← 返回 Skills 市场

Deep Research.Bak

Name: Deep Research.Bak
Author: emiltsoi

作者 emiltsoi · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ⚠ suspicious

总下载

当前安装

版本数

在 OpenClaw 中安装

/install deep-research-bak

功能描述

Conducts enterprise-grade research with multi-source synthesis, citation tracking, and verification. Produces citation-backed reports through a structured pi...

使用说明 (SKILL.md)

Deep Research

Core Purpose

Deliver citation-backed, verified research reports through a structured pipeline with source credibility scoring, evidence persistence, and progressive context management.

Autonomy Principle: Operate independently. Infer assumptions from context. Only stop for critical errors or incomprehensible queries.

Decision Tree

Request Analysis
+-- Simple lookup? --> STOP: Use WebSearch
+-- Debugging? --> STOP: Use standard tools
+-- Complex analysis needed? --> CONTINUE

Mode Selection
+-- Initial exploration --> quick (3 phases, 2-5 min)
+-- Standard research --> standard (6 phases, 5-10 min) [DEFAULT]
+-- Critical decision --> deep (8 phases, 10-20 min)
+-- Comprehensive review --> ultradeep (8+ phases, 20-45 min)

Default assumptions: Technical query = technical audience. Comparison = balanced perspective. Trend = recent 1-2 years.

Workflow Overview

Phase	Name	Quick	Standard	Deep	UltraDeep
1	SCOPE	Y	Y	Y	Y
2	PLAN	-	Y	Y	Y
3	RETRIEVE	Y	Y	Y	Y
4	TRIANGULATE	-	Y	Y	Y
4.5	OUTLINE REFINEMENT	-	Y	Y	Y
5	SYNTHESIZE	-	Y	Y	Y
6	CRITIQUE	-	-	Y	Y
7	REFINE	-	-	Y	Y
8	PACKAGE	Y	Y	Y	Y

Execution

On invocation, load relevant reference files:

Phase 1-7: Load methodology.md for detailed phase instructions
Phase 8 (Report): Load report-assembly.md for progressive generation
HTML/PDF output: Load html-generation.md
Quality checks: Load quality-gates.md
Long reports (>18K words): Load continuation.md

Templates:

Report structure: report_template.md
HTML styling: mckinsey_report_template.html

Scripts:

python scripts/validate_report.py --report [path]
python scripts/verify_citations.py --report [path]
python scripts/md_to_html.py [markdown_path]

Output Contract

Required sections:

Executive Summary (200-400 words)
Introduction (scope, methodology, assumptions)
Main Analysis (4-8 findings, 600-2,000 words each, cited)
Synthesis & Insights (patterns, implications)
Limitations & Caveats
Recommendations
Bibliography (COMPLETE - every citation, no placeholders)
Methodology Appendix

Output files (all to ~/Documents/[Topic]_Research_[YYYYMMDD]/):

Markdown (primary source)
HTML (McKinsey style, auto-opened)
PDF (professional print, auto-opened)

Quality standards:

10+ sources, 3+ per major claim
All claims cited immediately [N]
No placeholders, no fabricated citations
Prose-first (>=80%), bullets sparingly

When to Use / NOT Use

Use: Comprehensive analysis, technology comparisons, state-of-the-art reviews, multi-perspective investigation, market analysis.

Do NOT use: Simple lookups, debugging, 1-2 search answers, quick time-sensitive queries.

安全使用建议

This package appears to implement the advertised research pipeline, but it gives the agent permission to read and write files in your home directory and to spawn recursive continuation agents that load saved state. Before installing or enabling it: 1) review the remaining code files you haven't inspected (especially research_engine.py and verify_citations.py) for network calls or any calls that read arbitrary paths; 2) consider running the skill in a sandboxed account or VM so its file writes (~/Documents and ~/.claude/research_output) are isolated; 3) do not provide unrelated API keys or system credentials; only configure optional search-cli keys if you trust the provider setup; 4) if you allow the skill to run, audit created continuation_state and sources.json files and confirm the skill deletes or manages them per your data-retention policy. If you want, I can scan the remaining omitted scripts for network or file-access patterns and point out any risky lines.

功能分析

Type: OpenClaw Skill Name: deep-research-bak Version: 1.0.0 The skill bundle implements a highly autonomous research pipeline that utilizes risky capabilities, including recursive sub-agent spawning via the Task tool (detailed in methodology.md and continuation.md) and broad filesystem access for report assembly. While these features are aligned with the stated goal of 'Enterprise-grade research,' they introduce significant operational risk. Furthermore, scripts/verify_citations.py contains a potential SSRF vulnerability by performing network requests to arbitrary URLs extracted from research content without validation. The bundle also promotes the installation of an external third-party utility (search-cli) from a specific GitHub repository.

能力评估

✓ Purpose & Capability

Name/description align with the included files: citation manager, report validators, markdown→HTML, source evaluator, and an orchestration engine. Persisting reports and citations to disk and converting to PDF/HTML are coherent with an enterprise research skill.

⚠ Instruction Scope

SKILL.md explicitly instructs the agent to run shell commands (e.g., date), read and write files under ~/Documents and ~/.claude/research_output, spawn Task sub-agents that load continuation_state and report files, and 'Grep/Read for local documentation'. Those directives permit reading arbitrary local documents and recursive agent spawning; while plausible for deep research, they broaden what the agent may access and persist beyond the narrow task of web search and synthesis.

✓ Install Mechanism

No install spec (instruction-only) — lowest install risk. README suggests optional external tools (brew search-cli, pip install weasyprint) if the user opts in; those are optional and clearly documented.

ℹ Credentials

The skill declares no required env vars or credentials. However README and SKILL.md suggest optional multi-provider search-cli setup requiring provider API keys, and the skill will run network-based citation verification. The absence of declared env vars is consistent but you should not supply unrelated credentials. The skill does persist state to home directory.

⚠ Persistence & Privilege

The skill persists continuation_state and sources.json under ~/.claude/research_output and writes reports to ~/Documents. It also instructs spawning continuation agents (Task tool) that read/write those files recursively. 'always' is false, but the persistence and recursive agent spawning increase blast radius if misused or if underlying orchestration code performs unexpected reads or network calls.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install deep-research-bak
安装完成后，直接呼叫该 Skill 的名称或使用 /deep-research-bak 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release of deep-research: structured, citation-backed enterprise research. - Produces comprehensive research reports with multi-source synthesis and source credibility scoring. - Triggers only on complex research needs (e.g., "deep research," "compare X vs Y," "analyze trends"). - Follows a multi-phase workflow (scope, plan, retrieve, triangulate, synthesize, critique, package) with configurable depth. - Enforces stringent output standards: executive summary, detailed findings, full bibliography, and professional file exports. - Explicitly not for simple lookups, debugging, or single-answer queries.

元数据

Slug deep-research-bak

版本 1.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题