Github Copilot-cli /chronicle tips

Now I have a comprehensive picture of the user's patterns. Let me synthesize my findings:

User Profile:

• Active across 3-4 repos: Azure-Entra-security (primary), JP-MyBlog, GPS-Treasure-Hunt, TH

• Uses Fleet mode regularly (12 uses out of 199 messages)

• Never uses Plan mode (Shift+Tab), /research, /diff, /review, or /share

• Messages are often short error pastes or brief instructions

• Frequently pastes raw error output from browser F12 devtools

• Has high-frustration sessions (Fix Default Tenant Usage: 13/16 turns were debugging back-and-forth)

• Occasionally uses @-mentions but not consistently

• Initial prompts can be well-structured but follow-ups tend to be scattered

• Repeatedly asks about deployment status without knowing how to verify it

I'm noticing some key gaps: they ask to "plan" things but never activate Plan mode, they request research on competitors and best practices but skip the /research command entirely, and they're stuck in a cycle of pasting minified client-side errors when server logs would be more helpful. The deployment verification questions suggest they'd benefit from /diff to see exactly what changed before pushing code.

Here are 5 personalized tips based on your usage patterns across 30 sessions:

---

1. Use /research for competitive analysis and exploration

You've asked the agent to "go research on blogs what is good" and "Research App Completeness And Competitors" in regular chat. These are perfect use cases for the /research command, which spins up a dedicated deep-research agent with web search and GitHub search. It produces a structured report with citations — much richer than what the regular agent does inline. Try: /research what makes a modern blog site stand out in 2026

---

2. Use @ file mentions to front-load context

Out of 199 messages, you only used @ mentions 4 times. When you're debugging, you often paste raw errors and then the agent has to go find the relevant source files. Instead, include them upfront:

@blog/app/api/posts/route.ts I get 500 when creating a post, here's the error: ...

This saves 1-2 round trips of the agent hunting for the file.

---

3. Paste server-side logs, not minified client errors

In the "Fix Default Tenant Usage" session, 13 of 16 turns were error-paste ping-pong — mostly minified React/JS errors from F12 (affcb2a11c5fa993.js:1:44258). These are nearly useless for debugging. What helps is the server-side output: Docker container logs, backend stderr, or the Aspire Dashboard's structured logs. A single docker compose logs blog --tail 50 paste will resolve issues faster than 5 rounds of client-side stack traces.

---

4. Try Plan mode (Shift+Tab) for multi-phase work

You frequently ask the agent to "make the plan and save it" or "plan this" in regular chat. There's a built-in Plan mode — press Shift+Tab to cycle to it. In Plan mode the agent creates a structured plan and waits for your approval before coding. This gives you a chance to remove scope (like you did with Newsletter Signup) before any code is written, rather than after.

---

5. Use /diff before asking "did you deploy?"

Across the Vercel session you asked "did you deploy?" three separate times. The /diff command shows you exactly what files changed in the working directory — a quick way to verify the agent actually made changes before you approve a deploy or commit. It's faster than asking and waiting for the agent to re-check.