> ## Documentation Index > Fetch the complete documentation index at: https://docs.replit.com/llms.txt > Use this file to discover all available pages before exploring further. # August 21, 2026 > 3 min read ## What's new * [Work in Free Mode](#work-in-free-mode) * [Start with Conversations](#start-with-conversations) * [Automate recurring work with Routines](#automate-recurring-work-with-routines) * [Steer Agent while working](#steer-agent-while-working) * [Share and import Skills](#share-and-import-skills) * [Control Agent models across Workspaces](#control-agent-models-across-workspaces) * [Run black-box pen tests](#run-black-box-pen-tests) ## Agent ### Work in Free Mode Use Free Mode to chat, explore ideas, create, and build without using Power Mode or Max Mode credits. Free Mode uses Auto to select a model, and its limits depend on your plan. Check **Settings** > **Usage** to see your remaining allowance and reset timing. Learn how to [get the most out of Free Mode](/chat/free-mode). ### Start with Conversations Start with a private Conversation to ask questions, explore information, attach context, run a Routine, or describe an outcome. When the work becomes substantial, ask Replit to turn the Conversation into a Project. Your context and files carry forward. Sanitized Replit Workspace home with Projects, Conversations, Routines, suggested prompts, and the Conversation message box Learn more about [Conversations](/chat/conversations). ### Automate recurring work with Routines Schedule Replit to repeat work inside a Conversation. Each result returns to the same Conversation, where you can review it and continue working. Routines run in Power Mode or Max Mode and require a budget for each run. Routines page listing an active daily calendar summary on Slack and its schedule Learn more about [Routines](/chat/routines). ### Steer Agent while working Send a follow-up while Agent works to add new context to the active turn. Steer is the default in Conversations and Projects, and you can change either preference to Queue in **Settings** > **Personalization**. Press `Cmd+Enter` on Mac or `Ctrl+Enter` on Windows and Linux to use the opposite behavior for one message. Agent settings showing Conversation and Project send behavior with Steer and Queue options Learn more about [steering and queuing messages](/features/agent/steer-and-queue-messages). ### Import Skills from GitHub Import Skills from a public GitHub repository, folder, or file URL, then preview the Skill files before you add them. Saved Project Skills also support member access levels of **Required**, **Available**, or **No access**. Agent Workspace Skills picker listing ask-matt, claude-handoff, and code-review Learn more about [sharing and importing Agent Skills](/features/agent/skills#share-project-skills). ## Teams and Enterprise ### Control Agent models across Workspaces Enterprise account admins can set an account policy for each model provider, then choose which models are available in individual Workspaces. Use **Enable all**, **Enable selected**, or **Disabled** to match your organization's policy. Agent modes and model providers settings with provider policies and the Fast mode models control Learn more about [Enterprise model controls](/features/collaboration/enterprise-model-controls). ## Platform ### Run black-box pen tests Run a Level 3 security scan to combine three checks. It runs free dependency and package checks, Agent static analysis of your source code, and an external black-box test of the live Preview. Replit runs the source-code review and the outside test in parallel. Security and Privacy Center with the Select scan level menu expanded and Level 3 AI pentest selected Learn more about [black-box pen tests](/features/security/black-box-pen-tests).