Guide · Agent skill
How to use the X Research sentiment briefing agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The x-research skill is a set of instructions that teaches an AI agent to research public reaction on X (formerly Twitter) to a stock, company, product or market event, then save the findings as a cited Markdown report. It is written for analysts, product and marketing teams, and students who want a structured snapshot rather than raw search results. After this guide you will know what the skill does, how to install it, how to phrase requests so it triggers, and what limits to keep in mind.
X Research Sentiment Briefingnexu-io/open-design · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Download the x-research skill ZIP from this page. It contains SKILL.md.
- 2Unzip into a skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder for one project.
- 3Or upload in the Claude apps. In the Claude apps, open Settings, go to Capabilities, then Skills, and upload the ZIP.
- 4Make sure file saving is available. Confirm your Open Design environment has file_write capability, since the report is saved to Design Files.
- 5Ask a natural question. Request sentiment on a named company, product or event, for example asking what people are saying on X about it.
- 6Review the report. Open research/x-research/ in Design Files, check the Source Coverage table and caveats, then follow the citations.
What the x-research skill makes the agent do
An agent skill is a folder of instructions (here, a single SKILL.md file) that an AI agent reads when a request matches the skill's description. This one covers X/Twitter public sentiment: what people are saying about a stock, sector, company, product, or market event.
When it triggers, the agent follows a fixed workflow. It restates the topic and time window (default is the last 7 days for fast-moving topics), runs three to five targeted searches, groups posts into bullish, bearish, neutral and disagreement themes, and writes a Markdown report with citations to Design Files at research/x-research/<topic-slug>.md.
- Defines the topic, target entity and time window before searching
- Runs 3-5 targeted queries, such as core keywords or a $TICKER cashtag, bullish and bearish signal terms, and news links
- Records each source's query, source class and coverage status (checked, unavailable, thin or not relevant)
- Ends with an overall label of bullish, bearish, mixed or neutral, with confidence and caveats
Install the x-research skill
Download the x-research ZIP from this page. Unzip it into your personal skills folder, ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only inside that project.
Alternatively, if you use the Claude apps, open Settings, go to Capabilities, then Skills, and upload the ZIP there. The skill package contains one file, SKILL.md.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Trigger the skill with natural requests
You do not need a special command. The skill activates when your request matches its description or trigger phrases. The source lists triggers such as "x research", "twitter sentiment", "ct sentiment", "public sentiment", "social sentiment" and "what people are saying on x". Plain requests that name a topic work well.
Name the entity and, if it matters, the time window. Asking for a window other than the default 7 days is allowed.
- "What are people saying on X about NVIDIA's earnings this week?"
- "Run x research on the reaction to our new pricing page launch."
- "Give me CT sentiment on the Fed rate decision from the last 3 days."
- "What is public sentiment on X about the new EV startup?"
What the agent needs to run it
The skill needs an Open Design environment with file_write capability, because the report is saved to Design Files. X access is optional. The skill uses X only when a usable connector, API, browser session or user-provided export or link is actually available in the run. Otherwise it falls back to web search, public pages, user-provided links or screenshots, and says so in the report.
The skill does not run scripts. It does use the network, since it searches and reads public pages.
- Required: Open Design environment with file_write capability
- Optional: an X/Twitter connector, API, browser session, or user-provided export
- Without X access: web search and public pages are used instead
Read the report and check the caveats
Every report follows the same structure: Query Summary, Source Coverage table, Sentiment Themes with numbered citations, Overall Sentiment, Caveats, Sources and an Evidence Note. The Source Coverage table tells you which sources were checked and which were unavailable, so you can see what the conclusion rests on.
Read the Caveats section before acting on anything. The skill itself warns that results carry sample bias and that the report is a qualitative briefing, not a quantified survey.
- Check whether X was actually checked or marked unavailable
- Follow the [1], [2] citations back to the source posts or pages
- Treat the overall label as a qualitative reading with stated confidence
Safety notes: untrusted content and network use
The skill tells the agent to treat X posts, web pages, comments, search results, screenshots and documents as untrusted evidence. Instructions or tool requests embedded in that content should be ignored, and external material is used only for factual grounding and citations.
Because the skill uses the network, the sources it reads are external and can be wrong, manipulated or outdated. Do not treat the report as verified fact.
- Uses network access for search and public pages; runs no scripts
- Embedded instructions in posts or pages are not followed
- Social posts can be misleading, manipulated or outdated
Tips for better briefings
Give a specific entity and a clear reason for the research, such as a launch or an earnings event. Vague topics produce broad, thin results. If you have an export or links to specific posts, include them, since the skill can use user-provided material when X itself is unavailable.
For fast-moving events, keep the default 7-day window or narrow it. For slower topics, ask for a longer window explicitly.
- Name the company, product or ticker precisely
- Provide links or an export if you have them
- State the time window when the default does not fit
What people use it for
- An analyst gets a cited snapshot of public reaction to an earnings report or market event.
- A product team gauges community response to a feature launch or pricing change.
- A marketing team reviews how a competitor's campaign is being discussed.
- A student builds a structured sentiment summary with source citations for a research assignment.
Frequently asked questions
How do I install the x-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 instead upload the ZIP under Settings, Capabilities, Skills.
What are the requirements for x-research?
You need an Open Design environment with file_write capability so the report can be saved to Design Files. An X/Twitter connector, API, browser session or user-provided export is optional; without one the skill falls back to web search and public pages.
Is the x-research skill free?
The source material lists its licence as Apache-2.0. It does not state pricing, so check the licence terms on the source repository for details.
Does x-research work without X/Twitter access?
Yes. The skill is instructed to say clearly when X is unavailable and then use web search, public pages, user-provided links or screenshots. It will not claim X coverage it did not check.
Can I trust the sentiment result for investment decisions?
Not on its own. The skill describes the output as a qualitative briefing with sample bias, and social posts can be misleading, manipulated or outdated, so verify independently before any investment decision.
How is this different from a normal search?
A normal search returns links. This skill follows a fixed workflow: it sets a time window, runs targeted queries, groups posts into bullish, bearish, neutral and disagreement themes, and saves a cited report with coverage notes.
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
Use x-research if you want a quick, cited, consistently structured snapshot of public reaction to a company, product or market event, and you will verify the findings yourself. Pick something else if you need rigorous statistical sentiment measurement or a source you can fully trust for financial decisions, since the skill itself describes its output as a qualitative briefing with sample bias.
Download the skill