By SamMRoberts
Sentinel for vibe-coding: enforce assumption verification and change-control scope before implementation.
Use when: enforcing assumption discovery before implementation, validating assumption-gate artifacts, wiring assumption checks into Codex, Copilot, Claude, hooks, CI, or pre-commit workflows. Private stage agent of vibe-sentinel.
Use when: compiling vague implementation requests into enforceable scope contracts, validating change-control artifacts, checking file drift, or wiring scope enforcement into Codex, Copilot, Claude, hooks, CI, or pre-commit workflows. Private stage agent of vibe-sentinel.
Use when: enforcing vibe-coding guardrails (assumption gate + change-control contract) before implementation, validating either artifact, wiring scope enforcement and assumption verification into Codex, Copilot, Claude, hooks, CI, or pre-commit workflows.
Use before implementation, bug fixes, refactors, parser changes, Playwright changes, CI changes, or architecture changes to identify, verify, and record assumptions before editing code.
Use before implementation, bug fixes, refactors, parser changes, Playwright changes, CI changes, dependency changes, or architecture changes to compile the user request into an enforceable Change Control Contract.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A personal collection of agent customizations: plugins, instructions, custom agents, and the marketplace that distributes them. Built around the idea that agent quality comes from explicit workflows, deterministic validation, and small composable primitives — not large prompts.
agentic-engineering/
├── plugins/ # Installable plugins (this is what the marketplace serves)
│ ├── marketplace.json # Auto-generated marketplace descriptor
│ ├── agent-compat/ # Copilot custom agent compatibility for Codex/Claude
│ ├── agent-on-ramp-coach/ # Confidence-graded coding-agent adoption coach
│ ├── ux-gremlin/ # Hostile-but-realistic UX resilience test plans and Playwright generation
│ ├── vibe-sentinel/ # Assumption gate + change-control contract guardrails (vibe-coding sentinel)
│ └── web-ux-test/ # Workflow-enforced Playwright UX testing pack
├── agents/ # Standalone custom agents (not packaged as plugins)
├── instructions/ # Repository-scoped AGENTS.md instructions
│ ├── general-coding/ # Always-on coding discipline
│ └── agentic-tooling/ # Rules for editing agent tooling itself
├── scripts/
│ └── sync-marketplace.mjs # Regenerates plugins/marketplace.json
├── .githooks/ # Repo-managed git hooks (opt-in per clone)
├── .github/
│ └── workflows/ # CI: validate plugins, sync marketplace
└── AGENTS.md # Workspace-wide agent guidance
The two most active areas are plugins/ (full installable packs) and instructions/ (repo-scoped guidance that applies to every change made in this workspace).
Each subfolder under plugins/ is a self-contained pack that can be installed independently. Conventions are documented in instructions/agentic-tooling/repo/AGENTS.md, but the highlights:
Current plugins:
| Plugin | Purpose |
|---|---|
| agent-compat | Convert Copilot *.agent.md custom agents into Codex AGENTS.md and Claude custom-instructions.md overlays. |
| agent-on-ramp-coach | Help engineers adopt coding agents safely through confidence levels and reviewable sessions. |
| ux-gremlin | Turn happy-path web flows into hostile-but-realistic gremlin scenarios with a Playwright recipe DSL and executive reports. |
| vibe-sentinel | One plugin combining the assumption gate (verify assumptions before edits) and the change-control contract (lock scope, detect drift). Replaces the former assumption-killer and change-control-compiler plugins. |
| web-ux-test | Plan, run, classify, repair, and report structured web UX tests with a stateful CLI and MCP workflow server. |
The marketplace descriptor lives at plugins/marketplace.json and lists every plugin in plugins/ with a git-subdir source pointing back at this repository. Any Claude Code (or compatible) marketplace reader can install from it.
Add the entry below to your marketplace file (typically ~/.agents/plugins/marketplace.json for Claude Code):
{
"source": "https-fetch",
"url": "https://raw.githubusercontent.com/SamMRoberts/agentic-engineering/main/plugins/marketplace.json"
}
…or clone this repo and point the host at the file path directly. Then install any listed plugin from the host's marketplace UI.
The descriptor is generated, but each plugin can pin custom values by adding a marketplace block to its .codex-plugin/plugin.json (or root plugin.json):
{
"name": "my-plugin",
"marketplace": {
"ref": "v0.2.0",
"policy": { "installation": "PREVIEW", "authentication": "NONE" },
"category": "Quality"
}
}
Defaults are ref: main, policy: AVAILABLE / ON_INSTALL, and category inferred from interface.category.
plugins/marketplace.json is generated by scripts/sync-marketplace.mjs. Three layers keep it in step with plugins/:
npx claudepluginhub sammroberts/agentic-engineering --plugin vibe-sentinelCompatibility layer that converts Copilot *.agent.md custom agents into Codex and Claude instruction overlays.
Helps hesitant engineers adopt AI coding agents through safe confidence levels, read-only workflows, approval gates, and reviewable summaries.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Complete developer toolkit for Claude Code