agentic-engineering Plugin

Chain-aware loop driver. Works through implementation-plan tasks unattended — branch, build, ship a draft PR, dispatch the verifier, checkpoint, repeat — until the target is reached, the circuit breaker trips, or the task budget is spent. Never merges.

Entry point for the Agentic Engineering suite. Assesses which documentation-chain documents exist in this project, reports where the project stands, and routes to the right skill for the next step.

Independent verification specialist for the agentic-engineering chain. Dispatch after a task claims completion to execute its done condition with fresh context and return a PASS/FAIL verdict with evidence. Use proactively before any task is marked ✅ in STATUS — the producer must never grade its own

Use when bootstrapping a new repo for AI-assisted development, when no AGENTS.md / CLAUDE.md / copilot-instructions.md / .cursorrules exists yet, or when consolidating scattered tool-specific config into one source of truth. Triggers on "create AGENTS.md", "set up agent instructions", "new project setup", "configure Claude/Cursor/Copilot/Codex", "onboarding a repo", or noticing duplicated agent config across tool-specific files.

Use when a technical choice has multiple valid options and the decision has lasting consequences, when revisiting or superseding a prior architectural decision, or when about to lock in a technology, pattern, or constraint that will be hard to reverse. Triggers on "should we use X or Y?", "REST vs GraphQL", "monorepo vs polyrepo", "PostgreSQL vs DynamoDB", "event sourcing vs CRUD", technology selections, pattern choices, "let's just use X" (implicit decisions), or any moment alternatives exist and trade-offs are accepted.

Use when something said or decided in conversation makes a chain document (PRD, SPEC, ADR, IMPL PLAN, AGENTS.md) stale, contradicted, or incomplete, or when the user proposes a new feature or scope change mid-project. Triggers on "I want to add a feature", "let's also support X", "change the architecture", "switch from X to Y", "we decided to...", "the spec is out of date", "keep the docs in sync", "the code now does X but the doc says Y", or noticing any decision in the conversation that the docs don't yet reflect. The agent flags drift proactively and MUST get the user's permission before editing any fundamental document.

Use when an existing or legacy codebase needs to enter the agentic-engineering workflow but has no chain documents (or only scattered legacy docs — README, wiki, design notes). Triggers on "onboard this repo", "migrate this project to agentic-engineering", "reverse-engineer the docs", "adopt this workflow on an existing codebase", "we already have a codebase, set up the docs", "document what we already built", or pointing the suite at a brownfield project. Reverse-engineers the code into an as-built documentation chain; changes docs, never code.

Use when implementation work is about to touch code (branch before the first commit), when a built task needs to ship as a pull request, when defining how branches, commits, and PRs map to plan tasks, or when parallel agents share one repository. Triggers on "create a branch", "commit this", "open a PR", "ship this task", "how do we merge", task start or completion during an implementation plan, worktree setup, or noticing commits landing directly on main.

Use when specs are approved and work needs to be broken into independently executable tasks, when sequencing dependencies across modules, when sizing milestones, or when a project is about to start coding but lacks a milestone breakdown. Triggers on "let's plan the implementation", "break this into tasks", "sprint plan", "what should we build first", "milestones", "task ordering", "estimate the work", or transitioning from spec to code.

Use when a task or milestone claims completion and must be verified before being marked done, when acting as a verifier sub-agent checking a producer's work, or when a "done" claim needs evidence. Triggers on "verify this task", "is this actually done", "acceptance check", "run verification", any completion claim, any 🔍 → ✅ status transition, or dispatching a maker-checker review loop.

Use when starting a new project or feature from scratch, when a rough idea needs to be turned into a structured product spec, or when scope, target users, or success metrics are undefined. Triggers on "let's build X", "new project", "kick off", "write a PRD", "product brief", "what should we build", vague feature requests, or when no PRD exists yet for the work about to start.

Chain-aware unattended loop driver for the Agentic Engineering suite. Use ONLY when the user explicitly asks to run multiple implementation-plan tasks unattended — "run the loop", "run-loop M2", "work through milestone M2 without stopping", "run the next N tasks", "keep going until blocked". Never self-trigger for a single task or an ordinary coding request. Branch → build → draft PR → dispatch the verifier → checkpoint → repeat, until the target is reached, the circuit breaker trips (3 FAILs on one task), or the task budget is spent. Never merges.

Use at the start of every coding session (read docs/status.md before any other action), at the end of every session (append handoff log entry), or when checking current progress, last-done work, in-flight tasks, or what's blocked. Triggers on session start/end, "where are we", "pick up where we left off", "handoff", "what was done last session", "resume", or recovering after context loss / new agent session.

Use when the PRD is finalized and precise implementation contracts are needed, when defining API shapes, data models, state machines, error taxonomies, or acceptance criteria, or when ambiguous "handle this gracefully" requirements need to be hardened into testable specs. Triggers on "write the spec", "define the API", "what's the data model", "state diagram", "error contract", "acceptance criteria", or transitioning from PRD to implementation.

Entry point and router for the Agentic Engineering suite. Use when starting work in a project that uses (or should use) the documentation chain, when unsure which chain document or skill comes next, or when asked where the project stands. Triggers on "use agentic engineering", "using-agentic-engineering", "where does this project stand in the chain", "what's the next step", "assess the chain", "how do I start with this suite", or a request naming a chain step whose prerequisites aren't in place yet.