GitHub Copilot CLI and the Copilot SDK now support per-session AI credit limits, giving developers a practical guardrail for long-running agent tasks and unattended automation. What it does GitHub added AI credit session limits for Copilot CLI and the Copilot SDK. The practical value is simple: before you hand an agent a long task, you can set a credit cap so the session stops cleanly instead of running until the work is done or someone notices the bill. This belongs in a developer's automation checklist because AI coding agents are increasingly used for unattended refactors, test runs, investigations, and SDK-driven workflows. A per-session cap does not replace account budgets, but it gives each run a local spending boundary. How to try it Update GitHub Copilot CLI to version 1.0.66 or later, then use the new session-limit controls before starting work. For interactive CLI sessions, GitHub's docs show the /limits set command: For non-interactive CLI jobs, pass a maximum credit value on the command line: GitHub notes that the limit is a soft cap: a model response already in progress can finish, so final usage may slightly exceed the number you set. Their docs also advise that session limits work best above 30 AI credits because many model calls can cost more than 20 credits. Best use cases Budgeting unattended Copilot CLI jobs in CI-like local automation. Running exploratory codebase tasks without leaving an agent unbounded. Testing prompt cost before scaling a workflow across a team. Giving SDK-based agent features a per-run guardrail. Caveats This is a public preview feature, so behavior may change. It also controls a single session only; teams still need organization budgets, billing alerts, model-selection policies, and human review for expensive agent workflows. Treat the first few runs as calibration. Start with small, reversible tasks, check actual AI credit use afterward, and adjust the limit based on prompt size, repository size, selected model, and tool-call depth. Source links GitHub Changelog: https://github.blog/changelog/2026-07-01-set-ai-credit-session-limits-in-copilot-cli-and-sdk/ GitHub Docs: https://docs.github.com/en/copilot/tutorials/optimize-ai-usage GitHub CLI best practices: https://docs.github.com/copilot/how-tos/copilot-cli/cli-best-practices Session limit setup docs: https://docs.github.com/en/copilot/how-tos/copilot-cli/use-copilot-cli/set-session-limit

LinkLoot access: Free
Provider costs: Unknown

Review open

@ZachasADMIN

Open loot

Use VS Code browser tools to make Copilot test the UI it builds

No votes yet

#VS Code #GitHub Copilot+4

GitHub Copilot's browser tools in VS Code are now generally available, giving agents a real browser surface for building, clicking, inspecting, and checking web-app flows inside the editor. GitHub Copilot in VS Code can now use generally available browser tools, which makes it more useful for frontend and full-stack work where a text-only code edit is not enough. The practical move is simple: ask the agent to build or change a UI, then make it open the app, click through the flow, inspect the page, and fix what it finds. What it does The browser tools let a VS Code agent work with a real browser context while staying inside the editor workflow. The official guide shows the pattern: enable browser tools in the chat tools picker, ask the agent to build a small web app, start a local development server, open the page, interact with it, inspect console feedback, and iterate. This is strongest for UI work where the failure is visual or interactive: broken buttons, missing state updates, layout regressions, console errors, routing bugs, and flows that only make sense after a click. Practical workflow Where it helps most Small frontend features where a browser click-through catches obvious misses. Bug fixes involving forms, validation, routing, modals, auth walls, or loading states. PR review prep where the agent should produce evidence beyond a diff summary. Design-system changes that need quick spot checks across a few states. Documentation examples where the rendered page should match the written instructions. Caveats Keep the scope tight. Browser control does not replace Playwright, Cypress, accessibility tooling, or human design review. It is best as a fast interactive sanity check before formal tests. Treat the inspected page as untrusted input when it feeds back into an agent. A hostile page can display instructions, fake success messages, or misleading labels. The safe pattern is to ask the agent to report observable browser facts and test results, not obey instructions found inside the page. Source links GitHub changelog: https://github.blog/changelog/2026-07-01-browser-tools-for-github-copilot-in-vs-code-are-generally-available/ VS Code browser-agent testing guide: https://code.visualstudio.com/docs/agents/guides/browser-agent-testing-guide VS Code chat tools documentation: https://code.visualstudio.com/docs/chat/chat-tools

LinkLoot access: Free
Provider costs: Unknown

Review open

@ZachasADMIN

Open loot

Blog

Verwandte Artikel

Blog durchsuchen

AI & Automation

#github-copilot

Mehr aus diesem Thema

Cap GitHub Copilot CLI agent spend with AI credit session limits

Use VS Code browser tools to make Copilot test the UI it builds

Verwandte Artikel

Gemini 4 Argon goes GA in GitHub Copilot for paid plans at provider list pricing

Anthropic launches Claude Sonnet 5.5 with faster coding and Copilot access

GPT-6 Sol and Luna bring two new OpenAI reasoning tiers to the API

GitHub Copilot app adds centrally managed OpenTelemetry export

GitHub Copilot adds cost-aware model routing and agent workflows

GitHub Copilot sets September 28 defaults for agents, reviews, and billing

GPT-6 Astra reaches GitHub Copilot for long-horizon coding

Claude Fable 5.1 reaches GitHub Copilot with retention caveats

GitHub Copilot adds Gemini 3.8 Flash with usage-based billing

GitHub reopens Copilot Business signups with upfront billing

GitHub Copilot approvals turn code review into a merge-control decision

Grok 4.6 rolls into GitHub Copilot for longer coding tasks

Agent Plugins 1.0 reaches Copilot as vendors standardize agent extensions

Microsoft brings MAI-Code-1.1-Flash to Copilot at a lower token price

GitHub puts Copilot code review behind a default firewall

GitHub adds repository-level Copilot metrics for coding agents

Check Copilot policies before GPT-5.6 agents start spending AI credits

Cap Copilot agent spend per user from GitHub's billing UI

Run GitHub Copilot agents on desktop without a paid Copilot plan

Cap GitHub Copilot credit pools before one team burns the budget

Move Copilot workflows off Gemini 2.5 Pro and Gemini 3 Flash before July 31

GitHub Opens Copilot Agent Session Streaming for Enterprise Audit Trails

Try Kimi K2.7 Code in Copilot after your admin reviews open-weight risk

Run Copilot CLI in Actions without storing a PAT

Stream GitHub Copilot agent sessions before audit gaps grow

GitHub Copilot Vision is GA: what developers can attach now

Cap Copilot agent spend before long CLI and SDK runs

GitHub Copilot browser tools are GA in VS Code