pax-fullstack-harness

Diagnose failed or unreliable agent workflows by classifying bad assumptions, missing observations, tool misuse, wrapper failures, hidden repair loops, incomplete validation, handoff gaps, and repeated failure patterns. Use after a failed harness run, flaky agent output, or confusing validation loop.

Review agent-generated changes against project rules. Use when asked for a review, before opening a PR, after large Codex edits, or when checking diffs for correctness, security, architecture, tests, docs drift, dependencies, migrations, and release risk.

Generate or maintain a minimal high-signal AGENTS.md for coding agents. Use whenever a repo needs agent instructions, an existing AGENTS.md is too long or vague, or project docs/scripts need a concise routing map for Codex.

Define and mechanically enforce architecture boundaries for a repo. Use when a project needs allowed dependency rules, forbidden imports, layer documentation, dependency maps, circular dependency checks, or protection against agent-created architectural drift.

Scaffold a new TypeScript web, mobile, API, or full-stack project with Better-T-Stack, then harden it with the Pax Fullstack Harness. Use whenever the user wants a new TypeScript app scaffold, mentions Better-T-Stack, asks for a modern type-safe stack, or wants Codex to start a repo that is agent-ready from day one.

Use guided Chrome DevTools MCP setup and workflows for browser runtime debugging. Use when a web app needs console, network, DOM, accessibility, Lighthouse, performance, responsive, or live-browser investigation beyond normal tests.

Build PR validation, local CI reproduction, CI triage, and deployment gates. Use when a repo needs GitHub Actions beyond basic validation, coverage/artifact expectations, branch protection notes, CI failure triage docs, or release/deployment readiness gates.

Inventory and validate user-facing browser workflows with click-path evidence, accessible controls, Playwright coverage, console/network checks, and lifecycle verification. Use for UI work, web regressions, shadcn/ui changes, todo-style create/edit/delete flows, or browser acceptance testing.

Control dependency sprawl in agent-edited repos. Use when dependencies are added, removed, audited, approved, unused, risky, vulnerable, or need docs/quality/dependencies.md and scripts/check-dependencies.

Detect and prevent docs drift in agent-ready repos. Use when commands, environment variables, routes, migrations, schemas, public APIs, package scripts, architecture, or release behavior changes and docs may need updates.

Audit and reduce AI-generated codebase entropy. Use when a repo needs duplicate logic detection, dead-code checks, oversized file review, inconsistent pattern cleanup, stale TODO review, dependency drift checks, or a technical debt register.

Add or maintain a machine-readable feature state list for agent work. Use when a repo needs WIP=1 control, feature states, verification commands, pass-state evidence, or docs/features/feature_list.json.

Identify preferred local implementation patterns and encode them for agents to copy. Use when a repo has inconsistent patterns, needs canonical examples, templates, or docs/quality/golden-patterns.md updates.

Assess whether an existing repo is ready for Codex-heavy development. Use before large harness work, when a user asks for an audit, roadmap, maturity score, or wants to know which Pax Fullstack Harness skills to apply first.

Separate repo initialization from implementation. Use when a project needs startup readiness, first passing test evidence, task breakdown, scripts/init-check, or a clean initial checkpoint before feature work.

Make a project reproducible and inspectable locally for coding agents. Use when a repo needs dev server scripts, .env.example, Docker Compose, seed/reset commands, health checks, local logs, or runtime debugging instructions.

Manage database, schema, and API evolution safely. Use when a repo has migrations, database schemas, public API schemas, seed data, fixtures, compatibility requirements, rollback conventions, or needs scripts/check-migrations.

Generate scoped nested AGENTS.md files at meaningful repo boundaries. Use after scaffold or harness setup when apps, packages, docs, or service directories need local agent instructions without duplicating the root AGENTS.md.

Add local-first observability for coding agents. Use when a repo needs logging conventions, request or correlation IDs, diagnostics, scripts/logs, scripts/diagnose, debugging docs, metrics/tracing guidance, or smoke-test diagnostics.

Use OpenSpec for formal spec-driven development. Use when a feature, refactor, or product change should be proposed, designed, task-planned, validated, synced, or archived through openspec/ artifacts before or during implementation.

Require inspect-before-edit evidence for risky changes. Use before modifying architecture, data, auth, dependencies, migrations, generated workflows, validation wrappers, or broad UI/API behavior when the agent must prove it inspected relevant files and affected surfaces first.

Create or harden the baseline agent-ready structure for a repo. Use when a user asks to prepare a project for Codex, add an agent harness, standardize scripts/docs/CI, or make an existing repo easier and safer for coding agents to work in.

Generate thin project-local .codex/skills from a scaffolded repo's actual harness artifacts. Use after scaffold or harness setup when a repo should have local Codex workflows for feature state, session handoff, Definition of Done, runtime debugging, UI work, or review protocol without copying plugin skills.

Convert vague product or engineering requests into executable repo-local specs before coding. Use when a task is ambiguous, feature-sized, risky, cross-cutting, or needs acceptance criteria, non-goals, testing plan, rollout notes, or architecture impact.

Install executable quality gates for a repo and normalize validation behind stable scripts. Use when a project needs linting, formatting, type checking, tests, builds, doc checks, boundary checks, CI, or a single scripts/validate command for Codex.

Define and validate what ready-to-ship means beyond tests passing. Use when a repo needs release checklists, smoke tests, rollback plans, changelog/versioning guidance, deployment verification, or scripts/smoke-test.

Build operational repo-local documentation for coding agents. Use when a repo needs docs that explain product intent, architecture, commands, testing, quality gates, observability, security, release, or specs in a way Codex can act on.

Require current source, documentation, version, and rationale evidence before API, framework, dependency, security, integration, MCP, or external-service changes. Use when implementation depends on behavior that may have changed or should be sourced before coding.

Convert repeated review findings into durable harness checks. Use when code review catches recurring agent mistakes that should become docs, tests, lint rules, boundary checks, validation scripts, or quality policy.

Install secure-development basics for non-toy repos. Use when a repo needs secret scanning, dependency vulnerability checks, auth/session conventions, input validation guidance, safe environment variable handling, .gitignore protection, or a security review checklist.

Preserve agent state across sessions. Use when a repo needs PROGRESS.md, DECISIONS.md, handoff protocol, rebuild-cost reduction, session start or exit routines, or continuity for long-running Codex tasks.

Install, configure, and use shadcn/ui in React-style projects. Use when a repo needs shadcn/ui initialization, component installation, monorepo UI package setup, components.json review, registry or MCP guidance, or UI composition rules.

Audit reusable plugin skills, project-local .codex skills, AGENTS.md, nested instructions, context cost, stale guidance, duplicated rules, routing gaps, and daily-vs-library placement. Use when maintaining this harness, validating generated local skills, or checking whether agents can actually follow the available instructions.

Add process observability for agent work. Use when a task needs sprint contracts, evaluator rubrics, task traces, acceptance evidence, or structured feedback between generator and reviewer roles.

Generate and validate project-scoped Codex subagents for explicit read-heavy exploration, review, security, docs research, browser QA, and trajectory auditing. Use when a repo should expose .codex/agents specialists without bundling global agent config or spawning agents automatically.

Standardize Codex task planning and PR output. Use when work needs a task plan, PR description, validation evidence, risk assessment, rollback notes, docs/test summary, review rubric, or .github/pull_request_template.md.

Drive changes through TDD with unit tests and Playwright e2e tests. Use when a bug fix or feature should start with a failing reproduction, when browser workflows need e2e coverage, or when Playwright setup/test wrappers are needed.

Prevent premature completion claims. Use when a task needs Definition of Done evidence, three-layer validation, scripts/termination-check, or proof that static, runtime, e2e, docs, and clean-state checks passed.

Create a practical test strategy that coding agents can extend consistently. Use when a repo needs unit/integration/e2e conventions, fixtures, test command wrappers, examples, or a bug-fix rule that starts with a failing reproduction test.

Classify task traces, detect risky engineering trajectories, score advisory bad-state risk, and route to the least disruptive intervention. Use when a repo has docs/observability/task-trace.jsonl, trajectory risk policy, repeated agent drift, skipped validation, broad diffs, or needs shadow-mode trajectory supervision.