Agent Readiness

This repository is written to be read by coding agents, not just by people. Every file and directory below already exists in the public repo — this page just surfaces it, since most visitors will never browse GitHub to find it.

Agent Contract

What a coding agent reads before it makes any change, and in what order.

AGENTS.md

Universal, tool-agnostic agent contract — read by any coding agent (Codex, Cursor, Copilot, Claude Code) before touching this repo.

CLAUDE.md

Claude Code-specific layer on top of AGENTS.md — demo execution modes, the Agent Operating Contract, and a "what not to do" list learned from past sessions.

CONTEXT.md

Canonical domain vocabulary (45+ terms) so any agent uses precise, consistent language instead of guessing.

docs/adr/

Architecture Decision Records — the why behind the most surprising choices in the codebase, written after the fact.

specs/

Lightweight spec-before-code convention (inspired by GitHub Spec Kit) — what is about to be built and why, written before implementation.

Operating Surface

Reusable playbooks, session profiles, quality gates, and traceability — so agent work on this repo is repeatable and measurable, not ad hoc.

skills/

Runnable playbooks an agent can invoke by name: add a demo, run the full test suite, security review, agentic-SEO audit, recruiter/executive persona review, self-heal on a red CI run.

profiles/

Context-switching YAML files — which skills are active, which tools are permitted, and the focus area for a given session type (recruiter evaluation, testing, security audit, design).

evaluations/

Machine-readable quality gates per dimension — recruiter experience, security posture, agentic SEO, test health, accessibility, performance — each with a passing score, current score, and remediation steps.

runs/

Execution-trace directory — every significant skill run leaves a dated record, enabling measurement of agent reliability and regression rate over time.

None of this is theoretical — the demos on this site, its ADRs, and this page itself were built and reviewed by coding agents operating under this exact contract. See Governance for the runtime safety controls (guardrails, rate limiting, evals) applied to the AI features this contract governs.