Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Reckon: Make AI Attack Your Spec, Not Flatter It

    2

    Turns a raw idea or an existing PRD into a proceed / descope / kill verdict, using isolated bull-case and bear-case reviews after you commit your own take first.

    Free

    1 useSecurity scanned
    Reckon: Make AI Attack Your Spec, Not Flatter It

    Reckon: Make AI Attack Your Spec, Not Flatter It

    Example session with this skill installed

    Reckon this PRD before we commit a sprint to it: docs/forge-pulse-prd.md

    • Read your context and instructions
    • Compiled the reckon

    Reckon first asks: "Before I show you anything: in your own words, what do you think works about this spec, and what worries you most?" After you answer, it runs the isolated praise and critic reviews and returns:

    Reckoning Verdict — forge-pulse-fast-feedback-team-heartbeat (v1)

    ⚖️ Verdict: DESCOPE

    3 critical failure mode(s) and 3 weak market verdict(s) — fix before building.

    ✅ Strengths

    • Problem statement is empirically grounded with converging evidence from multiple independent sources [strong] · 🔒 DO NOT REGRESS — §problem
      The convergence of independent evidence makes the problem undeniable.
    • Quantitative baselines establish clear improvement targets [strong] · 🔒 DO NOT REGRESS — §problem
      Specific baselines enable falsifiable hypotheses.
    • Personas are dimensioned with empirical behavioral data rather than fictional archetypes [strong] · 🔒 DO NOT REGRESS — §personas
      Empirically-grounded personas create actionable design constraints.

    💡 Non-obvious insights

    • Leading indicators are distinguished from lagging outcomes — §success_metrics
      The success metrics section doesn't just list goals—it identifies early signals that predict long-term success.
    • Anonymity implementation recognizes tradeoff between perfect privacy and vote integrity — §technical_considerations
      The technical section explicitly states: 'Risk: Systems that don't track who is voting create possibility for voting abuse.'

    🔴 Failure modes (by severity)

    • [critical/high] Anonymous Voting Manipulation Will Destroy Trust in First Real Incident (failure_modes · §technical_considerations)
      Without cryptographic vote prevention (finding-014 shows one-time ring signatures required), first time someone stuffs votes with 20+ upvotes on their own note, entire team loses faith in anonymity system.
    • [critical/high] 1-2 Month Timeline With WebSocket, Real-Time Collaboration, and GDPR Compliance Is Fantasy (failure_modes · §scope)
      Finding-020 shows simple products take 1-3 months, real-time features add 2-3 weeks.
    • [critical/medium] Permanent Anonymity Prevents You From Moderating Harassment or Malicious Content (failure_modes · §technical_considerations)
      Technical section acknowledges 'permanent anonymity creates forensic challenge if malicious content posted.' First time someone posts 'Sarah is incompetent and should be fired' in anonymous note, you have zero recourse.

    🧭 Market skeptic

    Why care? [moderate] The pain is real (40-50% action completion, recurring themes, psychological safety barriers) but the willingness to pay is questionable.

    • Why now? [weak] The PRD provides no 'why now' argument.
    • Why win? [weak] Differentiation is Heartbeat Timeline + permanent anonymity + auto-carry actions.
    • Feature or product? [weak] This is a feature bundle that any existing retro tool could add in 4-8 weeks.

    📋 Overall

    This PRD exhibits three fatal flaws: (1) Solving symptom (low action completion) not cause (no organizational accountability culture), (2) Technical complexity (anonymous voting cryptography, WebSocket scaling, GDPR compliance) vastly underestimated relative to 1-2 month timeline, (3) Zero defensible moat against 10+ free competitors who can copy all features in 6-12 weeks.

    Critic confidence: 0.88 · Praise confidence: 0.92

    Then Reckon adds a short cognitive-forcing diff of your answer against these findings.

    Condensed for this listing: top strengths and critical failure modes only, with detail trimmed to the first sentence. The full verdict is in examples/expected-verdict.md inside the skill.

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

    About this skill

    Most AI flatters your spec or hedges into a "balanced" non-answer. Reckon makes it survive contact with reality.

    It runs two reviewers that never see each other: one builds the strongest case for your spec, the other the harshest case against it. You get a real bull case and a real bear case, not one watered-down take.

    How it works

    1. Start from anything. Give it a raw idea, and it interviews you and drafts a PRD with every assumption flagged. Or hand it an existing spec, PRD or design doc.
    2. Commit your own take first. Before you see any AI output, you write down what you think works and what worries you. This stops you anchoring on the machine.
    3. Isolated review. A praise engine and a critic run separately, so neither can soften the other.
    4. Validated output. Both reviews are checked against strict schemas, so results stay structured and consistent from run to run.
    5. Verdict. You get a severity-ranked proceed / descope / kill call, plus a diff of your judgment against the AI's: what you caught, what you missed, and where you disagree.

    What the critic checks

    • Failure modes, scale pain, user confusion risks and engineering regrets, each rated by severity and likelihood and tied to a spec section
    • Market skeptic: Why care? Why now? Why win? Feature or product?
    • Kill recommendation, reserved for fundamentally flawed ideas, not fixable problems

    What's inside

    • SKILL.md: the full flow
    • references/: three prompts (PRD synthesizer, praise engine, critic)
    • scripts/reckoning.mjs: a short, readable validator and renderer with no dependencies (Node 18+)
    • examples/: a real draft PRD and a real verdict

    Known limitations

    • No web research yet. PRDs drafted from an idea use only your answers, so citations stay empty and assumptions are flagged.
    • True isolation needs an agent that can run subagents. On other agents the two reviews run one after the other, and the output says so.
    • Rendering needs Node 18+. Without it, the agent formats the output by hand.

    Free and open source (MIT): github.com/alchemyguy/reckoning

    No web research yet: PRDs drafted from an idea use only your interview answers, so citations stay empty and assumptions are flagged. True isolation of the praise and critic reviews needs an agent that supports subagents; elsewhere they run one after the other and the output says so. Automatic validation and rendering need Node.js 18+.

    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 any payment.

    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

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Node.js

    Permissions

    Read Files
    Write Files
    Terminal / Shell

    File Scopes

    .reckoning/**
    Listed1 month ago

    What's inside

    Frequently Asked Questions

    A single prompt tends to flatter the spec or hedge into a "balanced" take. Reckon runs a praise engine and a critic in isolation, so neither can soften the other. It also makes you commit your own assessment before showing any AI output, then diffs your judgment against the AI's.

    Both. For a raw idea, Reckon runs up to three rounds of clarifying questions and drafts a PRD with every assumption flagged, then red-teams it. For an existing spec, PRD or design doc, it goes straight to the review.

    KILL when the critic judges the idea fundamentally flawed. DESCOPE when there is at least one critical failure mode or two or more weak market verdicts: fix those before building. PROCEED when there are no critical blockers, with the remaining risks listed.

    No. Reckon runs inside your own agent session. The bundled script makes no network calls and needs no npm install; it only reads and writes JSON files in a local .reckoning/ folder.

    Any agent that supports the SKILL.md standard, including Claude Code, Cursor, Codex CLI, GitHub Copilot and Gemini CLI. Agents that can run subagents get fully isolated reviews; on others the two reviews run one after the other. Want only part of the flow? See the companion skills spec-red-team, idea-to-prd and spec-premortem.

    New in Research & Analysis