clairvoyance-plugins

Clairvoyance

AI agents can write working code, but they don't stop to consider effective design unless asked. Clairvoyance is a set of skills inspired by John Ousterhout's A Philosophy of Software Design. Each skill gives your agent extrasensory perspective around software design, with concrete tests to see ahead of obstacles during implementation and review.

How It Works

Skills activate automatically and push your agent to ask questions like:

• Does this interface hide real complexity, or just pass things through?
• Can this method be understood without reading another one in a different file?
• Is error handling pushing work onto callers that the module could handle itself?

You can also invoke them directly. Use /red-flags to trigger a design smell scan, /deep-modules to check interface depth and /design-it-twice to compare alternatives before committing.

Installation

Note: Installation differs by platform. Claude Code has a built-in plugin marketplace. Codex and OpenCode require manual setup.

Claude Code (via Plugin Marketplace)

/plugin marketplace add codybrom/clairvoyance
/plugin install clairvoyance@clairvoyance-plugins

skills.sh

npx skills add codybrom/clairvoyance --skill '*'

Codex

Tell Codex:

Fetch and follow instructions from https://raw.githubusercontent.com/codybrom/clairvoyance/refs/heads/main/.codex/INSTALL.md

See .codex/INSTALL.md for detailed steps.

OpenCode

Tell OpenCode:

Fetch and follow instructions from https://raw.githubusercontent.com/codybrom/clairvoyance/refs/heads/main/.opencode/INSTALL.md

See .opencode/INSTALL.md for detailed steps.

llms.txt

Machine-readable skill index for LLM agents:

• clairvoyance.fyi/llms.txt — table of contents with descriptions
• clairvoyance.fyi/llms-full.txt — full content of all skills

What's Inside

Structure & Modules

Skill	Covers
deep-modules	Module depth, shallow modules, classitis, pass-through methods, interface vs implementation
module-boundaries	Merge vs split, conjoined methods, method splitting, dependency minimization
information-hiding	Information leakage, temporal decomposition, partial hiding, false encapsulation
pull-complexity-down	Caller burden, configuration parameters, the core asymmetry

Abstraction & Generality

Skill	Covers
abstraction-quality	Genuine vs false abstractions, layer boundaries, decorators
general-vs-special	Interface generality, special-general mixture, edge-case elimination
error-design	Define errors out of existence, exception masking, aggregation, just crash

Clarity & Communication

Skill	Covers
naming-obviousness	Isolation test, scope-length principle, consistency, avoid extra words
comments-docs	Comment types, comments-first workflow, cross-module documentation

Process & Evolution

Skill	Covers
strategic-mindset	Strategic vs tactical, investment rule, tactical tornado
design-it-twice	Generate alternatives, compare on criteria, synthesize
code-evolution	"Designed this way" standard, repetition, technical debt
complexity-recognition	Change amplification, cognitive load, unknown unknowns

Diagnostic

Skill	Covers
red-flags	Design smell scan covering structure, boundaries, documentation, naming, and process
design-review	Structured review funnel from complexity triage through structural, interface, and surface checks
diagnose	Routes a vague symptom or complaint to the most relevant skill via a decision tree

Attribution

These skills are adapted in part from the teachings of John Ousterhout, professor of computer science at Stanford University, and his book A Philosophy of Software Design. This project is not affiliated with, endorsed by, or sponsored by John Ousterhout, Stanford University, or the publishers of A Philosophy of Software Design.