Skills Explorer Add Skills

deep-research

An agent skill by daymade, from daymade/claude-code-skills. Tags: analysis, documentation, research, strategy, validation.

What it does

Generate format-controlled research reports with evidence tracking, citations, source governance, and multi-pass synthesis. This skill should be used when users request a research report, literature review, market or industry analysis, competitive landscape, policy or technical brief. Triggers: "帮我调研一下", "深度研究", "综述报告", "深入分析", "research this topic", "write a report on", "survey the literature on", "competitive analysis of", "技术选型分析", "竞品研究", "政策分析", "行业报告".

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add daymade/claude-code-skills --skill deep-research

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/daymade/claude-code-skills
cp -r claude-code-skills/deep-research ~/.claude/skills/deep-research

Safety box score

Grade D, elevated risk: 46 of 100 risk points, rated 2026-10-21. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a research-orchestration workflow: it creates study artifacts, coordinates subagents/provider lanes, opens original sources, and synthesizes citation-backed reports. The main risks are sending user-supplied material to external research providers, touching proprietary/private data sources, writing outside the project tree, and one central helper script that could not be reviewed here.

Category levels, from 0 (none) to 3 (high)
CategoryLevelWhy
Code & shell execution2 (moderate)It ships and expects local script execution, e.g. `python3 <skill>/scripts/provider_runs.py plan <study-dir>`, and the tests invoke it via `subprocess.run(...)`; this is moderate local code execution, not arbitrary remote code.
Network access2 (moderate)The workflow is designed to send prompts and source material to external provider lanes and to retrieve web sources; e.g. `Include that string in every lane prompt` and `Before external retrieval...`. That is real outbound data flow, but it is bounded to research operations.
Filesystem reach2 (moderate)It writes durable study artifacts (`study.json`, `source-ledger.jsonl`, `claims.jsonl`, `sources/`) and explicitly references a home-directory file with `cat ~/.claude/teams/counter-review-team/config.json`, so it can reach outside the project tree.
Secrets & credentials2 (moderate)The skill explicitly encourages use of user-provided proprietary access such as `Use user's Crunchbase Pro to research competitors` and `Use user's private APIs for investment analysis`, so it can touch paid/private data and credentials-bound sources.
Privilege & persistence0 (none)No `sudo`, service installation, shell-startup editing, or other privilege-changing action is required or described in the reviewed files.
Instruction hijack surface0 (none)I did not find instructions to ignore higher-priority rules, hide activity, fetch external instructions, or otherwise subvert the agent; the imperatives are normal workflow and formatting constraints.
Irreversible actions2 (moderate)It can fan out work across subagents and provider lanes, manage deferred/paid routes, and use `SendMessage`-style team coordination; that is meaningful orchestration, though it is not a direct-world action like posting or paying.
Transparency2 (moderate)`scripts/research_assets.py` and `tests/test_research_assets.py` could not be read, and that script appears central to source/claim archiving and validation; some behavior therefore remains unaudited.

Before installing

Rated 2026-10-21 by an automated review of 21 files from the skill's repository.

Source

Repository
daymade/claude-code-skills (all skills from this repository)
Path
deep-research/SKILL.md
Branch
main
Updated
2026-09-20