Questions? Call 1.800.716.8955

Claude Certified Associate: Foundations

Feature field guide

This guide collects practical know-how for each Claude feature the exam guide names. It describes what each feature is for and how to use it well, not exact limits or menu locations, which change often. Check Anthropic's help center for current details before relying on a specific setting.

Chat

What it is. A single conversation with Claude. Everything you type, paste or upload in that conversation is in context. Use it well.

Start a new chat for each new, unrelated task.

Front-load context: who you are, what you need and who it is for.

Upload files rather than pasting very long text when the content is a document. Edit and resend an earlier message to take the conversation in a new direction, instead of piling corrections on top.

Watch out for. Long, multi-topic chats that dilute focus; assuming Claude remembers other chats when memory is not enabled for that context.

Projects

What it is. A workspace with its own instructions and knowledge, where every chat inherits that context. Team and Enterprise plans can share Projects.

Use it well.

One Project per client, product, process or recurring deliverable.

Concise instructions; detailed reference material in knowledge.

Clear file names and one current version of each source.

A named owner and a review rhythm.

Watch out for. Stale files, contradictory instructions, oversharing sensitive files with a broad team.

Artifacts

What it is. Substantial, standalone output (documents, code, diagrams, simple interactive tools or visualizations) shown in a dedicated panel where you can view, iterate and share it. Use it well.

Ask for an artifact when the output is long, will be edited several times, or will be reused or shared.

Iterate in place: "In the artifact, shorten section 2 and add a summary table." Use simple interactive artifacts for internal tools such as calculators, checklists or quiz pages, and test them before sharing.

Watch out for. Sharing an artifact containing confidential content more widely than intended; treating a quick prototype as production software.

Research mode and web search

What they are. Web search lets Claude look up current information within a chat. Research mode runs a deeper, multi-step investigation across many sources and returns a cited report. Use them well.

Define scope: topic, date range, regions and preferred source types.

Ask for links and dates for each key claim.

Ask Claude to separate well-supported findings from weakly supported ones.

Open and check the sources behind any claim you will rely on.

Watch out for. Outdated or low-quality sources, citations that do not actually support the claim, and paywalled sources you cannot check.

File uploads and document analysis

What it is. Claude can read uploaded documents, spreadsheets, presentations and images, and answer questions about them.

Use it well.

Upload clean, text-based files where possible.

Tell Claude which file and section to focus on.

Ask for quotes and page or section references so you can verify.

For very large documents, share the relevant sections.

Watch out for. Poor scans, complex tables and charts that may be misread, and uploading data your policy does not allow.

Appendix d. feature field guide

Code execution and file creation

What it is. Claude can write and run code to analyze data, perform exact calculations and create files such as spreadsheets, documents, slide decks and charts. You do not need to read or write code yourself.

Use it well.

Say "use code" for any calculation that matters.

Ask Claude to show the calculation steps or summary statistics so you can sanity-check results. Ask for a downloadable file in the format your audience uses.

Spot-check a few results against the source system.

Watch out for. Messy source data such as duplicates, text-formatted numbers or hidden rows; trusting a chart without checking the underlying figures.

Connectors

What they are. Integrations that let Claude read, and in some cases act on, content in other tools such as Google Drive, Gmail and Google Calendar, plus many others built on the Model Context Protocol (MCP).

Use them well.

Enable only what a task needs, and only connectors your organization allows.

Prefer connectors for content that changes often.

Ask Claude to draft rather than send or change things, and review before acting. Disconnect services you no longer need.

Watch out for. Instructions embedded in fetched emails or documents (prompt injection); broader access than you intended; acting on stale or incorrect content without checking.

Memory

What it is. Claude can remember useful context and preferences across chats. You can view, edit and delete what it remembers. Projects keep their own separate memory.

Use it well.

Let memory hold durable preferences such as format and style.

Review what Claude remembers periodically and remove anything outdated or sensitive. Use incognito chats for content that should not be remembered.

Watch out for. Sensitive details persisting; outdated preferences affecting new work.

Skills

What they are. Folders of instructions, templates, scripts and resources that teach Claude how to perform a specific task, such as applying brand guidelines or building a standard report. Claude loads a Skill when the task calls for it. Organizations can provide Skills to their teams. Use them well.

Use Skills for procedures repeated across many contexts.

Keep each Skill focused on one task, with clear instructions and examples.

Have an owner maintain organizational Skills.

Watch out for. Outdated templates inside Skills; overlapping Skills that give conflicting guidance.

Extended thinking

What it is. A mode in which Claude reasons through a problem step by step before answering. Use it well. Turn it on for complex analysis, planning, multi-constraint problems and tricky judgment calls.

Watch out for. Using it on simple tasks, where it adds time and usage without improving results.

Incognito chats

What it is. A chat that is not saved to your history or used for memory.

Use it well. For sensitive or temporary discussions, on a tool your policy approves. Watch out for. Treating incognito as permission to share data your policy forbids. It changes what is saved, not what is allowed.

Admin and plan controls (awareness only)

On business plans, administrators can manage members, enable or restrict features and connectors, and apply security and data settings. As an associate, you do not need to configure these, but you should know they exist, follow them, and ask your administrator when a feature you need is restricted.

Appendix d. feature field guide

Use ← → to move between sections