Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add PANGKAIFENG/ai-product-manager-skills --skill research-topic-compilergit clone --depth 1 https://github.com/PANGKAIFENG/ai-product-manager-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/pangkaifeng/ai-product-manager-skills/research-topic-compiler)<a href="https://agentmods.dev/skills/pangkaifeng/ai-product-manager-skills/research-topic-compiler"><img src="https://agentmods.dev/badge/skills/pangkaifeng/ai-product-manager-skills/research-topic-compiler/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/pangkaifeng/ai-product-manager-skills/research-topic-compiler"><img src="https://agentmods.dev/badge/skills/pangkaifeng/ai-product-manager-skills/research-topic-compiler.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00319 | $0.08483 |
| Opus 5 | $0.00160 | $0.04241 |
| Sonnet 5 | $0.00064 | $0.01697 |
| Haiku 4.5 | $0.00032 | $0.00848 |
Grade A, and why
research-topic-compiler scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 333 lines — stays where its author put it; the contents beside it link to each section on GitHub.
产品研究编译器(research-topic-compiler)
中文速查
- 中文名:产品研究 / 证据与决策输入
- 英文稳定名:
research-topic-compiler - 分类:研究学习 / Obsidian 知识编译
- 你可以这样叫我:
系统研究这个主题、帮我整理到 Obsidian、做一个深度专题、研究行业最佳实践、概念解读、概念源流、PM 技术评审提问脚本、行业演进看板、把这个大白话拆成研究目标 - 适合:围绕产品问题做多渠道证据收集、竞品与替代方案比较、用户/市场信号、证据矩阵、阶段结论、候选池、研究看板和应用转化;也适合把用户的大白话、模糊主题或 Roadmap 前置想法转成研究目标、研究问题和输出要求
- 默认学习型 HTML:当用户要逐步理解、深入解释、对比自身现状并落到行动时,使用
learning-report-html,不要把研究控制面直接投影成 Dashboard。 - 不适合:创建或评审 Skill;普通新闻搜索或一次性摘要;明确要“选一个 / 给最终推荐 / 排除其他方案”时改用
decision-research
Overview
使用这个 Skill 把一个研究主题编译成可学习、可追溯、可继续扩展、并能按用户画像转化为实际工作判断的 Obsidian Research Project 或聊天内研究报告。
核心原则:
- 先把用户原话转成明确研究目标、研究问题和输出要求,再判断研究深度、渠道和样本量。
- 研究框架不是固定报告目录:先识别
Research Job,分别维护 Evidence/Explanation Framework;V0 只是可修订假设,最终报告必须从证据更新后的 Framework Vn 重新编译。 - 先解析用户画像,再决定解释方式、案例选择、实践任务和应用转化。
- Seed Corpus 是线索和初始假设来源,不默认是权威证据;二手核心 Claim 要追溯原始来源或披露无法追溯的影响。
- Normal Research 和 Application 围绕最高价值证据缺口迭代;每轮只执行一个能降低关键不确定性的 Next Best Evidence Action。
- Research 只拥有证据覆盖、来源、矛盾、置信度和残余 gap;最终推荐、排除逻辑、方案设计和 readiness 属于其他 owner。
- Obsidian 是内部基线和默认沉淀位置,不是唯一研究渠道。
- 外部渠道动态选择,不默认全开;根据主题类型、证据缺口、时效性和可信度要求启用。
- 需要扩源时把
Pre-Research Source Expansion作为候选发现策略:用公开搜索、垂直 API、RSS、产品/市场目录等渠道寻找能关闭当前 Gap 的来源,再筛选进入正式证据矩阵。 - 结论必须能回到证据矩阵;
05_研究报告是第一阅读入口,02_证据与卡片是按需深挖层。 - 系统学习不是重型课程仓库;默认保持轻量,只有触发条件满足时才建议独立学习包文件。
- 当研究会影响产品策略、商业化、工作台/连接器设计、企业 adoption 或其他高成本决策时,默认按高门槛应用研究处理,读取
references/applied-business-research-contract.md。 - 轻量概念解构属于本 Skill 的研究模式,不再单独使用独立概念看板 Skill;它适合快速建立概念源流、语义漂移、范式阶段和 PM 决策问题。
- 用户画像只影响解释深度、案例选择、输出结构和实践任务,不覆盖用户当前明确要求。
- 研究必须能转成行动:判断、方案、模板、任务、PRD、Workflow、Eval、Checklist、SOP、路线图或实践练习。
- 用户补充的新渠道可以进入渠道库,但要先判断适用主题、访问条件、证据强度和风险。
- 微信公众号、X、私域社区、付费库等渠道默认只能做公开候选发现;任何登录态读取、客户端转发、发送到 Obsidian 同步号或第三方服务的动作,都需要当前 run 的明确授权和可见确认点。
Input / Context Intake
启动研究前先收集或推断这些上下文;不要问本地文件能发现的信息,只在答案会改变研究范围、访问权限或写回位置时追问:
- 原始意图:用户原话、业务愿望、想产出的材料、隐含的后续动作。
- 研究主题:主题名称、用户要解决的决策或学习目标、是否已有种子资料。
- 预期产物:聊天内报告、Obsidian Research Project、更新已有专题、还是长期雷达。
- 深度约束:用户期望的速度、深度、样本量、是否需要
L4/L5级外部扩展。 - 内部基线:是否扫描 Obsidian、哪些 Vault/目录可用、是否只读
笔记同步助手。 - 渠道偏好:必须看的渠道、明确排除的渠道、是否需要产品研究、GitHub、官方文档、论文、社区或 X。
- 访问边界:登录、API token、付费报告、私密社区、公司内部资料和引用限制。
- 写回边界:目标目录、命名规则、是否允许新增渠道到
channel-registry.md。 - 用户画像:角色、领域、技术深度、目标类型、输出偏好、应用场景和最终决策需求;先按
User Context Resolution解析,不要默认每次追问。
What ships with it
60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- assets/semantic-editorial-template.html 9.3 KB
- evals/evals.json 46 KB
- evals/fixtures/anthropic-skill-loop/catalog.json 5.2 KB
- evals/fixtures/anthropic-skill-loop/seed-wechat.md 931 B
- evals/fixtures/anthropic-skill-loop/sources/agent-skills-spec.md 713 B
- evals/fixtures/anthropic-skill-loop/sources/anthropic-primary.md 748 B
- evals/fixtures/anthropic-skill-loop/sources/anthropic-same-lineage.md 604 B
- evals/fixtures/anthropic-skill-loop/sources/fake-google-official.md 647 B
- evals/fixtures/anthropic-skill-loop/sources/openai-peer.md 645 B
- evals/fixtures/anthropic-skill-loop/sources/repo-canonical.md 561 B
- evals/fixtures/anthropic-skill-loop/sources/repo-counterexample.md 662 B
- evals/fixtures/anthropic-skill-loop/sources/repo-independent.md 716 B
- evals/fixtures/anthropic-skill-loop/sources/repost-duplicates.md 351 B
- evals/fixtures/anthropic-skill-loop/sources/saturation-independent-a.md 501 B
- evals/fixtures/anthropic-skill-loop/sources/saturation-independent-b.md 515 B
- evals/fixtures/boundaries/authoritative-sufficient-pack.md 1.1 KB
- evals/fixtures/boundaries/l2-agent-skill-official-pack.md 786 B
- evals/fixtures/boundaries/openai-input-items.md 563 B
- evals/fixtures/learning-report-holdout/release-safety-pack.md 2.0 KB
- evals/fixtures/learning-report-transfer/activation-interviews.md 2.3 KB
- evals/fixtures/mcp-auth-transfer/catalog.json 1.7 KB
- evals/fixtures/mcp-auth-transfer/seed-blog.md 483 B
- evals/fixtures/mcp-auth-transfer/sources/mcp-counterexample.md 508 B
- evals/fixtures/mcp-auth-transfer/sources/mcp-independent-security.md 422 B
- evals/fixtures/mcp-auth-transfer/sources/mcp-sdk-implementation.md 433 B
- evals/fixtures/mcp-auth-transfer/sources/mcp-spec-auth.md 591 B
- evals/fixtures/research-dashboard/dashboard.html 9.3 KB
- evals/fixtures/research-dashboard/summary.md 1.7 KB
- evals/graders/dashboard-artifact-rubric.md 2.8 KB
- evals/graders/iterative-research-rubric.md 4.2 KB
- evals/graders/learning-report-rubric.md 3.2 KB
- evals/graders/research-framework-compilation-rubric.md 5.2 KB
- evals/README.md 3.8 KB
- references/applied-business-research-contract.md 3.0 KB
- references/browser-walkthrough-boundaries.md 2.6 KB
- references/candidate-backlog-schema.md 3.3 KB
- references/channel-registry.md 13 KB
- references/channel-selection-rubric.md 8.4 KB
- references/concept-lens-design-quality.md 4.4 KB
- references/concept-lens-html-dashboard-template.md 3.6 KB
- references/concept-lens-output-contract.md 1.8 KB
- references/concept-lens-paradigm-framework.md 4.3 KB
- references/concept-lens-source-and-factuality.md 3.2 KB
- references/core-loop-research-handoff.md 3.3 KB
- references/cross-session-handoff.md 4.8 KB
- references/default-user-profile.md 1.4 KB
- references/editorial-projection-gate.md 5.0 KB
- references/iterative-research-loop.md 19 KB
- references/learning-pack-standards.md 3.5 KB
- references/learning-report-output-contract.md 7.4 KB
- references/mode-routing-guide.md 7.6 KB
- references/mode-selection.md 4.0 KB
- references/obsidian-output-contract.md 7.4 KB
- references/post-research-exits.md 2.9 KB
- references/pre-research-source-expansion.md 9.2 KB
- references/product-decision-brief-template.md 2.6 KB
- references/product-decision-mode.md 3.3 KB
- references/product-evidence-channel-guide.md 3.6 KB
- references/project-context-intake.md 2.3 KB
- references/report-writing-standards.md 8.5 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 333 lines · 319 tokens per session scan A f807cd923384
research-topic-compiler is a skill published in the GitHub repository PANGKAIFENG/ai-product-manager-skills (11 stars, last pushed 13d ago), licensed MIT. It adds 319 tokens to every session and 8,483 once invoked, about $0.0016 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
baoyu-youtube-transcript
A tool for downloading the written captions, subtitles, chapter information, speaker labels, and cover image from a YouTube video using its URL or ID.
baoyu-comic
A tool for creating original educational comics that explain knowledge or ideas through multiple illustrated panels. It supports different art styles and tones and can create several comics in one batch.
baoyu-diagram
Create professional, dark-themed SVG diagrams of any type — architecture diagrams, flowcharts, sequence diagrams, structural diagrams, mind maps, timelines, illustrative/conceptual diagrams, and more. Use this skill whenever the user asks for any kind of technical or conceptual diagram, visualization of a system…
playwright-cli
Automates browser interactions for testing and validating your own web applications using playwright-cli. Use when you need terminal-first browser control for navigation, form filling, screenshots, tracing, bound browser sessions, debugging, or generating Playwright test code. Only use against applications you own or…
agent-wiki
Incremental LLM-friendly wiki generator for Obsidian note vaults. Use when: (1) Building wiki from notes, (2) Ingesting notes to wiki, (3) Obsidian LLM wiki, (4) Incremental knowledge base management. Triggers: 'build wiki from notes', 'ingest notes to wiki', 'Obsidian LLM wiki', 'incremental knowledge base'.
sequential-think
Multi-step reasoning engine for complex analysis and systematic problem solving. Use when: (1) Complex debugging scenarios with multiple layers, (2) Architectural analysis and system design, (3) Problems requiring hypothesis testing and validation, (4) Multi-component failure investigation, (5) Performance bottleneck…