GUIDES · INTERACTIVE LESSONS

Methodology in motion

A dozen standalone interactive lessons that explain Forgeplan methodology from the inside out: how decisions are scored, why R_eff equals min() not average, how a PRD becomes a Spec becomes a shipped feature, and how multiple AI agents stay out of each other's way.

Reading order

Built linearly — each lesson references concepts from the previous. Companion interactive pages (3D charts, calculators) follow their theory pairs.

1. 00
   level zero · part 1 · the basics theory

   What a decision log is and why a team needs one

   Level zero — no tool branding. What "decision log" means as an engineering practice (ADR, DDR formats), six mandatory fields of a record, who needs it, six use-cases across domains, and why this matters more in an AI-augmented workflow.
   →
2. 01
   level zero · part 2 · practice theory

   How to introduce decision-logging into your team

   Practical companion to the intro. When to write a record and when to skip it (four rules). Depth scale. Typical day. Three ways an AI agent reads your context. Five agent–record interactions. Anti-patterns.
   →
3. 02
   overview · artifacts theory

   How to manage decision artifacts in a team

   Overview between intro and detailed guides. What counts as an artifact (3 properties), one-file-one-responsibility rule, the rhythm of work, three signs of a live artifact, 5 anti-patterns. Links to formal cycle and lifecycle for depth.
   →
4. 03
   overview · adi theory

   ADI as a daily practice: three versions before fixing

   Overview of the "write three versions of the cause" technique as a daily habit. Origin in 19th-century medical diagnostics. When to apply, one-cycle walkthrough, why exactly three (statistical threshold). Links to the formal cycle.
   →
5. 04
   overview · ddr theory

   What a good DDR looks like: template with full example

   Overview of the DDR format. DDR vs ADR comparison, six mandatory fields, full markdown example (JWT vs sessions vs OAuth). Why length matters (200–400 words, one screen). Links to the formal cycle and lifecycle.
   →
6. 05
   overview · fpf theory

   FPF: the consistency map

   Seven kernel nodes of FPF - holons, context, distinctions, decay, open/closed world, F-G-R trust, reasoning cycle. How they connect and why keeping them consistent matters.
   →
7. 06
   overview · trust calculus theory

   Trust Calculus: F/G/R and its connection to DDR

   Trust Calculus in plain language: three axes (Formal / Granular / Reliable), weakest-link rule (not average), context-level penalty, the R_eff formula in one line, where F/G/R lives inside a DDR record. Bar chart for average-vs-min contrast. Links to the 3D interactive.
   →
8. 07
   overview · evidence theory

   Evidence and decay: why proofs have a shelf life

   Overview of evidence types (own measurement, peer benchmark, vendor docs, third-party experience), three dimensions of strength (F/G/R brief recap), what happens to evidence over time (decay formula), revisit triggers by decay. Links to trust-calculus and lifecycle.
   →
9. 08
   detailed · foundation theory

   How decisions are made and evidence weighted

   ADI reasoning (Abduction → Deduction → Induction), DDR field on every ADR, and the Trust Calculus framework. Why averaging trust is wrong and the weakest-link principle.
   →
10. 09
    companion · interactive interactive

    Trust Calculus in 3D F/G/R

    3D scene with seven evidence on a /search latency scenario. Plane of threshold F+G+R=12, hover cards, hypothesis selection by WLNK. Rotate the scene with mouse — projector-ready teaching artifact.
    →
11. 10
    one feature theory

    BMAD · 13-step PRD · Adversarial Review

    Four phases of the funnel (Brainstorming → Modeling → Architecting → Delivery), 13 PRD sections of which 3 are critical (Non-Goals, Risks, Open Questions), and adversarial review through fresh eyes as the third quality gate.
    →
12. 11
    one specification theory

    Spec · Delta · R_eff

    Spec vs PRD: verifiable behavior, not intent. Delta format ADDED/MODIFIED/REMOVED as git diff for requirements. R_eff = min(scores) with Evidence Decay — why average deceives, minimum does not.
    →
13. 12
    after activate theory

    Lifecycle · from draft to terminal

    State machine with 5 states and 6 transitions: draft → active → {superseded | deprecated | stale} → renew/reopen. Passport-with-visa analogy. Stale as "forgotten state" requiring explicit decision.
    →
14. 13
    meta framework theory

    FPF · First Principles Framework as the parent rail

    Six base concepts (Bounded Context, Trust Calculus, F-G-R, ADI Cycle, Category Error, Gamma Algebra) + three usage modes (decompose / evaluate / reason). Shows how Trust Calculus and ADI are not separate ideas but built-in moves inside the same parent framework.
    →
15. 14
    companion · interactive interactive

    DAG Explorer · artifacts in 3D

    Plotly 3D scatter with artifact nodes and edge lines. Knowledge Vault MVP as a real example: 22 artifacts, 30 connections, type filters. Shows patterns (healthy chain / blindspot / orphan / branching) and traceability in one picture.
    →
16. 15
    companion · interactive interactive

    Depth Calibrator · routing in real-time

    Three sliders (complexity / blast radius / reversibility) + three hard gates (public API / security / cross-team) → real-time depth tier + pipeline. Presets for typical scenarios (cosmetic / team feature / public API / auth rewrite). Move sliders to see the marker shift and artifact list change.
    →
17. 16
    agent ↔ tool language theory

    Hint Contract · 5 markers

    How the agent stopped hallucinating next steps. Five markers (Next: / Or: / Wait: / Done. / Fix:) + five contract rules (Presence / Actionability / Determinism / Conditionality / Consistency). Reference card.
    →
18. 17
    parallel work theory

    Multi-agent · dispatch · claim · release

    Coordinating 2-5 agents in one repo without conflicts. Planner cuts work into conflict-free buckets, worker takes one with TTL, release returns to pool. WIP=1 hard limit. Adversarial review always in a separate agent.
    →
19. 18
    the tool itself tool

    Forgeplan as the operational layer

    POS for methodologies: CLI + MCP + .forgeplan/ absorbs FPF, BMAD, OpenSpec, QuintCode. 10 artifact types (6 actively used) with dependency DAG. Four depth levels and three quality gates in one validate.
    →
20. 19
    tutorial interactive

    First Artifact · 20-minute hands-on walkthrough

    Step-by-step simulated terminal: init → health → route → new prd → validate → reason (ADI) → build → new evidence → link + score → activate. Progress bar with clickable points, live workspace side-panel, common-error blocks at each step.
    →

External documentation

Beyond these lessons: full Forgeplan reference.