Skip to content
Claude logo

Claude

AI assistant for writing, analysis, coding, and collaborative problem-solving.

WritingCodingAgents

How it works / How to use

AI assistant for writing, analysis, coding, and collaborative problem-solving. Start a chat on claude.ai or Claude Desktop, describe what you need, and refine in the same thread. Projects group related work with a shared knowledge base; artifacts, Claude Code, and Cowork depend on your plan.

  1. Open Claude on the web, iOS, Android, or Claude Desktop and start a new chat—or open a Project when the work belongs together.
  2. State your goal, audience, constraints, and any facts Claude should treat as true.
  3. Review the first reply; use follow-ups to refine instead of restarting from scratch.
  4. For ongoing work, add documents and project instructions to the project knowledge base so context carries across chats.

General chat and analysis

Lead with the question, then add context and the output shape you want (bullets, steps, pros/cons, short paragraph).

What to provide

  • The question, decision, or topic you need help with
  • Background facts Claude should treat as true
  • How detailed the answer should be (quick take vs careful explanation)

Details that improve the result

  • Name your level of expertise so the explanation matches it
  • Say what to optimize for (clarity, speed, rigor)
  • Ask Claude to flag uncertainty instead of guessing

Example prompt

Explain how SPF, DKIM, and DMARC work together for email deliverability. Audience: a developer who has never set up DNS. Output: 5 bullets plus one common misconfiguration. Mark anything you are not sure about.

If the first output is not good

  • Ask for one missing piece only, e.g. “Make the DNS example concrete for one provider.”
  • Request a shorter or longer version without repeating the whole prompt
  • Ask “What would you need from me to answer this more precisely?”

Common mistakes

  • Asking several unrelated questions in one message
  • Treating the first answer as final without checking obvious facts
  • Leaving out constraints that matter (deadline, region, tone, budget)

Writing and rewriting

State the writing job first, then paste material under a clear label such as “Draft:” or “Source notes:”.

What to provide

  • The draft, outline, or source notes (paste in chat or upload)
  • Audience, purpose, and channel (email, blog, product copy)
  • Length, tone, and words or claims to avoid

Details that improve the result

  • Include one example of voice you like
  • List facts that must appear verbatim
  • Say whether Claude should rewrite, shorten, expand, or fix grammar only

Example prompt

Rewrite this product-update email for existing customers. Tone: clear and direct, not salesy. Length: under 180 words. Keep the release date and pricing exactly as written. Do not add metrics.

Draft:
[paste draft]

If the first output is not good

  • “Keep the structure; make the opening more concrete.”
  • “Shorten by 30% and remove adjectives.”
  • “Give me two alternate subject lines under 45 characters.”

Common mistakes

  • Asking for polish without saying what “better” means
  • Forgetting to say what must not be changed
  • Mixing multiple formats (tweet + blog + FAQ) in one request

Projects with knowledge and instructions

Create or open a Project, upload context to the knowledge base, set project instructions, then start focused chats inside that project.

What to provide

  • Documents, code, or notes to upload to the project knowledge base
  • Project instructions for tone, role, or response style
  • The specific task for this chat within the project

Details that improve the result

  • Remember: context is not shared across chats unless it is in the project knowledge base
  • Free accounts can create up to five projects
  • On paid plans, large knowledge bases can expand via retrieval augmented generation (RAG)

Example prompt

Project instructions: Answer as a senior PM. Use concise bullets. Never invent metrics.

Task: From the uploaded PRD, list open questions for engineering before sprint planning. Output: table with columns Question, Why it matters, Suggested owner.

If the first output is not good

  • “Add this new doc to the knowledge base and re-check the open questions.”
  • “Tighten project instructions to require citations from uploaded files only.”
  • “Move this chat into the project from chat history.”

Common mistakes

  • Expecting Claude to remember details from other chats without adding them to project knowledge
  • Uploading many files without saying which is authoritative
  • Skipping project instructions when tone or role matters every time

Artifacts (apps, documents, and visual content)

Ask Claude to build the artifact in chat; it appears in a dedicated panel beside the conversation. Refine by describing changes or, for Markdown, highlight text and use “Edit with Claude.”

What to provide

  • A clear description of the artifact you want (tool, doc, visualization, code snippet)
  • Constraints on format, interactivity, or length
  • Any source data or style references

Details that improve the result

  • Artifacts are supported on Free, Pro, Max, Team, and Enterprise in Claude and Claude Desktop
  • Use Publish to save an artifact to your sidebar and get a shareable link (public publish on Free, Pro, Max)
  • Team and Enterprise artifacts can be shared within your organization

Example prompt

Create an interactive quiz artifact on basic SQL joins. Five questions, immediate feedback after each answer, no external libraries. Keep styling minimal and readable on mobile.

If the first output is not good

  • “Change question 3 to use a real-world inventory example.”
  • “Export-friendly Markdown version with the same content.”
  • Highlight a paragraph and use Edit with Claude for a localized fix

Common mistakes

  • Expecting sidebar access without publishing first
  • Asking for a large artifact without breaking it into steps
  • Sharing Team artifacts with people who lack project access

Coding and software tasks

Describe the task, paste relevant code or error output, and ask for code plus a brief explanation or usage example.

What to provide

  • Language, framework, and runtime
  • The exact behavior or bug to address
  • Inputs, outputs, edge cases, and style constraints

Details that improve the result

  • Include error messages verbatim
  • Say whether you want a patch, a full function, or a review
  • Claude Code on paid plans shares the same usage pool as chat for terminal-based agentic coding

Example prompt

In TypeScript, write a function `parseIsoDate(value: string): Date` that accepts YYYY-MM-DD only and throws a typed error for invalid input. Include one passing and one failing usage example. No external libraries.

If the first output is not good

  • “Handle empty string explicitly.”
  • “Show the fix as a diff against my snippet.”
  • “Suggest a regression test.”

Common mistakes

  • Omitting the language or framework
  • Pasting huge files without pointing to the relevant section
  • Accepting code without running or reading it

File and document analysis

Upload the file, name it, state the operation, and ask for a structured response (table, list, extraction).

What to provide

  • The file (PDF, DOCX, spreadsheet, presentation, code, or text)
  • The task: summarize, compare, extract, or analyze
  • Rules for quotes, numbers, and what must not be invented

Details that improve the result

  • For spreadsheets, say which columns or sheets matter
  • Ask for counts, quotes, or headings when you need extraction
  • Say whether Claude should use only the uploaded file or may use general knowledge

Example prompt

I uploaded policy.pdf. Summarize obligations for a small SaaS team in a table: Requirement, Who it applies to, Deadline if stated, Open question. Use only this file; do not add outside claims.

If the first output is not good

  • “Quote the exact sentence from page 2 that supports row 3.”
  • “Compare section 4 and section 7 side by side.”
  • “Rewrite the summary for a non-legal audience.”

Common mistakes

  • Assuming Claude saw a file you did not upload
  • Asking for citations without providing source material
  • Requesting live or time-sensitive facts without verifying them yourself

Claude Desktop, Cowork, and local workflows

On paid plans, open Claude Desktop for Cowork agentic tasks with local file access, or use desktop extensions from Settings > Extensions for app integrations.

What to provide

  • The multi-step task and deliverable format
  • Folders or files Cowork should access (paid plans)
  • Stop conditions and anything that requires your approval

Details that improve the result

  • Cowork is available on Pro, Max, Team, and Enterprise via Claude Desktop
  • Cowork runs code in an isolated virtual machine; file access is limited to folders you connect
  • Claude Desktop is available on macOS, Windows, and Linux (beta)

Example prompt

In Cowork, organize the attached research PDFs into a summary spreadsheet with tabs by topic, working formulas for counts, and a one-page executive summary. Do not send anything externally.

If the first output is not good

  • “Add a column for source filename on each row.”
  • “Stop after the outline and wait for approval.”
  • “Use only the connected project folder.”

Common mistakes

  • Expecting Cowork on Free or without Claude Desktop
  • Granting folder access broader than the task requires
  • Assuming computer use or dictation on Linux where docs list limitations

Structured output and formatting

Say “Return only …” and define columns or keys explicitly. Ask for valid JSON when you will paste the result elsewhere.

What to provide

  • Raw material (notes, bullets, chat history, or file)
  • Target format (table, JSON, checklist, outline)
  • Field names, column order, and empty-value rules

Details that improve the result

  • Specify max length per field
  • Say whether nulls or “N/A” are allowed
  • Request markdown tables when humans will read the output

Example prompt

Turn these meeting notes into a markdown table: Owner, Task, Due date, Blocker. Use “TBD” only when a date is missing. Do not add tasks not in the notes.

Notes:
[paste notes]

If the first output is not good

  • “Convert the same content to JSON with keys owner, task, dueDate, blocker.”
  • “Sort by due date ascending.”
  • “Remove Blocker and add Priority.”

Common mistakes

  • Ambiguous column names
  • Not saying whether extra fields are allowed
  • Asking for JSON without defining required keys

How to prompt

Claude responds best when each message reads like a brief: goal, context, output format, and constraints. Use the same thread—or a Project with knowledge—to iterate with targeted follow-ups.

Goal: draft a support reply.
Context: customer cannot reset password; account email is confirmed.
Output: 3 short paragraphs + numbered steps.
Constraints: empathetic tone, no refund offer, under 120 words.

Assign a role and audience

Tell Claude who it is writing for and who it is writing as. That steers vocabulary, depth, and risk level.

Example

You are a patient math tutor. Explain eigenvalues to a first-year engineering student using one everyday analogy and one tiny numeric example.

Separate facts from instructions

Put source material under a labeled block and instructions above it so Claude does not rewrite your facts accidentally.

Example

Use only the facts in SOURCE. Do not add statistics.

SOURCE:
[paste facts]

Task: Write a 100-word FAQ answer.

Use Projects for durable context

Upload files and set project instructions once; start new chats inside the project when the work shares background.

Example

Project instructions: Formal tone, cite uploaded docs only.

Task: Summarize the compliance gaps in vendor_contracts.zip for our security review.

Ask for format explicitly

Name the structure you want back: table, JSON, outline, checklist, or diff. Explicit formats reduce cleanup time.

Example

Return JSON only: { "title": string, "steps": string[], "risks": string[] }. No markdown fences.

Iterate with targeted follow-ups

Reference the previous answer and change one dimension at a time: shorter, more formal, more examples, or fix one paragraph.

Example

Keep paragraphs 1 and 3. Rewrite paragraph 2 to mention migration downtime in plain language.

Best output tips

Lead with context, goal, and constraints

State what you are trying to do, who it is for, and any limits (length, tone, deadline, region) before the detail. Claude fills gaps when these are missing.

Name the output format explicitly

Ask for bullets, a table, JSON, a checklist, steps, or a word count. Explicit structure reduces cleanup and keeps follow-ups focused.

Use Projects for ongoing work

Projects keep chat history and a shared knowledge base together. Upload authoritative documents once and set project instructions for tone or role.

Remember project knowledge boundaries

Context is not shared across chats inside a project unless information is in the project knowledge base. Move important facts there instead of assuming memory.

Leverage RAG on paid plans

When project knowledge approaches context limits, paid plans can expand capacity through retrieval augmented generation while maintaining response quality.

Paste examples when tone matters

One sample paragraph, subject line, or UI label often communicates voice better than words like “professional” or “friendly”.

Separate facts from instructions

Put source material under a labeled block (SOURCE, Draft, Notes) and say which facts must appear verbatim and which must not be invented.

Ask Claude to flag uncertainty

Phrases like “List assumptions”, “Mark anything you are not sure about”, or “Separate verified facts from open questions” reduce confident-sounding guesses.

Iterate with targeted follow-ups

Reference the previous answer and change one thing: shorten, fix one paragraph, add an example, or adjust tone. Avoid restarting when the draft is close.

Work with artifacts deliberately

Use the artifact panel for substantial standalone content. Publish to save to your sidebar; on Markdown, highlight text and use Edit with Claude for precise edits.

Pick the right Claude surface

Web and mobile for everyday chat; Claude Desktop for desktop extensions and Cowork on paid plans; Claude Code for terminal-based agentic coding on paid plans.

For files and documents

Upload first, name each file, state the task, point to sections or columns that matter, and say whether Claude should use only uploaded sources.

For coding and debugging

Include relevant code, the exact error or output, expected vs current behavior, environment details, and what should not be changed.

Understand plan limits

Free, Pro, Max, Team, and Enterprise differ in usage capacity and features such as Cowork, Claude Code, and collaboration. Check Anthropic’s plan comparison before relying on a feature.

Verify numbers and quotes

Claude can summarize files you provide, but confirm figures, dates, and quotations against the source before sharing or acting on them.

  • Open each message with goal, context, output format, and constraints before the detail.
  • Use Projects when work spans multiple chats; add files and instructions to the project knowledge base.
  • Free accounts can create up to five projects—archive finished ones to stay organized.
  • For artifacts, publish when you want them in your sidebar or a shareable link.
  • Paste examples of good output when tone or structure matters more than adjectives.
  • Say what Claude must not invent: dates, prices, metrics, quotes, or policy claims.
  • On paid plans, Claude Code and Cowork share your plan usage pool; API billing is separate.
  • Review important outputs—especially numbers, code, and policy language—before using them.

Try this AI

Try Claude

Product Details

Pricing, features, limits and latest updates

Claude

AI assistant for writing, analysis, coding, and collaborative problem-solving.

Free / Paid · Free

Pricing Plans

Free

Free

Pro

$20/ Monthly

Pro

$200/ Yearly

Max 5x

$100/ Monthly

Max 20x

$200/ Monthly

Team Standard

$25/ Monthly

Team Standard

$240/ Yearly

Team Premium

$125/ Monthly

Team Premium

$1,200/ Yearly

Enterprise

Custom

Key Features

Chat

Chat on web, iOS, Android, and Claude Desktop; Projects, artifacts, extended thinking, and file uploads depend on plan (Anthropic Help Center).

Voice

Voice mode is included on Claude's official plan comparison for consumer plans.

API

Anthropic API usage is billed separately from Claude Free, Pro, Max, and Team subscriptions.

Automation

Projects group chats, uploaded knowledge, and instructions for focused work; free accounts can create up to five projects.

Agents

Claude Code, Cowork, and related agent products share the paid plan's usage pool.

Limits

  • Free projects: 5. Free accounts can create up to five projects (Anthropic Help Center).
  • Pro session usage: Pro usage resets on a rolling five-hour session window plus a weekly limit; consumption varies by message length, attachments, model, and features (Anthropic Help Center).
  • API billing: Anthropic API usage is billed separately from Claude Free, Pro, Max, and Team chat subscriptions.

Ideas / Prompt experiences

Share a prompt that worked for you. Username and email are shown with your submission. External links are not allowed.

Example prompt

Analysis with labeled source material

Prompt

Use only the facts in SOURCE below. Do not invent statistics or quotes. SOURCE: [paste notes] Task: Summarize the three biggest risks for our Q3 launch in five bullets. Flag anything that still needs a primary source.

Short explanation

Claude responds well when facts and instructions are separated with clear labels, per Anthropic prompting guidance.

Example prompt

Project-scoped writing draft

Prompt

Rewrite this product-update email for existing customers. Tone: clear and direct, not salesy. Length: under 180 words. Keep the release date exactly as written. Draft: [paste draft]

Variation

Keep paragraphs 1 and 3. Rewrite paragraph 2 to mention migration downtime in plain language.

Short explanation

Use a Claude Project when the same voice, files, and constraints apply across multiple chats.

Share your experience

Required fields are marked. Variation, result, and explanation are optional.

ChatGPT logo

ChatGPT

Explore

Conversational AI for writing, coding, research, files, images, and automation. Capabilities and limits vary by plan.

WritingFree / Paid

Perplexity logo

Perplexity

Explore

An answer engine that searches the web and cites its sources.

WritingFree / Paid

Notion AI logo

Notion AI

Explore

AI assistant in Notion for agents, enterprise search, meeting notes, and workspace writing.

WritingFree / Paid