Copilot Instructions
Project Overview
This is a personal website and blog (“Operating Intelligence”) built with Quarto, published to vidal.biz on Azure Static Web Apps via GitHub Actions.
Build & Preview
quarto preview # Local dev server with live reload
quarto render # Full site build to _site/Python 3.12 is used via direnv/pyenv (see .envrc). Set QUARTO_PYTHON=$(which python) if Quarto doesn’t find the venv Python.
Architecture
_quarto.yml— Site-wide config: navbar, footer, theme (quartz), metadata, Google Analytics, social cards.index.qmd— Landing page with author bio (uses thetrestlesabout template).posts.qmd— Blog listing page. Aggregates content from two sources:- Local posts in
posts/(.qmdor.mdfiles in date-prefixed subdirectories) - External posts defined in
posts/external.yaml(links to articles hosted elsewhere)
- Local posts in
talks.qmd— Talks listing page with inline metadata (no subdirectories).posts/_metadata.yml— Shared defaults for all posts:freeze: true, banner title blocks, TOC, KaTeX math, citations enabled._freeze/— Quarto freeze directory for cached computational output. Committed to git._extensions/— Quarto extensions:fontawesome,iconify,academicons.
Content Conventions
Adding a new local blog post
Create a directory posts/YYYY-MM-DD-slug/ containing an index.qmd (or index.md) file with YAML front matter:
---
title: "Post Title"
author: "Cedric Vidal"
date: "YYYY-MM-DD"
categories:
- Category
tags:
- tag1
image: "header.webp"
---Place images in the same directory as the post.
Adding an external blog post
Append an entry to posts/external.yaml:
- path: https://example.com/article
image: /posts/thumbnails/image.png
title: "Title"
description: "Description"
date: "YYYY-MM-DD"
categories:
- Category
tags:
- tagThumbnail images go in posts/thumbnails/.
Deployment
CI is in .github/workflows/publish-swa.yml. On push to main or PR, GitHub Actions renders with Quarto and deploys to Azure Static Web Apps. PR branches get preview deployments that are cleaned up on close.
Blog Positioning & Structure Guidelines
Purpose
This blog exists to build long-term authority in:
- Designing intelligent systems
- Evaluating agentic systems
- Operating AI in production
The goal is not volume. The goal is durable intellectual positioning.
This blog should compound credibility across: - Founder path - Big tech leadership - Consulting - Research-adjacent audiences
Core Principles
1. Source of Truth
- The canonical version of every post lives on vidal.biz
- Other platforms (LinkedIn, Substack, X, dev platforms) are distribution layers
- Content is written once, adapted for distribution
- Never write posts that only work if someone clicks away
2. Brand Positioning
The blog must signal:
- Systems thinking
- Architectural clarity
- Evaluation rigor
- Production realism
Avoid signaling:
- AI hype
- Feature evangelism
- Tool tutorials without framing
- News commentary without insight
The tone should feel:
- Calm
- Structured
- Technical but clear
- Durable
3. Creative Independence
When writing blog posts, articles, or any creative content:
- Do not look at other branches or PRs for inspiration
- Do not let other work-in-progress influence the content
- Each piece should reflect original thinking, not derivative of uncommitted work
- Do research the web for related blog posts or articles to use as references and grounding
This ensures authenticity and prevents cross-contamination of ideas across parallel efforts, while still grounding content in existing published work.
Content Pillars
The blog is structured around four pillars.
Each post must belong to exactly one primary category.
1. Infrastructure
Focus: - How intelligent systems are built and run - Pipelines, deployment, scaling - Distillation workflows - Fine-tuning infra - Cloud architecture - Cost tradeoffs - State, queues, retries, orchestration
Litmus test: > Is this about how to build or operate the system?
If yes → Infrastructure.
Distillation posts default here unless primarily about measurement.
2. Evaluation
Focus: - Measuring agent behavior - Reliability metrics - Propensity vs efficacy - Scenario design - Benchmark critique - Multi-turn evaluation - Failure analysis
Litmus test: > Is this about measuring or validating behavior?
If yes → Evaluation.
Evaluation is epistemology. Infrastructure is mechanics.
3. Systems
Focus: - Architectural reasoning - Component interaction - Feedback loops - State and memory design - Multi-agent coordination - Failure modes across layers - Control theory framing
Litmus test: > Is this about how intelligent components interact over time?
If yes → Systems.
Systems is about structure and emergent behavior.
4. Strategy
Focus: - Ecosystem shifts - Incentive changes - Vendor positioning - Founder implications - Enterprise implications - Market dynamics - Directional thinking
Litmus test: > Is this about what this means, not just how it works?
If yes → Strategy.
Strategy must be technically grounded. Avoid shallow commentary.
What Not To Do
Do not:
- Create new top-level categories for tools
- Create categories like “Azure”, “Distillation”, “LoRA”
- Publish generic AI summaries
- Publish reactive commentary without insight
- Publish purely tactical debugging posts
Every post must increase:
- Authority
- Coherence
- Positioning clarity
Homepage Structure
The homepage should communicate in under 10 seconds:
- Who this is for
- What intellectual domain is owned
- What lens is applied
Suggested framing:
Cedric Vidal Operating Intelligence
Writing about building, evaluating, and scaling intelligent systems in production.
Visual Guidelines
- Reduce visual noise
- Favor whitespace over gradients
- Minimize decorative AI artwork
- Typography over visual intensity
- Categories visible, tag spam hidden
The site should feel like: - A body of work - Not a content feed
Distribution Strategy
For each post:
- Publish canonical version on vidal.biz
- Adapt for:
- LinkedIn (native condensed version)
- X (thread)
- Substack (full or adapted)
- Link back to canonical source when appropriate
Goal: Maximum reach, preserved ownership.
Quality Filter
Before publishing, ask:
- Does this increase my authority?
- Does this clarify my thesis?
- Does this demonstrate depth?
- Would a founder or staff engineer learn something non-obvious?
If not, reconsider publishing.
Long-Term Goal
In 3 years, this blog should clearly signal:
“This person understands how intelligent systems are designed, evaluated, and operated at scale.”
Everything published should move toward that outcome.