Guide · Agent skill
How to use the GitHub Deep Research agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The GitHub Deep Research skill is an agent skill that tells an AI agent how to investigate a GitHub project in several rounds and write a structured markdown report. It is aimed at developers evaluating open-source projects and at analysts or students who need a documented history and competitive view of a repository. After this guide you will know what the skill does, how to install it, how to phrase a request so it triggers, and what it needs to run.
GitHub Deep Researchbytedance/deer-flow · MIT Download the skill Quick start
- 1Download the ZIP. Download the github-deep-research skill as a ZIP from this page.
- 2Unzip into a skills folder. Extract it into ~/.claude/skills/ for personal use or into .claude/skills/ inside a project.
- 3Or upload in the Claude apps. Go to Settings → Capabilities → Skills and upload the ZIP instead.
- 4Confirm Python 3 and tools. Make sure Python 3 is available and that your agent has web search and web fetch tools.
- 5Ask for a report. Send a request that includes a GitHub repository URL or owner and repo name, and ask for a deep research report.
- 6Review the output. Open the saved research_{topic}_{YYYYMMDD}.md file and check the confidence ratings and citations before relying on it.
What an agent skill is and what this one does
An agent skill is a folder of instructions and supporting files that an AI agent loads when a request matches the skill's description. This skill comes from the bytedance/deer-flow project, under the MIT licence, and lives in skills/public/github-deep-research.
When it is active, the agent follows a fixed research process on a GitHub project. It first runs a bundled Python script that pulls repository facts from the GitHub API, including summary, README, file tree, commits, issues, pull requests, releases and contributors. It then runs several web searches and page fetches on architecture, competitors, community discussion and roadmap.
- Round 1: GitHub API data via scripts/github_api.py
- Round 2: 3-5 discovery web searches for overview, official repo and main competitors
- Round 3: 5-10 deeper searches and page fetches on architecture, timeline, community sentiment
- Round 4: commit, issue, PR and contributor analysis for feature evolution
- Output: a markdown report with executive summary, timeline, metrics tables, Mermaid diagrams, confidence ratings and inline citations
Install the github-deep-research skill
Download the skill as a ZIP from this page. Then choose one of the following methods.
For a personal install across all projects, unzip the folder into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder. In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/
- Project only: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Check the requirements before you start
The skill needs a few things in place. It runs a Python 3 script that uses the requests library, and falls back to Python's built-in urllib if requests is not installed. The host agent must also provide web search and web fetch tools, because most of the research happens there.
A GitHub token is optional. The source says it gives higher API rate limits, but the token handling is not shown in the excerpt, so check the full script before relying on it.
- Python 3
- requests library (optional; urllib is used as fallback)
- Web search and web fetch tools in the host agent
- Optional GitHub token for higher rate limits (verify in the full script)
Trigger the skill with a natural request
The skill's description lists the situations it should handle: comprehensive analysis, timeline reconstruction, competitive analysis, and in-depth investigation of GitHub. It also triggers on a GitHub repository URL or on open-source project names.
Plain requests work well. Include the repository URL or owner and repo name, and say what kind of report you want.
- Do a deep research report on https://github.com/owner/repo, covering its timeline and main competitors.
- Analyse the history of the owner/repo project from its commits and issues and write a structured report.
- Compare this open-source project with its alternatives and tell me whether it is worth adopting.
What the report contains
The report follows the template in assets/report_template.md. It opens with a metadata block (date, confidence level, subject), then an executive summary, a chronological timeline, topic-specific analysis, metrics and comparisons, strengths and weaknesses, sources, a confidence assessment and a methodology section.
Claims from external sources carry inline citations in the form [citation:Title](URL). Confidence is rated High (90%+), Medium (70-89%) or Low (50-69%) based on source quality. The report is saved as research_{topic}_{YYYYMMDD}.md.
Understand the safety notes
The skill runs a script and uses the network, so review it before installing from any source. The GitHub script calls the public GitHub API. The excerpt shows no credentials, so private repositories will fail or return limited data.
The skill itself warns that web sources can be outdated or biased. Treat the report as a research draft and verify important facts, especially for adoption decisions.
- Runs scripts: yes (scripts/github_api.py)
- Uses network: yes (GitHub API, web search, web fetch)
- Private repositories are not supported by the excerpt's public-API approach
What people use it for
- Deciding whether an open-source library is worth adopting for a project
- Reconstructing how a repository evolved from its commits, issues and releases
- Comparing a project with its competitors for a university or analyst report
- Reviewing a project's contributor activity and community discussion
Frequently asked questions
How do I install the GitHub Deep Research skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can upload the ZIP under Settings → Capabilities → Skills.
What do I need to run the github-deep-research skill?
You need Python 3, and the agent needs web search and web fetch tools. The requests library is optional because the script falls back to Python's built-in urllib. A GitHub token is optional for higher rate limits, but the excerpt does not show how it is used.
Is the GitHub Deep Research skill free?
The source identifies the licence as MIT, which is shown on this page. It does not describe pricing, so any costs depend on your own agent setup and tools.
Can it research private GitHub repositories?
Not reliably. The script calls the public GitHub API, and the excerpt shows no credentials, so access to private repositories will fail or be limited.
Why does the skill not trigger on my request?
It triggers on GitHub repository URLs, open-source project names and requests for comprehensive analysis, timeline reconstruction or competitive analysis. Include a repository URL or clearly ask for a deep research report.
How is this different from a quick web search?
The skill runs several rounds: GitHub API data first, then discovery, deeper investigation and commit and issue analysis. It produces a long structured report with citations, which is slower than a quick factual lookup.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits developers evaluating an open-source project and analysts who need a documented timeline and competitor view of a repository. Skip it for quick lookups, private repositories, or decisions that need verified facts without human review, since the skill itself warns that web sources can be outdated or biased.
Download the skill