Daily Dose of GHCP
Bite-sized GitHub Copilot features, tips, and discoveries — published daily.
All Posts
Spec-Driven Development with GitHub Copilot and Spec Kit
Without a shared source of truth, requirements drift and AI-generated code misses the mark.
GitHub Spec Kit gives GitHub Copilot the structure it needs — constitution, spec, plan, tasks —
to generate production-ready code that stays aligned with intent from day one.
Read more →
Integrate MCP with Copilot
AI tools need a standard way to act on your behalf, not just reply.
By adopting Model Context Protocol, GitHub Copilot's agent mode discovers
tools, selects the right one, and executes it — giving you an AI that
operates, not only answers.
Read more →
Subagents: How Copilot Delegates Work in VS Code
When the main agent's context fills up, quality drops. Subagents fix this by running
focused subtasks in isolated context windows and returning only a summary.
Learn the four key properties, usage patterns, custom agent orchestration, and the
coordinator-worker pattern.
Read more →
GitHub Agentic Workflows: CI/CD Meets AI
Write GitHub Actions automations in plain Markdown and let an AI coding agent execute them.
Explore the two-file model, the Continuous AI pattern, Safe Outputs security,
and when agentic workflows augment — not replace — traditional CI/CD.
Read more →
Copilot Customization in VS Code and CLI
A simple guide to the four customization concepts that matter most: custom instructions, prompt files,
custom agents, and agent skills. Includes beginner-friendly examples and a clear VS Code vs CLI comparison.
Read more →
Autopilot in IDE and CLI
Autopilot sounds similar across surfaces, but it behaves differently in VS Code and Copilot CLI.
This guide compares Autopilot vs Agent mode in IDE, then breaks down CLI Autopilot vs
Read more →
12 March 2026
💬 Copilot in IDE
🌐 Beginners
--yolo/--allow-all and --no-ask-user with a practical decision table.
GitHub Copilot for Azure
GitHub Copilot for Azure brings Azure-aware prompts, tool calls, and live cloud context into your IDE.
Learn where it runs, what tasks it is built for, and the prompt habits that help it choose Azure MCP tools instead of giving generic cloud advice.
Read more →
Understanding Copilot's Internal Architecture
What actually happens between you hitting Enter and the first token appearing?
Explore the three-tier architecture, the Copilot Proxy's role, how prompts are assembled,
model selection with Auto mode, tool-call ReAct loops, and the two content-safety checkpoints.
Read more →
Best Practices for Running Multiple GHCP Agents in One Repository
Running 10+ Copilot agents concurrently leads to merge conflicts, stale branches, and PR collisions.
Learn a five-layer architecture — Planner Agent, domain ownership, PR queue, worktrees, and small PRs —
that reduces conflicts by up to 80%.
Read more →
Creating GitHub Issues with Copilot
Stop filing issues by hand. Paste a plain-language description or a screenshot into
Copilot Chat at github.com, and Copilot drafts a structured issue — title, body, labels,
and assignees — ready for your review. Works for single tasks, whole to-do lists, and full epic trees.
Read more →
Managing Context for AI in Copilot Chat
The quality of Copilot's answers depends entirely on what context it receives.
Learn how to use #-mentions, @-mentions, codebase search, #fetch, vision input,
browser elements, and context compaction to keep Copilot on target.
Read more →
Copilot CLI System Architecture Deep Dive
The CLI is a layered, modular agentic system with six primary layers.
Understand how input routing, core processing, tool execution, and state management
collaborate to enable complex, multi-step terminal-based AI assistance.
Read more →
Compaction in IDE: How Context Stays Useful
A continuation of the CLI and SDK session story, this post focuses on IDE compaction and why it keeps long chats productive.
Learn when automatic and manual compaction help, where to use
Read more →
/compact, and how to keep multi-turn threads clean without losing key decisions.
Compaction and Infinite Sessions in CLI and SDK
Copilot CLI can run long conversations without context-limit anxiety.
Learn how compaction/checkpoints work in CLI and how Copilot SDK persists,
resumes, and tunes infinite sessions with configurable thresholds.
Read more →
Tracking Coding Agent Sessions
Copilot coding agent sessions run autonomously in the background — and GitHub gives you
seven surfaces to watch, steer, and stop them. Learn how to use the agents tab, GitHub CLI,
Raycast, VS Code, JetBrains, Eclipse, and GitHub Mobile to stay in control.
Read more →
Inside GHCP: How the Execution Engine Works
The LLM is not at the bottom of a linear pipeline — it sits at the centre.
Tool calls and context compaction are feedback loops driven by the model itself.
Learn the accurate mental model for how Copilot's execution engine operates.
Read more →
GHCP Across Multiple Surfaces
GitHub Copilot doesn't live in one place. Understand the four surfaces — GitHub.com, IDE, CLI,
and SDK — and when to reach for each one to get the most from Copilot across your entire workflow.
Read more →
Context Window in GHCP Part 2
Part 1 explained the mechanics — now here's what to actually do with that knowledge.
Eight actionable practices covering inline completions, Chat context references,
Read more →
copilot-instructions.md hygiene, CLI session scoping, Coding Agent issue design,
model selection, and Copilot Spaces.
Context Window in GHCP Part 1
What is a context window, why does it matter even more for agents, and how does GitHub Copilot
actually manage it? Explore model limits vs. effective context, how sources are selected and
filtered, and practical tips for working with large codebases.
Read more →