More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+22 more

    Planpick: one plan, pick by number, ship in waves

    1

    Surveys the code read-only, writes one plan with numbered options (one recommended, cost and risk on each), waits for your pick, then builds in tested waves with a status update after each.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Planpick: one plan, pick by number, ship in waves

    Planpick: one plan, pick by number, ship in waves

    Example session with this skill installed

    Propose a plan to add search to the product list. Let's discuss first, no code yet.

    • Read your context and instructions
    • Compiled the planpick

    Planning, no code yet. Goal restated: shoppers find a product by name in under two seconds.

    Surveyed app/products, lib/db and tests read-only. Second pass refuted 1 of 4 findings (the "unused Fuse.js dependency" is used in admin), so 3 confirmed findings feed the plan.

    GOAL        Product list search by name, fast on 2k products
CONTEXT     - Products load via a server action, no query param
            - Postgres already has a products.name index
            - No tests cover the product list
OPTIONS
  1. Server-side ILIKE in the existing action  (RECOMMENDED)
     cost: ~4 h, 6 files   risk: cache drift on the list page
  2. New REST /api/search layer
     cost: ~8 h, 11 files  risk: second fetch path to maintain
  3. Do nothing, browser find
     cost: 0               risk: no mobile search
WAVES       W1: query param + action filter + tests, commit
            W2: debounced input + empty state, commit
            W3: Playwright search spec, full e2e, commit
OWNER-ONLY  none
GLOSSARY    ILIKE: case-insensitive text match in Postgres
            cache drift: list page showing results from an older query
ASK         Reply with numbers.

    When you reply, I will echo your pick and lock the plan. "proceed" locks it only; I will then ask "Start wave 1?" and wait.

    Example heartbeat after wave 1:

    WAVE 1/3 DONE · tests 48/48 green · typecheck ok · commit a1b2c3d
shipped   query param, action filter
next      wave 2: debounced input, empty state
blocked   none

    Next steps

    • Reply "1" to take the recommendation, or another number.
    • Ask "what does cache drift mean?" for any term before picking.
    • Say "start wave 1" after the plan is locked.

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

    About this skill

    You said "let's discuss first" and your agent started editing files. Or it handed back ten options with no recommendation, took "ok, proceed" as permission to build, and left you typing "what's the status?" again and again.

    Planpick is a discuss-first planning loop: one plan, numbered options, your pick by number, then tested waves with a status update after each.

    What it does

    • Read-only survey by dimension (correctness, security, performance, accessibility, test coverage) or by module, with a second pass that tries to refute each finding before it enters the plan.
    • One plan file in a fixed format: goal, context, numbered options with exactly one RECOMMENDED, cost and risk on every option, waves, owner-only items and a glossary line per internal term.
    • Pick by number accepts short answers like "1 3" and echoes the chosen IDs back before locking.
    • Proceed locks the plan only. It replies "Plan locked. Start wave 1?" and waits for an explicit build signal.
    • Waves with a gate add one regression test per fix, run targeted tests plus typecheck and lint, and commit locally only on green.
    • Heartbeat after every wave without being asked: shipped, next, blocked.

    How it works

    1. Enter planning and restate the goal in one line.
    2. Survey and verify findings against the code.
    3. Write the plan and ask by number.
    4. Lock, then wait for the build signal.
    5. Execute P0 first, heartbeat and commit per wave, resume from the first unchecked item.

    Frameworks & tools

    Uses git, tsc --noEmit and ESLint in the gate, plus the stack's test kit (Vitest, Playwright, pytest, PHPUnit or Pest). Claude Code plan mode and AskUserQuestion where available, with chat equivalents for other agents.

    Why this beats prompting it yourself

    A plain prompt drifts into code during discussion. Planpick separates locking the plan from permission to build, and keeps one plan file as the source of truth so a new session resumes without re-explaining.

    Use cases

    • Feature or refactor bigger than one obvious edit.
    • Pre-deploy audit with a test kit.
    • Backlog review "what's left?"
    • Build vs buy choices scored on weighted criteria, with a do-nothing line.

    Known limitations

    Not for one-line fixes with a clear spec. Pushes or opens a PR only when you say so. Full e2e runs only on the last wave or on request.

    How to use

    Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill passes our trust checks below.

    Security scanned

    Passed our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scanned

    Verified clean 1 day ago

    • Passed all security checks, Safe to install

    Requires

    git
    tsc
    ESLint
    Vitest
    Playwright
    pytest
    PHPUnit
    Pest

    Listed1 day ago

    What's inside

    Frequently Asked Questions