Apache 2.0 · 100% Open Source · Runs on Your Machine

The AI that never forgets,
never lies.

Most AI tools have brilliant assistants with amnesia. Forge has memory.

Free forever · local · no account needed · vector database · knowledge graph · confidence decay · contradiction detection · 25 personas · 190+ skills · 150+ commands

8
Lifecycle Phases
150+
CLI Commands
25
Canonical Personas
190+
Skills
Works with your AI host
Claude Code OpenAI Codex CLI Gemini CLI GitHub Copilot (preview) Cursor (preview)
// The Problem

Every developer hits this at week 3.

You've been building with an AI assistant for weeks. It knows your stack. It knows your decisions. It knows why you chose Postgres over MySQL, why you killed that microservice, why that edge case exists.

Then one day - it doesn't.

It suggests the pattern you explicitly rejected. It forgets the constraint your CTO flagged. It confidently tells you something that used to be true but isn't anymore.

That's not an AI problem. That's a memory problem.
And until Forge, nobody solved it.

// Forge Anvil Features

Memory treated as infrastructure, not an afterthought.

Every other tool stores memories forever and treats a note from 3 months ago the same as something you said this morning. Forge doesn't.

!

Never Get Gaslighted

The moment Forge receives a memory that conflicts with an existing one, it flags both. You see the conflict. You resolve it. Your AI works from verified facts - not a pile of stale notes it has to guess from.

~t

Memory That Ages Honestly

Every memory has a confidence score that decays over time. A fact you verified today scores 1.0. The same fact 6 weeks later, unverified, scores 0.6 - and your AI tells you before using it to make a decision.

Vector Database Engine

Forge Anvil is a vector database embedded directly in your project - zero setup, no API keys. Every memory is embedded and searched by meaning, not keywords. Ask "what did we decide about auth?" and it finds the answer even if those exact words never appear.

Every Fact Has a Receipt

When Forge retrieves a memory, it gives you the fact plus confidence score, status, verified date, age, and any warnings. Your AI knows not just what to say - it knows how sure to be.

Your Data Stays Yours

A pure JSON store embedded in your project. No server, no Docker, no API key, no vendor to trust. If forgeguru.ai disappeared tomorrow, your Anvil keeps working. Your machine is the source of truth.

Open Source to the Core

Forge Anvil store: Apache 2.0. TF-IDF embeddings: pure JS, zero deps. Every line of the memory engine is auditable. No black boxes. No "trust us." Read it, fork it, audit it - the source is the truth.

$_

One CLI to Rule Your Memory

Everything your AI knows - visible and controllable from the terminal. Strike, draw, seek, temper, fractures, assay, cast. No magic. No mystery.

Knowledge Graph

Forge doesn't just store memories - it links them. ADRs, decisions, constraints, and retro learnings form a traversable knowledge graph. Ask "what depends on this decision?" and trace the full causal chain across your project history.

// The Persona System

25 personas. One apex operator.

Every Forge project is staffed by a roster of typed personas - each with a pre-flight checklist, an explicit output contract, and a defined scope. Solo developer wears all hats sequentially. Collaborators own hats individually. The roster is canonical; headcount is variable.

Blacksmith Apex Persona - Earned, Not Assigned Can impersonate any persona with full audit trail. 5-tier ladder: Apprentice → Smith → Master Smith → Master Forger → Blacksmith.
@analystResearch
@pmProduct
@architectSystems
@designerUX / Tokens
@engineerBuild
@qaTesting
@sreReliability
@trust-officerSecurity
@compliance-officerRegulatory
@auditorEvidence
@marketerGrowth
@cs-leadCustomer
@researcherDomain
@cfoFinance
@people-leadHiring
@orchestratorDual-Ledger
@nuke-operatorWind-Down
@red-teamAdversarial
@security-auditorCVE / SBOM
@performanceLatency / CWV
@investigatorDebug / Forensics
@edge-case-hunterBranching Paths
@wcag-auditorAccessibility
@doc-syncDoc Drift

Every persona enforces at the tool level — not just a prompt preamble. An @analyst literally cannot call Edit on source files. Pair two roles with --pair for adversarial review.

$ forge persona declare architect --pair=security # dual-role adversarial review Loaded: Winston, principal architect · Tools restricted to: Read, Grep, Skill, Bash(read-only) Pair role: security · adjudicates every design decision · refusals logged $ forge blacksmith as @qa forge quench # impersonate with full audit trail IMPERSONATE_BEGIN · persona=qa · by=blacksmith · ts=2026-05-15T09:12:00Z · audit=logged

Blacksmith 5-Tier Ladder — earned, not assigned

Tier 1
Apprentice
Tier 2
Smith
Tier 3
Master Smith
Tier 4
Master Forger
Tier 5 ⬡
Blacksmith
// The Forge Pyramid

Three protocols that govern everything else.

Every protocol in Forge operates inside the constraints of three apex protocols. No other protocol can control, coordinate, or override across protocol boundaries — only these three can. Together they form the load-bearing triangle of the system.

blacksmith
Authority
lifecycle
Structure
decision
Permanence

All other protocols defer to this triangle for authority, sequencing, and record-keeping.

№1 blacksmith Authority

Root control over everything else. Can override, impersonate any persona, nuke anvils, bypass gates, and force-clear bands. The only protocol that sits above other protocols — nothing in Forge operates outside its reach.

god-mode TTL two-key safety Class A/B/C ops persona impersonation append-only audit log
№2 lifecycle Structure

The 8-phase spine every project runs inside. Brief → Survey → Strike → Shape → Temper → Quench → Stamp → Ship. Without it, no other protocol has a sequence to attach to. Blacksmith governs it; lifecycle sequences everything below.

8 phase verbs entry gates phase-back routing .forge-phase state exit criteria
№3 decision Permanence

The record that survives even a nuke. ADRs and decision logs (DEC-NNN) outlast any project, persona, or restore operation. You can restore an anvil — you cannot un-make a recorded decision. The institutional memory that makes the system learn.

DEC-NNN log one-way / two-way doors DACI framework kill criteria ADR sub-type
# Authority — Blacksmith takes root control with TTL $ forge blacksmith god-mode --ttl 600 # 10-min window, auto-expires BLACKSMITH god-mode ENABLED · expires 2026-05-15T22:04:00Z · audit=logging # Structure — lifecycle gates enforce phase sequence $ forge quench metal # runs 80+ B-* dims, rolls up to 6 bands POLISH ██████░░░░ 65 RED · B-LEGAL 4.0 · forge bands metal POLISH to drill # Permanence — decision record survives nuke + restore $ forge decision new metal # scaffold DEC-NNN, DACI, door-type DEC-007 opened · door=ONE-WAY · review-trigger required before closing
// Get Started in 90 Seconds

From zero to verified memory in four commands.

Install the CLI, kindle your Anvil, smelt existing docs into memories, and strike your first ground truth.

Quick start

Four commands. Your AI has memory. It never confidently uses information it shouldn't trust.

Apache 2.0 · Node 20+ · macOS / Linux / Windows

# install $ npm install -g @forge/cli # kindle your anvil $ forge anvil kindle # smelt existing docs into memories $ forge anvil smelt ./docs/decisions/ # strike your first ground truth $ forge anvil strike "This project uses Next.js 15 with App Router" # check memory health $ forge anvil assay ✓ 47 memories · 0 conflicts · 2 stale · health: good
// The Full Anvil CLI Surface
$ forge anvil strike "We use row-level security in Supabase for all user data" $ forge anvil draw --type project $ forge anvil seek "auth decisions" $ forge anvil temper mem_01J3X... # reset confidence to 1.0 $ forge anvil fractures # list everything needing resolution $ forge anvil assay # health: memory count, stale, conflicts $ forge anvil cast # dump to markdown (offline portability)
// Confidence Decay - The Feature No One Else Has
AgeStatusConfidenceAI Behavior
0-7 daysFresh0.9 - 1.0Used directly
8-30 daysAging0.7 - 0.9Used with timestamp
31-60 daysStale0.4 - 0.7Surfaced with warning
60+ daysArchive< 0.4Flagged, not used

We invented the half-life of AI memory. No other AI dev tool does this.

// The 8-Phase FORGE Lifecycle
1. DiscoverWhat's the problem
2. FrameProblem brief
3. PlanPRD + ADRs + Design
4. ReviewCouncil pressure-test
5. BuildEngineer implements
6. VerifyQA + Auditor + evals
7. Ship ⬢HARD GATE
8. RetroLoop closes

One hard gate: verify → ship. Everything else is advisory. Single-gate methods get respected.

// The 6 Audit Bands

Every dimension rolls up to six numbers you can hold in your head.

80+ B-* audit dimensions collapse into exactly 6 named bands - designed to fit Miller's 7±2 so a tired operator can read the verdict at 6am without translation. Anvil verdict = worst band. Three-color system: GREEN ≥85, YELLOW 70-84, RED <70.

HEAT
"Is work actually happening?"
Measures activity cadence and momentum. No touches in 14+ days triggers Cold Forge failure mode. Tracks commit frequency, phase progression, and operator engagement.
HARDNESS
"Does it hold up under pressure?"
Tests, coverage targets, SLO adherence. Shipping without the verify phase triggers Quench Skip. Coverage is a CI gate, not a suggestion. Flaky tests are bugs.
FIT
"Are we building what we said?"
Scope drift vs. the original brief. Every out-of-scope feature needs an ADR or it hits FIT. Tracks story alignment, brief coverage, and undocumented pivots.
WEAR
"Are we accumulating drag?"
Cost per customer, toil score, tech debt budget, LLM token attribution. Anvil Sprawl (5+ projects, none shipping) is a WEAR failure. FinOps tracked to the cent.
GRAIN
"Is the code well-structured?"
Architecture integrity, dependency hygiene, naming conventions, DTCG token conformance. ADR coverage. No magic strings. GLOSSARY.md aligned to types/index.ts.
POLISH
"Is it complete?"
Documentation freshness, UX quality, accessibility, legal compliance, AI-tell density (CQS ≥85). RUNBOOK sections older than 90 days fail automatically.

Anvil verdict = worst band score. A GREEN product with a RED WEAR band is a RED anvil.

GREEN≥85 - ship YELLOW70-84 - flag RED<70 - block
// Why Forge Wins

Structural advantages no competitor can copy overnight.

These are not marketing spin. They are architectural decisions baked in from day one.

Infrastructure, Not a Feature

No other AI dev methodology has a local, anti-hallucination memory DB baked in. Every other tool - BMAD, Cursor, Copilot Workspace, Devin - treats memory as a scratchpad. Forge treats it as infrastructure. Memory with confidence decay, conflict detection, and a knowledge graph is not something you bolt on later. It has to be designed in from day one.

One Sentence That Sticks

"Forge remembers your project accurately - it verifies, ages, and conflicts-checks every memory." Every developer has been burned by an AI that confidently used stale context. Forge is the first methodology that solves that structurally, not just by asking the model to "try harder."

Full Transparency

Ground truth lives on your machine, not a black-box cloud. Every memory Forge stores is a plain file on your laptop - readable, exportable, and deletable without asking permission. Apache 2.0 means the entire memory engine is auditable. The AI your security-conscious colleagues will actually approve.

Forkable and Extendable

Apache 2.0 means you can fork the whole thing, swap the embedding engine, extend the knowledge graph, or build adapters for new AI hosts. The methodology is plain text. The writs are versioned files. Nothing is locked in a proprietary format. You own the stack.

Host Portable

Built on Claude Code. Adapters for Codex CLI, Gemini CLI, GitHub Copilot, Cursor. Per-model overlays for Opus, Sonnet, GPT-5.4, o-series. The methodology stays the same; behavior adapts to your host and your model.

Compounds with Every Ship

The learning loop closes automatically. Past pitfalls inform new reviews. Every retro produces typed learnings with confidence + decay that future review-phase personas auto-query. Loop closure rate ≥10% tracked. Your methodology becomes your moat.

// The Blacksmith Authority System

The safest way to give an AI root access.

Destructive operations in Forge require earning, not just having, authority. The Blacksmith subsystem is a sudo-style, TTL'd authority layer with two-key safety, full audit trail, and automatic expiry. Authority never silently falls through. Authority never propagates to sub-shells.

Two-key Safety Model

Both environment variables must be set simultaneously to authorize destructive operations. Neither alone is sufficient. This prevents accidental automation from triggering irreversible actions.

# requires BOTH simultaneously - neither alone is enough $ FORGE_BLACKSMITH=1 FORGE_BLACKSMITH_YES=1 forge blacksmith nuke my-project AUTO-TARBALL: /archive/my-project-20260514-091200.tar.gz ... ✓ NUKE: my-project deleted. Audit log preserved at ~/.forge/blacksmith/audit.log # impersonate any persona - always logged $ forge blacksmith as @compliance-officer forge quench IMPERSONATE_BEGIN · persona=compliance-officer · by=blacksmith · ts=2026-05-14T09:12:00Z # auto-expiring gate bypass $ forge blacksmith bypass-gate my-project BYPASS_GATE granted for 24h. Expires: 2026-05-15T09:12:00Z
Class A
Destructive
nuke, reset, purge-buried. Auto-tarballs workspace before deletion. Tar failure aborts - workspace untouched. Never auto-pruned.
Class B
Override
override-band, bypass-gate, ship --dangerously. TTL-gated. Logged with reason string. Auto-expires - no manual cleanup required.
Class C
Impersonation
as <persona> <cmd>. IMPERSONATE_BEGIN/END event pairs in audit log. FORGE_IMPERSONATED_BY env var propagated downstream.

Audit log: ~/.forge/blacksmith/audit.log · mode 0600 · 15+ event types · survives anvil deletion · forensic record persists permanently.

Apprentice Smith Master Smith Master Forger Blacksmith - earned by running 2+ Anvils
// Named Failure Modes

Forge names the failure before it happens.

Seven named anti-patterns with associated band triggers. If your project is heading toward one, the audit catches it before it becomes a crisis.

Cold Forge

→ HEAT failure

No commits, no phase progression in 14+ days. The project has gone cold.

Over-tempered

→ HARDNESS failure

Soft-pass audits. Tests exist but don't actually verify the acceptance criteria.

Slag Drift

→ FIT failure

Shipping features outside the original brief without an ADR to record the decision.

Spark-blindness

→ POLISH failure

RUNBOOK and README last updated 90+ days ago. Operator working from memory.

Anvil Sprawl

→ WEAR failure

5+ active projects, none shipping. Portfolio spreading attention below the threshold of progress.

Quench Skip

→ HARDNESS failure

Shipped without running the verify phase. fship-safe blocks this - unless bypassed.

Hammer Fatigue

→ WEAR + HEAT

Same anvil, no forward progress. Effort spent without moving the phase needle.

// vs. The Alternatives

The missing middle.

AI-first tools give you speed but no structure or memory. Spec-driven methods (BMAD, GSD) give you structure but still no memory. Forge is the only methodology that ships all three.

Capability Vanilla Claude / Cursor BMAD / GSD FORGE
Ground-truth memoryNone - resets every sessionNone - docs drift fastAnvil: confidence decay + contradiction detection, local JSON
Memory health dashboardNoneNoneforge anvil assay - count, stale %, conflicts, confidence map
Session handoffContext lost on closeManual notesquench / kindle / where - structured exit, cold resume in <60s
Hard gates05-10 (slow, bypassed)1 - verify then ship
Typed personasNoneRoles, not enforceable17 canonical + Blacksmith apex persona
Cross-model auditNoneNoneGPT reviews Claude - 15-20 catches per plan
Adaptive gatingNoneAlways-on or skipped entirelyZero-findings personas auto-skip, hit-rate tracked
Learning loopNoneRetros, not auto-appliedTyped learnings, decaying, auto-injected into future reviews
Host portabilityTool-lockedPlatform configs, no adapter systemClaude / Codex / Gemini / Copilot / Cursor via adapter transforms
Audit dimensionsNoneManual checklists80+ B-* dims - 6 bands - one verdict
Operational knowledge baseNoneAgent role files41 writs - versioned, cross-referenced, CLI-accessible
Reusable skill libraryNoneAgents, not portable190+ certified skills, multi-host slash-command handlers
Rebuild resilienceNoneNoneGURU: 3-zone spec, Cold Iron drills, frequency ratchet, cron-verified
Egress / security hooksNoneNoneAllowlist + deny-unreviewed-skills + HITL checkpoint
Compliance evidenceNoneManualSOC 2 pack, EU AI Act, SLSA Build L2
Root authority modelNoneNoneBlacksmith: two-key safety, TTL, full audit log
CLI depthNoneNone190+ commands across 3 tiers
Open sourceClosed / proprietaryMIT / openApache 2.0 - read it, fork it, trust it
// The Writ System

Forge's built-in protocol library. 7 categories. One coherent system.

Forge ships with a structured library of writs — versioned, cross-referenced operating protocols covering every function a modern AI-native team needs. Not templates. Not starters. Each writ ships with an Implementation Kit, 30-Day Adoption Playbook, Maturity Scorecard (L0–L5), Worked Example, and CI-runnable Audit Fixtures. Every protocol is v1.0.1 and points to the Forge Pyramid for authority, sequencing, and record-keeping.

CategoryWhat it coversProtocols
forge_writ PyramidThe 3 apex protocols that govern all others — authority, structure, permanenceblacksmith · lifecycle · decision
operation_writForge operational layer — personas, agents, skills, knowledge, retro, nukeforge · agent · brain · skills · knowledge · operator_self · retro · nuke · cli_reference
technical_writEngineering — ops, testing, AI, data, architecture, release, migration, backup13 protocols
platform_writMobile and design — iOS, Android, product design3 protocols
compliance_writLegal and regulatory — security, audit, legal, compliance4 protocols
marketing_writGrowth and brand — marketing, brand, communication, research, sales, customer6 protocols
business_writFinance and operations — finance, finops, support, crisis, people, startup6 protocols
$ forge view writ # list all protocols by category $ forge view writ --category forge_writ # the 3 pyramid protocols $ forge view writ blacksmith # apex authority protocol $ forge view writ --category technical # all 13 engineering protocols
// Security & Compliance

Built for the security team, not around it.

Forge ships with a complete security and compliance stack - not bolted on, baked in. 150+ dedicated skills for regulatory automation, supply chain integrity, and enterprise readiness.

Always Local

Ground Truth Never Leaves

All memory lives in your project directory as a plain JSON file. No server, no vendor, no exfiltration risk. We architecturally cannot see your project's context.

  • 100% local by default - no cloud required
  • Readable, exportable, deletable without asking permission
  • Survives vendor shutdown - your Anvil keeps working
  • detect-secrets pre-commit hook auto-installed
Evidence Ready

SOC 2 & Regulatory

Forge generates real compliance artifacts, not documentation theater. The audit log is forensic-grade. The evidence pack is dated and integrity-verified.

  • forge evidence - SOC 2 Type II evidence pack
  • Append-only audit log (mode 0600)
  • EU AI Act: classify, check, ui-check skills
  • GDPR-friendly telemetry (EU PostHog, self-hostable)
  • DPA review, DPOP audit, data classification skills
Supply Chain

SLSA & Provenance

Every release ships with cryptographic provenance. NPM packages are signed. SBOM is generated and attested. Supply chain attacks are a known class - Forge treats them that way.

  • SLSA Build L2 - forge-slsa-build-l2
  • SBOM generation + attestation
  • NPM provenance signing on every release
  • forge-supply-chain-verify
  • forge-secret-rotate + forge-secret-rotate-dynamic
// 150+ Security & Compliance Skills
# security forge-threat-model forge-llm-red-team forge-fuzz-test forge-mutation-test forge-zero-trust-audit forge-passkey-deploy forge-k8s-pss-audit forge-ai-tell-scan # supply chain forge-sbom-generate forge-sbom-attest forge-slsa-build-l2 forge-supply-chain-verify forge-secret-rotate forge-secret-rotate-dynamic # regulatory forge-eu-ai-act-check forge-eu-ai-act-classify forge-eu-ai-act-ui-check forge-dpa-review forge-dpop-audit forge-continuous-compliance forge-data-classify
// 130+ Commands - Three Tiers

The deepest CLI surface in AI development tooling.

Two binaries. Seventy-eight aliases. Three access tiers. Tab completion across every command, subcommand, project slug, story ID, and persona name.

// forge anvil - memory

kindle strike draw seek temper fractures assay cast smelt

// forge - lifecycle verbs

brief survey strike shape temper quench stamp ship

// forge - project & audit

init audit audit --fix audit-all bands doctor adr new adr graph story new gen tests persona declare portfolio recon-all backup-all evidence

// forge blacksmith - authority

god-mode as <persona> nuke restore bypass-gate override-band ship --dangerously audit-log
$ forge open my-saas --claude # pipe project briefing to Claude CLI - morning briefing workflow $ forge deploy precheck # pre-flight checklist before deploy $ forge deploy verify # curl health endpoints; non-zero on failure $ forge memory search "auth" # search across ALL projects $ forge status --json # machine-readable output for CI $ forge completion --shell zsh # tab completion for everything $ forge telemetry status # see exactly what data (if any) is sent
// Ecosystem & Roadmap

A platform, not just a tool.

Forge ships today with Claude Code, Codex CLI, and Gemini CLI. The roadmap extends it into every corner of the developer workflow - IDE, project management, deployment, mobile, and community.

Today

Core Platform

Claude Code + Codex CLI + Gemini CLI. 150+ commands. 17 personas. Forge Anvil memory. 41 writs. 6-band audit. GitHub Action. Astro docs site. CI/CD pipeline.

Claude Code Codex CLI Gemini CLI GitHub Action
Month 3

IDE & Project Management

Linear ticket cross-linking. Notion connector. VS Code extension with NICHE/ROI sidebar. Cursor MDC rules (shipping now in preview). GitHub Copilot adapter.

Linear Notion VS Code GitHub Copilot
Year 1

One-Command Bootstrap & Community

Vercel + Supabase + Fly.io one-command provision. Obsidian plugin. Slack and Discord apps. Raycast extension. Smithy community platform and certification ladder (Smith $399 → Master Forger $14,999).

Vercel Supabase Obsidian Slack Raycast Certification
Year 2+

Platform & API

Public REST API. Webhooks bridge. Plugin marketplace with 3-tier quality system. iOS and Android native apps. forgedev.ai developer community hub.

REST API Plugin Marketplace iOS Android forgedev.ai
// The Forge Anvil Guarantee

Your AI will never confidently use information it shouldn't trust.

If a memory is stale, you'll know. If two memories conflict, you'll see it. If a fact hasn't been verified in 60 days, it won't silently poison your next decision.

Other tools give you a smarter AI.

Forge gives you an AI you can trust.

// Open Source

Free. Forever. No vendor lock-in.

Forge is Apache 2.0 - read it, fork it, trust it. Install in 30 seconds. Your memory lives on your machine. No account, no subscription, no black box.

★ Star on GitHub → See the Install Guide
$ npm install -g @forge/cli

Want updates on new releases and community news?

No spam. Unsubscribe anytime. Apache 2.0 - read the source.