read existing `research\u002Ficp.md` if it exists (prior run — ask user if they want to refine or start fresh)\n- Read any existing `research\u002Fcustomer-discovery-*.md` intermediate artifacts (from prior framework runs)\n\n**Provisional product-path evidence.** When a referenced product path is not present in `research\u002F.progress.yaml`, do not treat it as a canonical active path. Require an explicit provisional-path evidence reference before using it as source context. If no such evidence exists, ask the user whether to proceed or to run `$idea-scope-brief` first.\n\n### 3. Marketplace Side Preflight\n\nBefore identifying ICP candidates, resolve market-structure side coverage:\n\n- Read any `Market Structure Handoff` or marketplace\u002Fplatform\u002FB2B2C\u002Fmulti-sided notes from the idea brief, especially `## ICP Readiness`. Treat side names and value exchange as hypotheses, not validated ICPs.\n- If no idea brief exists, infer likely sides from `$ARGUMENTS`, README, specs, codebase context, or product description when the concept suggests a marketplace\u002Fplatform\u002FB2B2C\u002Fmulti-sided model.\n- During broad market research, validate or refute the classification with evidence.\n- Before candidate generation, write a side-coverage note in the working packet.\n- Candidate generation must cover each material side or explicitly explain why a side is excluded.\n\n### 4. Candidate Bootstrapping\n\nThis step runs in **state E** only (skipped when `research\u002Ficp.md` already exists and user chooses to refine).\n\n**Broad market research** using web search with **8–12 diverse query strategies** (direct persona, pain point, market segment, trend, competitor user, forum\u002Fcommunity, job posting, industry report, switching trigger, adjacent market, geographic\u002Fregulatory, business model, WTP signal, named account searches).\n\n**Classify the business model** into one or more of: B2B SaaS (PLG), B2B SaaS (SLG), B2C, B2C subscription, marketplace\u002Fplatform, B2B2C, D2C, open-source\u002Fopen-core, API\u002Fdeveloper-first.\n\n**Identify 2–5 ICP candidates.** For each, note: who they are, pain evidence, accessibility, value delivery potential, WTP signal strength.\n\nWrite candidates to `research\u002F_working\u002Fpreliminary-customer-discovery-research.md`.\n\n**Checkpoint — Present candidates to the user.** Show ICP candidates with evidence. Ask for corrections and missing segments. Incorporate feedback before framework selection.\n\n### 5a. State E — Build Framework Multi-Select Page\n\nBuild the framework multi-select `review` alignment page with mode explanation, ICP candidates summary, available evidence, multi-select framework section (with mode-appropriate defaults pre-checked), a loop explanation (the selected set is the scope-and-candidate approval gate; each selected framework is then run inline, one findings page per framework, and the run advances by re-invoking `$customer-discovery`), and the approval gate.\n\nThis multi-select approval **is** the Stage-1 scope approval for the whole selected set. Stop for compiled YAML. Do **not** write the run manifest or run any framework in this session — that is state C.\n\n### 5b. State C — Run Next Pending Framework (inline)\n\nThis session consumes the approved multi-select YAML (state 0→C) or advances after a prior framework's approval. At the **head** of the session, do the light bookkeeping first:\n\n1. **Write\u002Fappend the run manifest** if it does not yet exist: `research\u002F_working\u002Fcustomer-discovery-run.yaml` (flat) or `research\u002F{slug}\u002F_working\u002Fcustomer-discovery-run.yaml` (product-path), recording `selected_frameworks` with each framework's `slug` and canonical `intermediate` path. Include only frameworks the user selected.\n2. **Archive the preliminary interview handoff** at this selection-commit point (move it under `docs\u002Fhistory\u002Farchive\u002FYYYY-MM-DD\u002FHHMMSS\u002F\u003Coriginal-working-path>`). A rejected multi-select page must still rebuild from the handoff, so do not archive it earlier.\n3. **If a prior framework's reviewed content was just approved** by the pasted YAML, write its canonical intermediate `research\u002Fcustomer-discovery-{fw}.md` (or `research\u002F{slug}\u002Fcustomer-discovery-{fw}.md`) from the already-reviewed working packet, and archive that framework's working packet and superseded review page.\n\nThen run the **one heavy phase**: determine the next pending framework (first selected framework whose canonical intermediate does not yet exist), then **load and follow that framework subskill's `SKILL.md` inline, entering at its research stage (Stage 2)** — the multi-select approval already satisfied the framework's Stage-1 scope gate, so perform the research, write its working packet, and build a single findings `review` page. Stop for that framework's compiled YAML.\n\n**Advance the loop by self-re-invocation.** The confirmed-page handoff and the terminal message name `$customer-discovery` and tell the user to start a fresh Codex session and re-invoke, reporting progress as \"k of N frameworks complete\". Do not emit cross-skill routing here — that happens only after synthesis (step 9).\n\n### 6. State B — Synthesis (auto-detected; also `$customer-discovery --synthesize`)\n\nEnter synthesis when the run manifest exists, **all** selected framework intermediates exist, and no canonical `research\u002Ficp.md` yet exists. An explicit `$customer-discovery --synthesize` also forces this state. Read all intermediate framework outputs (`research\u002Fcustomer-discovery-{slug}.md`). At least one must exist.\n\n**Synthesis mapping** — how framework outputs merge into the canonical 9+1 section format:\n\n| Canonical Section | Primary Source | Secondary Sources |\n|---|---|---|\n| Customer Profile | w3-hypothesis (WHO), pmf-engine (HXC) | seven-dimensions (Ready\u002FWilling\u002FAble) |\n| User Profile(s) | w3-hypothesis (WHO detail), pmf-engine (daily users) | jtbd-needs (job performers) |\n| Trigger Events | four-forces (Push), five-rings (Priority Initiative) | jtbd-needs (unmet outcomes) |\n| Current State Journey | jtbd-needs (current job map) | four-forces (Habit) |\n| Pain Map | jtbd-needs (underserved outcomes) | four-forces (Push intensity) |\n| Current Alternatives | four-forces (Habit, current solution language) | five-rings (Decision Criteria) |\n| Market Sizing | Aggregated across frameworks | seven-dimensions (Acquisition Efficiency) |\n| Stated Value Drivers | w3-hypothesis (WHAT\u002FWHY), jtbd-needs (desired outcomes) | five-rings (Success Factors), pmf-engine |\n| Customer ↔ User Dynamics | seven-dimensions (post-sale dimensions) | five-rings (Buyer's Journey) |\n| Discovery & Evaluation | five-rings (all 5 rings) | four-forces (Active Looking) |\n\n**Cross-ICP handling**: Merge per-candidate evidence across frameworks, select primary ICP using seven-dimensions scoring (or value × accessibility matrix), populate Additional ICPs and Cross-ICP Analysis sections.\n\nBuild alignment page for synthesis approval. After approval: write `research\u002Ficp.md` and `research\u002Ficp-search-log.md`, then on this canonical write **archive the run manifest** (`customer-discovery-run.yaml`) and synthesis working packet under `docs\u002Fhistory\u002Farchive\u002FYYYY-MM-DD\u002FHHMMSS\u002F\u003Coriginal-working-path>`, **update `research\u002F.progress.yaml`** `pipeline_stage` to `customer-discovery`, and emit the downstream next-step routing (step 9).\n\n### 7. State C via Discovery Shortcut (`$customer-discovery discovery`)\n\nPre-product shortcut — short-circuits states E→C with a fixed framework set (`w3-hypothesis`, `jtbd-needs`, `four-forces`). Build a `review` alignment page for the shortcut plan and require final compiled YAML approval. After approval, write the fixed framework set straight into the run manifest (`research\u002F_working\u002Fcustomer-discovery-run.yaml` or the product-path equivalent), then enter **state C** and run framework 1 inline per step 5b. The loop advances by re-invoking `$customer-discovery`.\n\n### 8. State C via Validation Shortcut (`$customer-discovery validate`)\n\nProduct-exists shortcut — short-circuits states E→C with a fixed framework set (`pmf-engine`, `seven-dimensions`). Build a `review` alignment page for the shortcut plan and require final compiled YAML approval. After approval, write the fixed framework set straight into the run manifest, then enter **state C** and run framework 1 inline per step 5b. The loop advances by re-invoking `$customer-discovery`.\n\n### 9. Next Steps (after synthesis only)\n\nRecommend the first matching condition as **Recommended**, remaining as **Other options**:\n\n- ALWAYS: `$competitive-analysis` — Research competitors and market gaps for this ICP\n- IF no `specs\u002F`: `$competitive-analysis` — Map the competitive landscape\n- IF `specs\u002F` exist but no `research\u002Fjourney-map.md`: check pack availability for `customer-lifecycle`, recommend `$journey-map` if enabled, or `npx skillpacks install customer-lifecycle` from the project shell if not enabled\n- IF codebase exists: check pack availability for `business-ops`, recommend `$mvp-gap` if enabled, or `npx skillpacks install business-ops` from the project shell if not enabled\n- IF `research\u002Fcompetitive-analysis.md` exists: check pack availability for `product-design`, recommend `$brainstorm` if enabled, or `npx skillpacks install product-design` from the project shell if not enabled\n\n### 10. Downstream Impact Check\n\nAfter writing, check existing downstream documents for conflicts. Classify as None\u002FMinor\u002FMajor. For Major, recommend `$reconcile-research`.\n\n## Output\n\n### State E output: framework multi-select `review` page + working packet\n\nThe run manifest `research\u002F_working\u002Fcustomer-discovery-run.yaml` is written at the head of the first state-C session (on multi-select approval), not in state E.\n\n### State C output: per-framework findings `review` page + `research\u002Fcustomer-discovery-{framework}.md`\n\n### State B output (synthesis): `research\u002Ficp.md` (or `research\u002F{slug}\u002Ficp.md`)\n\nCanonical 9+1 section format preserved for downstream compatibility. See claude mirror for full template.\n\n### `research\u002Ficp-search-log.md`\n\nRaw research log with all queries, findings, evidence, and data gaps.\n\n### `research\u002F.progress.yaml`\n\nProduct-path manifest updated when secondary ICPs create parked or promotable paths.\n\n## Task Classification\n\n- Immediately actionable → `tasks\u002Ftodo.md`\n- Human-only external actions → `tasks\u002Fmanual-todo.md`\n- Condition-gated records → `tasks\u002Frecord-todo.md`\n- Cadence-based reviews → `tasks\u002Frecurring-todo.md`\n\n## Constraints\n\n- **Parent self-advances one phase per invocation** and follows the next pending framework's subskill inline (entering at its research stage). It bootstraps candidates, builds the multi-select selection, runs each selected framework inline, and synthesizes. The run manifest records the selected framework set; progress is the existence of canonical intermediates. The loop advances by re-invoking `$customer-discovery` (fresh Codex session between sessions).\n- **Synthesis requires at least one framework output.**\n- **Mode detection is evidence-based.**\n- **Stay in problem space.** No features, architecture, UI, or technical solutions.\n- **Evidence-based.** Every claim must trace back to research evidence.\n- **Primary ICP must use the canonical 9 top-level `##` sections** for downstream compatibility.\n- **Section 10 captures behavioural signals only.**\n- **WTP is evidence, not pricing strategy.** Route raw signals to `$monetization`.\n- **Do not overwrite existing `research\u002Ficp.md`** without asking.\n- **Minimum 8 web search queries** before identifying ICP candidates.\n- **Present before writing.**\n\n## Alignment Page\n\nWhen this skill produces durable deliverables (research, specs, plans, reports, prototypes, or any document output), build a full-depth HTML alignment page following `ALIGNMENT-PAGE.md` in this skill's directory. Output: `alignment\u002Fcustomer-discovery-{topic}.html`.\n\n## Default Shipping Contract\n\nFollow the shared shipping contract convention in CLAUDE.md.\n",{"name":5,"type":28,"version":45,"invocation":33,"description":7,"visual_tier":37,"argument-hint":46,"context_intake":38},"v1.10","[optional: \"discovery\" | \"validate\" | \"--synthesize\" | concept\u002Fidea, spec file path]","master","96e2912af886dd1606b1b97df707715aef062b02","packs\u002Fbusiness-research\u002Fcodex\u002Fcustomer-discovery\u002Farchive\u002Fv1.10",0,"https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills","https:\u002F\u002Fagentic-skills-blush.vercel.app","https:\u002F\u002Fgithub.com\u002FGeorgeQLe","georgeqle.com","George Le is a futurist and serial tech entrepreneur who wants to work toward a world where humanity is free from robotic tasks.","2026-11-10 04:07:24",{"source":8,"skillName":5,"commands":58,"prompt":68,"zipAvailable":69},[59,62,65],{"tool":60,"command":61},"npx","npx skills add GeorgeQLe\u002Fagentic-skills --skill customer-discovery",{"tool":63,"command":64},"bunx","bunx skills add GeorgeQLe\u002Fagentic-skills --skill customer-discovery",{"tool":66,"command":67},"pnpm","pnpm dlx skills add GeorgeQLe\u002Fagentic-skills --skill customer-discovery","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Fpacks\u002Fbusiness-research\u002Fcodex\u002Fcustomer-discovery\u002Farchive\u002Fv1.10\u002FSKILL.md",true,[],[72,85,95,107,120,131],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":40},"01M6G0NJES15RJKZE6YZ0TSYGF","source-checkout-allowed","Valid source-checkout-only fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fsource-checkout-allowed\u002FSKILL.md",[],[79,80,81,82,83,84],"source checkout allowed","checkout","fixture","only","v0","valid",{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":40},"01M6G0NFZ3R1MNB1E3DVTM0WB7","deck-install-valid","Valid deck install fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fdeck-install\u002FSKILL.md",[],[92,93,94,84,81,83],"deck install valid","deck","install",{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":101,"publishedAt":40},"01M6G0ND3QZQC4MDZ9NVB97PT2","codex-pack-guard-valid","Valid Codex Pack Availability Guard fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fcodex-pack-guard\u002FSKILL.md",[],[102,103,104,84,105,106,81,83],"codex pack guard valid","pack","guard","codex","availability",{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":116,"publishedAt":40},"01M6G0NA9YB3NKGZKD3S96JMNB","codex-missing-skill-valid","Valid Codex missing skill fallback fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fcodex-missing-skill-fallback\u002FSKILL.md",[113],{"categoryNo":114,"categoryName":115,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发",[117,118,84,105,83,119,81],"codex missing skill valid","missing","fallback",{"skillNo":121,"name":122,"nameZh":6,"description":123,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":124,"categories":125,"keywords":127,"publishedAt":40},"01M6G0N7PC0EKFD40ZWQAXSAQ9","codex-base-skill-fallback-valid","Valid Codex base skill fallback fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fcodex-base-skill-fallback\u002FSKILL.md",[126],{"categoryNo":114,"categoryName":115,"taxonomyType":21,"taxonomyTypeDesc":22},[84,105,119,128,129,130,83,81],"base skill","codex base skill fallback valid","base",{"skillNo":132,"name":133,"nameZh":6,"description":134,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":135,"categories":136,"keywords":137,"publishedAt":40},"01M6G0N4VRR2YKZQ6N3Y6S2BQA","claude-pack-guard-valid","Valid Claude Pack Availability Guard fixture.","https:\u002F\u002Fgithub.com\u002FGeorgeQLe\u002Fagentic-skills\u002Fblob\u002Fmaster\u002Ftests\u002Ffixtures\u002Fskill-install-routing\u002Fvalid\u002Fclaude-pack-guard\u002FSKILL.md",[],[84,104,103,138,106,83,81],"claude pack guard valid"]