dsh-idea

English | 中文

dsh-idea 是一个面向外部机会、市场变化和真实需求发现的 DeepSeek Harness 插件包。

dsh-idea is a DeepSeek Harness bundle for evidence-backed external opportunity discovery and product discovery.

Public six-plugin collaboration contract: SUITE.md.

DSH compatibility and installation

Closed-loop handoff guide: owned receipts, evidence gates and feedback. Start cross-plugin work with idea_handoff_receive; an accepted receipt is not approval or completion.

Tested against DeepSeek Harness 0.1.5-rc.2, the npm latest channel checked on 2026-09-22, with Cordis 4.0.2. Node.js must satisfy ^22.19.0 || >=24.0.0. DSH peer packages are pinned to the tested version; other release channels need a fresh compatibility check.

npm install -g @deepseek-ai/dsh@0.1.5-rc.2
dsh --version
dsh plugin --profile web add github:winyh/dsh-idea
dsh --profile web --dump-config
dsh web

Install into the profile you actually launch: web for dsh web, or replace it consistently with your custom profile name. Installing into default does not enable the plugin in web. Restart the running profile after an update.

A GitHub source installation uses prepare to build the entry point. If pnpm blocks it, add the exact package key printed by pnpm to that profile's pnpm-workspace.yaml under allowBuilds and repeat the installation after reviewing the source. Pin a reviewed Git commit for reproducibility. A built tarball from pnpm pack can instead be installed with dsh plugin --profile web add ./package.tgz.

The profile supplies tools and fs. It also supplies web and its search/fetch providers. The standard DSH base includes them; custom profiles must configure them explicitly. This plugin shares the profile's provider lifecycle, proxy and transport limits. Private docs/ files remain excluded from Git and package contents.

For maintainers, run pnpm install --frozen-lockfile and pnpm run verify. Verification includes type checking, lint, unit tests, build, package checks and plugin:runtime:validate: a keyless Cordis Loader test of the built plugin with real DSH services, model-visible tools, file reads, argument/output validation, cancellation and unload cleanup.

协作可靠

idea_artifact_review 在机会交给产品前校验工件版本、稳定 ID、内容指纹和有效期；它不会把机会草案直接升级为产品需求。

idea_artifact_review validates version, stable ID, content fingerprint and freshness before an opportunity is handed to product; it never turns an opportunity draft directly into a product requirement.

It turns “find an idea” into a repeatable operating loop:

signal → problem → opportunity → candidate idea → smallest useful test

The package is external-first. It can scan user-selected public pages for bounded source snapshots, then normalize optional Markdown, CSV, TSV, JSON and JSONL research records, filter and rank pain-point signals, cluster them into opportunity themes, draft multiple solution directions and generate a falsifiable experiment plan.

Plugin Positioning and Collaboration Navigation

dsh-idea is the opportunity-discovery layer in the six-plugin system. It turns external signals into testable opportunities and does not treat unvalidated ideas as product requirements.

• Owns: External market signals, real pain points, target users and contexts, opportunity themes, candidate directions and smallest falsifiable experiments.
• Inputs: User-selected public pages, community/competitor/review evidence and locally imported research records.
• Outputs: idea_review, Opportunity Solution Trees, candidate directions, interview guides and threshold-based experiment plans for dsh-product, with commercial viability reviewed by dsh-business.
• Does not own: Product delivery, commercial pricing, sales closing, growth analytics or SEO/GEO content execution. Popularity is not demand proof.

Positioning Architecture: Commercial Strategy Layer + Four-Stage Core Flow

The six plugins work together to turn a real demand signal into a deliverable product, reach target customers through marketing, and use monetization results to drive product iteration or discover new opportunities.

flowchart TB
    S["dsh-business<br/>Commercial strategy layer<br/>Value · model · pricing · profit"]

    subgraph MAIN["Four-stage core flow"]
        direction LR
        A["1. Demand<br/>dsh-idea"] --> B["2. Product<br/>dsh-product"]
        B --> C["3. Marketing<br/>dsh-geo + dsh-growth"]
        C --> D["4. Monetization execution<br/>dsh-sales"]
    end

    S -. "Sets commercial direction and guardrails" .-> A
    D --> R["Feedback<br/>Deals · renewals · revenue · cost"]
    R -->|Product iteration| B
    R -->|New discovery| A

    classDef strategy fill:#FFF4D6,stroke:#B7791F,color:#5C4500
    classDef stage fill:#E8F1FF,stroke:#3366CC,color:#173A7A
    classDef feedback fill:#E8F7EE,stroke:#2F855A,color:#1C4532
    class S strategy
    class A,B,C,D stage
    class R feedback

This plugin owns the demand stage: establish who has what pain in which context, how they solve it today and why it matters now; then hand the opportunity to dsh-product while dsh-business reviews commercial value. New problems, unmet needs and user contexts found during monetization return here for further research.

Plugin Navigation

Plugin	Clear responsibility	Direct link
dsh-idea	External opportunities, demand signals, candidate directions and smallest useful tests (this plugin)	README
dsh-product	Product definition, POC/MVP, release gates and PMF	README
dsh-business	Cross-cutting commercial strategy, value, pricing and profitability	README
dsh-sales	Monetization execution: qualification, deal progression, closing, expansion and renewal	README
dsh-growth	Acquisition, activation, retention, revenue analysis and growth experiments	README
dsh-geo	SEO/GEO/AEO, content production and search/answer-engine discoverability	README

Recommended Handoffs

Output from this plugin	Hand off to	Handoff question
Opportunity themes, target users, contexts and pain evidence	dsh-product	Can we build a smallest useful product and test the key assumptions?
Value, payment signals and competitive/substitute evidence	dsh-business	Which model and pricing structure can be profitable?
Validated user language, problems and search signals	dsh-geo	Which content will help target users find and trust us?
Experiment results and new user feedback	Back to this plugin	Which opportunity, user or context needs more research?

idea_opportunity_handoff converts an idea_review result into the versioned opportunity-handoff contract (handoffVersion: 1.0). It preserves the source, target user, scene, problem, current workarounds, evidence, riskiest assumption and validation experiment for dsh-product. A partial handoff is an explicit research gap, not a product requirement.

What it does

• Scan public HTTP(S) pages from communities, issue trackers, reviews, trend pages, competitor pages or research reports without credentials.
• Run an onboarding/readiness check for an idea-discovery workspace and its external research evidence.
• Import optional structured or Markdown-based pain-point records without losing source lineage.
• Filter by keyword, target user, scene, source type, date, pain score and repeat count.
• Rank signals by latest, repeated, painful or verifiable.
• Build opportunity themes from user × scene groups instead of turning each complaint into a feature.
• Draft multiple solution lenses: automation, copilot, workflow, marketplace and education.
• Generate an experiment with audience, steps, evidence, thresholds, duration and decision rule.
• Keep facts, signals, inferences and validated findings separate.
• Generate source-specific research plans for Product Hunt, TrustMRR, Reddit, G2, GitHub, Google Trends and other external sources.
• Score evidence by behavior, workaround, cost, repetition, cross-source support, commercial signals and “why now”.
• Generate JTBD, The Mom Test, switching and critical-event interview guides.
• Convert an idea_review result into an Opportunity Solution Tree with multiple solutions and experiments.

The external opportunity radar is intentionally only a research entry point. It does not perform unrestricted crawling, claim that a hot topic proves demand, or replace customer discovery and validation.

Included tools

Tool	Purpose
idea_external_scan	Fetch bounded public URL snapshots and return external evidence limits
idea_source_plan	Recommend external sources, search recipes, evidence fields and platform biases for a topic
idea_opportunity_radar	Combine external source planning with optional user-provided public URL snapshots
idea_onboarding	Check workspace readiness, signal coverage, evidence gaps and the next discovery step
idea_signal_import	Normalize a local Markdown/CSV/TSV/JSON/JSONL source into signal cards
idea_evidence_score	Score signal evidence strength and demand stage without claiming demand proof
idea_interview_guide	Generate JTBD, Mom Test, switching or critical-event interview scripts
idea_signal_filter	Filter and rank signals with evidence quality attached
idea_opportunity_map	Cluster signals into opportunity themes with risks and workarounds
idea_candidate_draft	Draft multiple candidate directions from opportunity themes
idea_experiment_plan	Turn one candidate into a smallest useful validation experiment
idea_review	Run the complete signal → opportunity → candidate → experiment review
idea_ost	Turn an idea review into an Opportunity Solution Tree
idea_report	Render an idea_review result as a shareable Markdown report

Standard method

1. Define the external market, region, target user, desired outcome and constraints.
2. Scan or research multiple public sources with URL, date and evidence limits.
3. Extract behavior, workaround, cost and “why now” signals.
4. Filter a research view by topic, user, scene, source, time and evidence strength.
5. Cluster signals into opportunities and separate them from solutions.
6. Draft more than one solution direction.
7. Test the riskiest assumption with interviews, concierge service, prototype, landing page or preorder.
8. Decide whether to continue, adjust or pause based on explicit thresholds.

Input example

problem,user,scene,source,sourceUrl,behavior,workaround,cost,painScore,repeatCount,capturedAt
每周手工整理报告,运营人员,内容运营,访谈,https://example.com/interview,复制多个表格再合并,Excel + 脚本,每周 4 小时,8,5,2026-08-20

External boundary

External scanning is limited to user-selected public HTTP(S) URLs. The plugin does not use cookies, credentials, private accounts, unrestricted crawling, live posting or outreach. Local files under defaultRoot remain available as optional caches, exports and supplementary evidence.

References

• skills/idea-discovery/SKILL.md: operating contract, method selector and output rules.
• references/methods-and-resources.md: method cards, signal library, pain-point radar and official resources.
• assets/templates/: copyable context, signal, opportunity and experiment templates.
• docs/项目架构与使用指南.md: architecture and tool workflow.

License

MIT