AI product Open source

Claude Code

Claude Code is Anthropic's agentic AI coding tool for the terminal, IDEs, and GitHub. It uses natural-language commands to examine and understand codebases, create and read files, execute commands, run tests, explain code, manage Git workflows, and assist with routine development tasks. It can load persistent project context, run custom slash commands, use plugins with custom commands and agents, and operate with configurable autonomy while leaving actions such as final pull-request merging to a human. The official repository documents installation for macOS, Linux, and Windows, and identifies npm installation as deprecated.

View repository Visit site Mentioned in 113 videos ↓

What Claude Code is used for

99 uses taken from transcripts — each links to the moment in the video.

  • Local agent harness used to run observe-think-act loops, read project files, use tools, and produce finished outputs.

  • Plans and directs the website-scraping, classification, and data-enrichment workflow through prompts and Apify integration.

  • Used as the skill-based workflow for turning existing knowledge into a digital product.

  • Used to understand the codebase, estimate rough levels of effort, inspect product logic, and investigate how the product behaves.

  • Performed SEO and blog-automation work and helped build websites and applications.

  • Vibe-codes browser-based linkable tools, demonstrated by creating an inventory calculator.

  • Executes described software ideas and supports coding and research workflows.

  • Provides a coding-agent environment used alongside independent agent labs.

  • Runs reusable skills, creates content, builds the product landing page, and coordinates connected workflows.

  • An intermediate implementation option in which graph steps write files and leave a paper trail.

  • Helps build personalized applications and automate workflows, including the grant-application process.

  • Coding agent used as an example of planning software tasks, navigating repositories, editing files, executing commands, and iterating from feedback.

  • An AI coding assistant used to reduce cognitive load and automatically apply a team's coding style.

  • Coding agent included in the agent stack used alongside Paper.

  • AI coding agent used in the workflow that TDD Guard initially supported.

  • Agent harness used to review Codex's work, research contracts and previous agreements, critique thumbnails, and potentially verify recurring tasks.

  • AI coding assistant used to build and modify the complex clipping application.

  • Used to help code and SEO-optimize a directory, with the creator checking for SEO elements the AI missed.

  • Agentic coding harness used to access codebases, run tools, execute long-running tasks and orchestrate other agents.

  • AI-assisted coding and document workflows; used with reversions, version checks, and verification loops after generated changes.

  • Example of using a Markdown file named Claude.md as persistent semantic memory containing project architecture, coding conventions, build commands, frameworks, and prohibited practices.

  • Coding engine placed under IBM's software-development harness.

  • Runs agent skills and their executable scripts.

  • Runs the explainer-video workflow: topic selection, fact verification, scriptwriting, voiceover generation requests, scene planning, visual specification, and project organization.

  • Used in a terminal-based workflow to build the marketing OS, process marketing tasks, and maintain feedback loops.

  • A coding agent that, according to the discussion, was tricked by the ghostjacking technique in nine out of ten tests.

  • Provides local coding-agent logs for usage and task auditing.

  • Provides AI coding sessions that Wallfacer indexes and makes searchable.

  • Can receive and answer requests from the Penako canvas.

  • Coding agent used in the shared agent workspace.

  • Runs the Specify workflow and implements AI-generated development tasks.

  • Daily AI agent harness for answering questions, analyzing documents, scripting videos, completing small coding projects, and handling connected personal tasks.

  • An AI coding agent used by Apple at large scale. The video mentions it as an example of how enterprises are already using frontier-model-based coding tools.

  • Runs the coding agent in bypass-permissions mode, triages repository issues, implements fixes, verifies changes, and creates pull requests.

  • An autonomous coding agent used to build software from intent-driven prompts, run commands, test projects, manage Git repositories, and create draft pull requests. The video shows it running on Google Cloud through Vertex AI.

  • AI coding agent used to install Ponytail locally and generate and compare weather dashboard applications with and without the plugin.

  • It is described as an AI agent interface that developers may use instead of a CLI, web UI, or platform portal. Viktor also uses it as a way to interact with custom agents and says it is preferable to simple custom-agent harnesses.

  • Provides the agent CLI used for the orchestrator, coder, and release profiles.

  • Runs as a terminal-based coding agent in the persistent sessions.

  • An AI coding and troubleshooting agent that can analyze problems, propose solutions, create Git branches and pull requests, and operate with configurable levels of autonomy while leaving final merging to a human.

  • Provided the skills and personal rules that Muse Code imported, and served as a comparison environment for coding-agent work.

  • A coding-oriented AI tool presented as being specialized for software engineering, including code planning and testing, with a terminal-based interface.

  • An AI coding tool that lets people create software without writing every line themselves.

  • A command-line agent used through natural language to create and read files, control the computer, load persistent project context, analyze an Obsidian vault, run custom slash commands, generate ideas, and build commands.

  • Anthropic's coding agent and development tool, frequently raised in enterprise software-development discussions. The video describes it as a model-locked offering whose users can switch to other tools.

  • Provides coding-agent functionality for memorax-code, refactoring-ui-skill, ask, fire-your-seo-agency, open-pstack, headcount, and lemalog workflows.

  • Coding-agent platform extended by plugins and used by several multi-agent tools.

  • An AI coding tool that helps engineers write code and build software. The speakers describe using it for coding projects and compare its reactive, knowledge-enhancing workflow with more autonomous systems.

  • Coding-agent environment used to run Fable and other models against real software projects.

  • An AI coding agent used by Speechify engineers to work on software tasks and orchestrate longer-running development workflows.

  • Provides the coding-agent behavior being audited and used through T3 Code, including skills, slash commands, SDK behavior, and multi-agent work.

  • Coding tool used by engineers in the enterprise usage example, where rate limits affected when work could be performed.

  • Agent harness used to run Fable, manage coding workflows, and apply skills during software development.

  • Runs the I-have-ADHD skill and provides the environment for Hyper Research.

  • Acts as the primary coding-agent harness for demonstrating agentic engineering workflows, including plan mode, sub-agents, loops, and code generation.

  • An agent used to investigate production incidents by querying server errors, inspecting request data, searching the codebase, analyzing logs, and tracing requests across systems. The speaker also uses it to prepare incident context, postmortems, and on-call handoffs.

  • Named as a coding-agent example intended to use Stack Overflow for Agents.

  • Coding harness whose configuration, agent files, compaction behavior, permissions, memory, and context settings were discussed and used with Fable.

  • Coding agent used to work across repositories, update a stale local clone, create a worktree, make fixes, file and verify a pull request, and generate front-end and game projects.

  • Anthropic's agentic coding tool for the terminal. It understands a codebase, handles natural-language coding tasks and Git workflows, and can also be used from an IDE or GitHub.

  • Andrew Ng describes Claude Code as his personal favorite AI coding tool, used to write software more quickly.

  • Implemented fixes after GPT 6.1 Soul reviewed code and identified problems in a T3 Code worktree setup flow.

  • Coding-agent sessions used for adversarial review workflows and demonstrated collaborating with other coding-agent instances.

  • Used for coding tasks where the user selects a model according to the required level of intelligence and token cost.

  • Used as an example of a coding agent that accelerates development while increasing code volume and complexity; also mentioned in rule-based troubleshooting automations.

  • Runs coding-agent tasks, including attempts to access browser history, bank information, and network resources.

  • Executes coding-agent tasks using either agentic search or a constrained vector-search implementation.

  • The coding-agent environment used in the workshop repository to run setup skills, voice-directed coding tasks, goals, loops, hooks, and scheduled tasks.

  • Diagnoses production alerts and investigates possible root causes.

  • Used to browse a conference website and extract schedule information while its harness handled tool-call results.

  • Reads failing traces and evaluation explanations, then proposes or applies prompt and application improvements.

  • Acts as the coding agent that adds Arize AX observability, retrieves and analyzes traces, identifies quality problems, and implements the price-filtering fix.

  • Used to create the presentation.

  • Coding agent selectable in Tessl Launch to execute automated workflows.

  • Terminal coding agent connected to the platform; creates a conversational space as owner and invites other agents to collaborate on an engineering task.

  • Anthropic's coding agent; the speaker's primary tool for shipping serious engineering work

  • Agent harness used to build the MVP MCP server; its ask-user-question tools are used in the rebuilt version

  • Generic coding agent used as the baseline comparison against the context-graph-connected agent in the evals

  • AI coding agent participants use to build projects like the guided admin support tool

  • Coding agent used to run the Haiku demos, sub-agent workflows and PR auditing.

  • Used by the speaker, who does not code, to build online apps that let workshop participants practice the leadership skills being taught.

  • Example of a coding agent running on laptops that hits all three lethal trifecta conditions

  • AI coding CLI whose auth is handled centrally by the gateway via tailnet identity.

  • Agent brain that called Bright Data's MCP: fetched a protected Glassdoor page successfully (vs blocked without MCP) and built a social media profile listener

  • Coding agent / MCP host that offloads sensitive document tasks to the private cluster

  • Coding agent alternative supported by the workflow

  • Coding harness example where prompt engineering with privacy in mind matters

  • AI coding tool pointed at the flash-examples repository so it learns the platform and helps build.

  • The coding agent used in the demo: investigates traces, diagnoses failures, edits the application code, builds datasets and runs experiments.

  • AI coding agent used to apply annotated changes and run verification skills.

  • A non-coder participant pointed Claude Code at the problem to build and optimize the CLI agentically.

  • Coding agent used to set up the mini Parameter Golf repo from its README, kick off the Weco run, and steer the search via a bridged session to the Weco dashboard.

  • Coding agent used to write and run Simulan automation code in the terminal

  • Coding agent typically run on the laptop as a pair programmer, or glued to GitHub Actions for scheduled and event-triggered runs; default harness in Agent Canvas local mode.

  • Coding agent attendees use for the hands-on portion to write code for a more complex financial agent.

  • Terminal agent recommended as the on-ramp for Claude Code users: a bundled Managed Agents skill builds a first agent and calls the API from natural language

  • Suggested as good at building an eval harness for you, given quality data

  • The coding agent driving the build — a workshop skill in the repo walks it through each act while it implements the chat UI, skill, subagents, custom tool and outcome handling.

  • Coding agent the speaker used heavily (max plans) over two years of building apps and workflows; the tool whose outputs he says converge on the median

Videos mentioning Claude Code

113 in the library.