sn-deep-research

用于用户请求深度研究、系统性研究、竞品分析、方案对比、趋势分析或事实核查时。**遇到以下任一情况就主动使用本 skill，不要自行搜几条就回答**：①用户出现触发词：深度研究 / 深度调研 / 深入研究 / 全面研究 / 系统研究 / 调研 / 调查 / 尽调 / 行业研究 / 市场研究 / 竞品分析 / 政策研究 / 技术研究 / 趋势研究 / 事实核查 / 写一份研究报告 / 调研报告 / 深度报告 / research / deep research；②请求需要跨多来源取证、多维度对比、交叉验证才能给出可靠结论；③用户要求产出报告、白皮书、行业分析或尽调文档；④话题涉及最新政策/市场/产品/价格/法规，需要系统核查。明确要求核验来源的单点事实可走 quick；无核验要求的简单常识问答不使用。模糊或宽泛的"研究/了解一下 X"也优先触发。仅不用于：一句话摘要、已给定单一来源的整理、纯文字润色改写。

Quality

87%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

85%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A highly actionable, rigorously sequenced orchestration spec with explicit validation gates, feedback loops, and clean delegation of per-role detail to separate files. Its main weakness is length — the controller prose is verbose and could be tightened without losing operational precision.

Suggestions

Compress the §4.1 clarification/format prose into tighter rules or a table; several paragraphs restate the same gating logic and could be consolidated to cut length.

Move the §6 controller-context table earlier or fold its key read-boundary rules into §1, since much of it repeats constraints already stated in §2 and §5.

Dimension	Reasoning	Score
Conciseness	The body is operational (contracts, paths, flags) rather than explaining concepts Claude already knows, but at ~580 lines it carries redundant prose and elaboration that could be tightened, so it is mostly efficient but not lean.	2 / 3
Actionability	Concrete, copy-paste-ready guidance throughout: full payload templates with exact absolute-path placeholders, exact validator invocations with flags (e.g., «--require-version 1.2 --expected-mode {initial\|supplement}»), and specific output paths per stage.	3 / 3
Workflow Clarity	Clear per-tier pipelines in §4 with explicit validation gates (validate_plan/validate_evidence/validate_outline), a failure-and-retry table with routes and ceilings (§4.3), and feedback loops (revise→revalidate) for risky batch/render operations.	3 / 3
Progressive Disclosure	A reading map and numbered §1–§6 sections give clear navigation, and per-role detail is split one level deep into «agents/*.md» with schemas and scripts referenced by path; the four referenced scripts (validate_evidence/validate_plan/validate_outline/source_snapshot) are present in the bundle.	3 / 3
	Total	11 / 12 Passed

Description

90%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A trigger-rich, well-scoped description that clearly signals when to invoke the skill and distinguishes it from adjacent quick-research tasks. The what-it-does side is present but phrased as use-case categories rather than a crisp list of discrete operations, and the description is on the verbose side.

Suggestions

Lead with a concise one-sentence statement of what the skill does (e.g., 'Orchestrates a multi-agent pipeline that gathers cross-verified evidence and produces a structured research report') before the trigger conditions, to sharpen the 'what' half of completeness.

Trim the long inline trigger-word list into a few representative triggers plus 'etc.' to reduce verbosity without losing the use-when signal.

Dimension	Reasoning	Score
Specificity	Names the domain and several concrete action types («多来源取证», «多维度对比», «交叉验证», «产出报告/白皮书/行业分析/尽调文档»), but these are framed as use-case categories rather than discrete operations, so not the full multi-action list of the level-3 anchor.	2 / 3
Completeness	Explicitly answers what (multi-source evidence gathering, cross-verification, report production) and when (four numbered trigger conditions plus a long trigger-word list and explicit exclusions), with explicit «使用本 skill» guidance.	3 / 3
Trigger Term Quality	Excellent coverage of natural trigger terms users would actually say («深度研究», «深度调研», «调研», «调查», «尽调», «竞品分析», «事实核查», «写一份研究报告», «research», «deep research»), spanning common variations.	3 / 3
Distinctiveness Conflict Risk	Clear niche (deep research orchestration) with distinct triggers and explicit exclusions («一句话摘要», «已给定单一来源的整理», «纯文字润色改写», simple common-sense Q&A), making wrong-skill conflicts unlikely.	3 / 3
	Total	11 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (586 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: OpenSenseNova/SenseNova-Skills
Path: skills/sn-deep-research/SKILL.md
Commit: 179fea1

Reviewed: 2 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.