Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    WaterfallAI project — project-manager

    1

    Run a phased project with AI role agents: phases, human-approved gates, baselines, change control and traceability on the ex-workflow engine.

    Free

    Security scanned
    WaterfallAI project — project-manager

    WaterfallAI project — project-manager

    Example session with this skill installed

    Run a phased project with gates: goal "migrate the reporting module". Start with the requirements phase; raise each gate for me to approve.

    • Read your context and instructions
    • Compiled the waterfallai project

    Summary

    The first phase below is the engine's own command loop as the skill drives it (the sequence the bundled test runs), narrated by the Project Manager persona. Every line comes from a wf command's JSON output.

    Requirements phase

    Pim (PM): Bee, the requirements for "migrate the reporting module"?

    Bee (BA): R1 "export runs on a schedule" (acceptance: a job exists and writes the file nightly), R2 "legacy report formats still open" — two items, phase requirements.

    — you confirm —

    wf add R1.json --by business-analyst   → item.added { phase: "requirements" }
    wf add R2.json --by business-analyst   → item.added { phase: "requirements" }
    wf ceremony kickoff k.json --by project-manager   → phase.opened { phase: "requirements", version: 1, items: [R1, R2] }
    wf next --max 2   → briefs R1@work, R2@work, role business-analyst
    wf complete R1 h1.json --tokens-in 58400 --tokens-out 0 --duration-ms 38000 --model haiku   → R1 → review (evidence srs)
    wf next   → R1@review, role qa
    wf complete R1 r1.json --role qa …   → item.completed, R1 → done
    … same for R2 …
    

    Pim: Requirements: 2 of 2 done. Deliverable srs present on both. Raising the gate.

    wf ceremony gateReview g.json --by project-manager   → phase.gate-pending { phase: "requirements", gate: "requirements-gate", pending: "p1" }
    wf decide p1 approve --by project-manager   → refused: gate-unmet — "project-manager" may not decide phase gate "requirements-gate" (p1); it awaits a human
    

    Pim: Gate p1 awaits you: SRS from Bee (R1, R2), both reviewed by Q. Approve or reject?
    — you approve —

    wf decide p1 approve --by human --note "ok"   → phase.baselined { phase: "requirements", version: 1, approvedBy: "human", items: [R1, R2] }
    wf ceremony kickoff k2.json --by project-manager   → phase.opened { phase: "design", version: 1 }
    wf ceremony phaseRetrospective pr.json --by project-manager   → phase.retrospective { phase: "requirements", lessons: [...], agreements: [...] }
    wf status   → report: phases[requirements: baselined, design: open, …], coverage { traced: 0, total: 2 }
    

    Next steps

    • Design phase: Arch proposes design items that trace to R1 and R2; requirement-traced is checked at the design and verification gates.
    • A change to a baselined requirement becomes a change-request with impactAnalysis { summary, reopen, weight } → changeControlBoard → your decision reopens the listed phases.
    • wf metrics: phaseCompletion, scheduleVariance, requirementCoverage, changeRequests, reworkWeight.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    WaterfallAI project runs a phased project of AI role agents inside your coding agent (Claude Code, Codex CLI, Cursor or any host that can run a prompt and node). The orchestrator is the project-manager skill; the business-analyst, architect, developer, qa and release-manager role skills are nested under roles/, so one install gives you a working team.

    What it does

    • Plans each phase with its deliverer: the business analyst proposes requirement items, the architect design items that trace to requirements, the developer implementation items, QA verification items, the release manager deployment items; you confirm, the engine stores them with their phase.
    • Opens phases in order (kickoff) and dispatches briefs only inside the open phase; QA reviews delivered items.
    • Raises each phase gate for you: when every item is done, the project manager collects the phase deliverables (srs, design-doc + interface-spec, code-baseline, test-report + the engine-generated traceability-matrix, release-notes + deployment-record) and raises the gate; a person approves or rejects. Approval baselines the phase; baselined items can no longer be edited.
    • Runs change control: a change to baselined work is a change-request item with an impact analysis (summary, phases to reopen, weight); the board recommends, you decide, the engine reopens exactly the listed phases.
    • Checks traceability: every requirement must trace to design and verification items before those gates; coverage is reported in wf status.
    • Holds a retrospective per phase whose adopted lessons become working agreements every later brief carries.

    How it works

    The engine is scripts/wf.js, a bundled build of the ex-workflow state machine (phases, gates, baselines, change control, traceability, phase metrics; published on npm as ex-workflow, MIT). All state lives in .workflow/ in your project; nothing leaves your machine. The skill is prose: it tells the host which wf command to run, how to build each worker prompt, and what to say to you.

    What's inside

    SKILL.md (the orchestrator), policy.json (the six default phases with QA delivering verification, every gate human-approved, personas, host model map), reference/ (every command with its JSON shapes, phase and gate narration, the worker contract, team voice), five role folders under roles/ (a SKILL.md plus the engine role manifest each), scripts/wf.js + scripts/package.json.

    Known limitations

    Every gate is human-approved by default: the project manager raises it and a person decides. With QA delivering verification items, QA also reviews them (the engine's single review role). Budgets are sized from the host's token counter; a host that reports no usage runs with estimates. This orchestrator has been exercised through the engine's full command loop (kickoff, deliver, review, gate, baseline, change control, retrospective) but not yet on a multi-day run with real workers; the Scrum sibling listing has.

    Support

    Message the creator on Agensi; updates ship as new listing versions.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Every gate is human-approved by default: the project manager raises it and a person decides. With QA delivering verification items, QA also reviews them (single review role in the preset). Budgets are sized from the host's token counter; a host that reports no usage runs with estimates. Exercised through the engine's full command loop including change control, not yet on a multi-day run with real workers.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 18 days ago

    • Free with an Agensi account

    Software
    Node.js 18 or later

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    File Scopes

    .workflow/**

    The bundled scripts/wf.js is the MIT ex-workflow engine; it reads and writes only .workflow/ under the project root. No network.

    Listed24 days ago
    Updated15 days ago

    What's inside

    Frequently Asked Questions

    A prompted manager skips gates, edits baselined work and approves its own phases. Here the bundled ex-workflow engine (scripts/wf.js) owns the project: phases open in order, a gate is raised by the project manager but decided by a person, approval baselines the phase and the engine refuses edits to baselined items (baseline-immutable), and only an approved change request reopens a phase. The orchestrator runs wf commands and narrates what they return.

    SKILL.md (the project-manager orchestrator), policy.json (the six default phases with QA delivering verification, human-approved gates, personas, host model map), reference/ (commands.md with every wf command and the Waterfall ceremony inputs and payloads, phases.md narration for kickoff, gates, change control and retrospectives, handoff.md worker contract, team.md), roles/business-analyst, roles/architect, roles/developer, roles/qa, roles/release-manager (a SKILL.md plus the engine role manifest each), and scripts/wf.js with scripts/package.json (the bundled MIT engine). 19 files, no network, all state under .workflow/.

    Any host that follows the SKILL.md standard, can run a prompt and has Node.js 18+ on PATH. Tested on Claude Code, where workers are dispatched as subagents; a host without subagents plays each role inline with the same prompt. policy.json maps small/medium/large tiers to host model names (the defaults never use the largest models); a non-Claude host maps them to its own.

    When every item of the open phase is done, the project manager runs gateReview with the phase deliverables collected from the handoffs; the engine raises a pending decision that only a person may approve or reject (the PM is refused with gate-unmet). Approval baselines the phase; the engine then refuses edits to its items. A change to baselined work is a change-request item with impactAnalysis { summary, reopen, weight }; changeControlBoard records the recommendation, your decision reopens exactly the listed phases with a fresh kickoff. Every requirement must trace to design and verification items before those gates (requirement-traced); the engine generates the traceability-matrix deliverable and reports coverage in wf status.

    The command loop (kickoff, deliver, review, gate, human approval, baseline, next phase, change control, retrospective) runs end to end in the package's test suite against the bundled engine; a multi-day run with real workers has not been done yet (the Scrum sibling listing has three such sprints). The engine is the npm package ex-workflow (MIT); each listing version bundles one engine version, buyers get new versions as updates, and your .workflow/state.json loads unchanged across versions. For problems or questions, message the creator on Agensi.

    New in Agents & Orchestration

    Free