GitHub - pablotech80/psdm-framework: Specification-first governance framework for AI-assisted software projects.

Keep AI work inside the objective, repository, scope, and authority you defined.

Preserve direction across Codex chats in the same project. Stop unauthorized transitions. Verify the result stayed in bounds.

Riscala is an open-source, local, dependency-free governance layer for developers building software with AI. Beta.6 focuses on same-project Codex chat continuity: when you open a new Codex chat in the same repository, it can recover the active repository, objective, work mode, allowed scope, prohibitions, stop conditions, handoff, and exact next action without you rebuilding context manually.

The governing rule is simple: a new request does not automatically expand prior authority.

Documentation: start with the dependency-free web guide at docs/index.html, then use the Markdown reference for deeper governance contracts.

Riscala does not replace your coding agent, tell it how to program, choose product direction, or approve its own work. Coding agents can reason about implementation; the beta.6 target is to check their proposed actions against the boundary you defined and require an explicit transition when repository, objective, mode, or authority changes.

PSDM is the internal governance method. Its risk-scaled controls remain available when a change reaches security, data, AI, infrastructure, or production boundaries, but low-risk daily work does not require initialization or mandatory artifacts.

Riscala is currently beta software and is still distributed through the compatibility package @ptechsolution/psdm-framework. It is a local CLI and GitHub Action, not a hosted platform.

What beta.6 does:

records the active repository, objective, mode, scope, forbidden actions, and stop conditions;
persists that boundary outside the chat;
lets a fresh Codex chat in the same project recover the current handoff and next action;
requires an explicit transition when repository, objective, mode, scope, or authority changes.

Install and create the active boundary for the current task:

npm install -g @ptechsolution/psdm-framework@beta
riscala work init "add search to the customer list" --mode implement --files src/search/**,tests/search/**
riscala work show
riscala adapters init codex

After meaningful progress, the Codex adapter requires the coding agent—not the developer—to record a compact handoff:

riscala work handoff \
  --completed "implemented scope enforcement" \
  --validation "focused tests passed" \
  --decisions "keep enforcement deterministic" \
  --questions "none" \
  --pending "fresh-chat validation" \
  --next "open a new chat without context"

The current handoff replaces stale progress while lifecycle history records that a handoff occurred. A fresh Codex chat in the same project must run riscala work show, recover the exact next action, and follow it unless it conflicts with the active boundary or a newer explicit developer instruction.

Riscala keeps a revisioned canonical copy in the local user configuration and treats .riscala/ACTIVE_WORK.md as its visible mirror. riscala work show refreshes stale snapshots before an agent evaluates authority. This supports local continuity between Codex chats that use the same project and machine; remote or cross-device continuity is not implied.

adapters init codex is the primary beta.6 path. The broader adapters init command can also install compatible boundaries for Claude Code, Cursor, Windsurf, OpenCode, and Antigravity, but those are secondary compatibility surfaces, not the core beta.6 workflow.

Before acting, every adapter must explicitly compare the new request with the active repository, objective, mode, allowed paths/actions, forbidden actions, and preservation rules. Each dimension is classified as aligned, conflicting, or unresolved. Aligned work continues without extra approval noise; a conflict or material uncertainty stops for a transition or missing decision. This is the control that correctly blocked a recovered next action when it belonged to a different objective.

Repository comparison is anchored before following any requested cd, target path, or workspace switch. The requested repository cannot supply the Active Work used to authorize entry into itself; changing repositories always requires resolving a transition from the initial repository boundary first.

The npm @beta tag installs 1.0.0-beta.6.

The compatibility executable remains available:

psdm audit

Why Riscala Exists
Supporting Judgment Loop
What Riscala Provides
Status
Install
CLI
Quick Start
Model And Tool Independence
Knowledge As Code Layer
Common Workflows
Local Validation
Contributing And Security
What PSDM Is Not
Examples
Configuration
Feature Artifacts
Backend And Platform Governance
Change Levels
Production Gate
GitHub Action
Design Principles
Current Limitations

Why Riscala Exists

AI can produce correct code and still be ungovernable: it can change the wrong repository, turn an experiment into implementation, expand scope, or continue after the developer's authority ended. Long conversations may recover control, but that direction is easily lost when the next chat starts.

An AGENTS.md provides general repository rules. It does not preserve the concrete boundary of today's task. Riscala keeps that active work compact and explicit while leaving final authority with the developer.

The beta.6 direction is defined in docs/ACTIVE_WORK.md: continuity preserves direction; control stops or verifies actions at the boundary.

Supporting Judgment Loop

Inside an authorized Active Work boundary, the existing read-only judgment path remains available:

flowchart TD
    A[Developer intent] --> B[Riscala impact]
    B --> C[Developer chooses direction and constraints]
    C --> D[AI-assisted implementation]
    D --> E[Stage expected files]
    E --> F[Riscala review]
    F --> G{Evidence matches intent?}
    G -->|No| C
    G -->|Yes| H[Developer validates and decides]

What Riscala Provides

A compact Active Work contract for continuity across Codex chats in the same project and explicit transitions when authority changes.
Before-action boundary checks and after-action compliance verification as the beta.6 target direction.
A read-only Judgment Brief before coding: observed facts, inferred impact, options, trade-offs, uncertainty, and decisions reserved for you.
A staged Decision Review after coding: expected versus actual scope, sensitive surfaces, dependency changes, and missing evidence.
learn, balanced, and concise explanation density for junior, senior, and staff-level workflows without changing authority or safety semantics.
Optional PSDM adoption for durable specifications, architecture, security, testing, deployment, operations, and CI controls when the project needs them.
Stable JSON output and compatibility with the existing psdm executable and npm package.

Status

Latest published beta: 1.0.0-beta.6.

Current main tracks the beta.6 line centered on governable Active Work. The existing judgment commands and PSDM compatibility surface remain available.

Install

Install the published beta from npm:

npm install -g @ptechsolution/psdm-framework@beta

Then run:

riscala help

The npm @beta tag installs 1.0.0-beta.6.

The existing psdm executable remains supported with identical commands and behavior during the Riscala migration. Both riscala help and psdm help work from the npm beta package.

For development from this checkout:

npm install -g .

Then run:

riscala help

The command reference below uses the primary riscala executable. Use psdm only when compatibility with older automation is needed; command behavior is identical.

CLI

Active Work

Create or read the compact Active Work boundary used across Codex chats in the same project:

riscala work init "add Google OAuth while preserving passwords" --mode design --files src/auth/**,tests/auth/**
riscala work show

work init creates .riscala/ACTIVE_WORK.md once and refuses to overwrite it. The file records repository, objective, mode, optional allowed paths, forbidden work, invariants, stop conditions, and categorized context. When --files is present, /status compares staged, unstaged, and untracked paths with that boundary and reports exact violations. Without --files, Riscala keeps the lightweight whole-repository behavior.

Build a read-only Judgment Brief from a proposed change without initializing PSDM artifacts:

riscala impact "add Google OAuth login while preserving passwords"

Adjust explanation density without changing the underlying facts or safety semantics:

riscala impact "add Google OAuth login" --guidance learn
riscala impact "add Google OAuth login" --guidance balanced
riscala impact "add Google OAuth login" --guidance concise
riscala impact "add Google OAuth login" --json

The brief separates observed repository evidence, inferred impact, options, advisory recommendation, uncertainty, and decisions reserved for the developer. impact is read-only, does not require riscala init, and never creates or simulates an owner decision.

After implementation, compare a CLI-declared expected file scope with the real Git index:

git add src/auth/login.mjs tests/auth/login.test.mjs
riscala review "add Google OAuth login" --staged \
  --files src/auth/login.mjs,tests/auth/login.test.mjs

review reports staged files outside the declared scope, expected files that are missing, observed and unexpected auth/schema/AI/config/deployment surfaces, package dependency changes, and the absence of supplied validation results. scope_aligned_evidence_unverified means only that staged scope aligns; it does not claim validation or approval. The Change Envelope is an advisory CLI input with authorityVerified: false; Decision Review never approves, commits, or establishes human authority.

Human-readable review output restores the language persisted in .riscala/ACTIVE_WORK.md; JSON keys remain English. In nested projects or monorepos, expected paths entered relative to the selected project are normalized to the Git root before comparison, avoiding false scope drift.

Repository

riscala audit [target] [--json] [--feature <name>] [--config <path>]
riscala check [target] [--json] [--feature <name>] [--config <path>]
riscala validate [target] [--json] [--feature <name>] [--config <path>]
riscala report [target] [--json] [--feature <name>] [--config <path>]
riscala inspect --staged [--json] [--target <path>] [--config <path>]

Interactive Shell

riscala shell [target] [--config <path>]
riscala action prepare git.commit [--target <path>] [--config <path>] [--json]
riscala approval verify git.commit --receipt <path> [--target <path>] [--config <path>] [--json]
riscala approval enforce git.commit [--receipt <path>] [--target <path>] [--config <path>] [--json]
riscala hook <install|remove|status> pre-commit [--target <path>] [--json]

The dependency-free shell is an operational governance console. It manages Active Work while showing branch, working-tree state, and active PSDM policy:

/help
/work [inspect|experiment|design|implement|release] <objective>
/work transition <mode> <new objective>
/work continue
/work close
/language es|en
/init preview
/init confirm
/uninstall preview
/uninstall confirm
/status
/check
/report
/inspect
/classify <change description>
/pr-checklist <change description>
/hook-status
/action
/advanced
/exit

The main palette keeps the everyday workflow concise. /advanced opens an arrow-navigable submenu for /approval, /audit, /impact, /review, and /validate. These commands remain directly executable and script-compatible; only their menu presentation changes.

When AGENTS.md already exists, /init confirm preserves its content and appends only missing PSDM Required Reading, Boundaries, and Escalation sections. Repeating initialization does not duplicate the integration block.

For new or untouched PSDM documents, /init confirm inspects a bounded repository inventory and generates a project-specific baseline from detected languages, frameworks, directories, validation commands, deployment signals, and sensitive surfaces. It excludes secrets, .env files, data directories, virtual environments, dependencies, and build output. Existing edited documents are never overwritten; unknown business decisions remain explicit review questions.

/uninstall preview lists the Riscala state, untouched templates, configuration, and adapter blocks that can be removed from the current project. /uninstall confirm modifies the repository: it removes recognized Riscala-managed files and blocks while preserving user-modified documents. It does not delete application code or uninstall the global npm package.

The first screen restores .riscala/ACTIVE_WORK.md. If none exists, /work <objective> creates it with implement mode by default. /work transition records a proposed boundary without applying it; /work continue accepts it explicitly; /work close ends the work. Timestamped history remains in the file. Source code is changed by Codex or your preferred coding agent.

The shell starts in Spanish when the system locale begins with es; otherwise it uses English. /language es|en or /lenguaje es|en stores a global Riscala preference, applies it immediately, and restores it across repositories. Language changes presentation, never policy meaning or JSON keys.

Interactive terminals use Ptech cyan (#00A8E8) with a light accent (#38BDF8) for the Riscala frame and prompt. Color is automatically disabled for pipes, non-TTY output, TERM=dumb, and the NO_COLOR convention.

Type / at the interactive prompt to open the dependency-free command palette. Filter by typing, navigate with ↑/↓, open submenus with → or Enter, return with ← or Esc, and complete with Tab. Piped sessions preserve the original line-oriented behavior.

The Active Work boundary is intentionally stricter than a chat suggestion. In a verified continuity test, a fresh Codex chat recovered the recorded next action but refused to implement it because it fell outside the still-active objective. That stop was correct: the next task required an explicit Active Work transition. Riscala therefore preserved developer control across the chat boundary instead of silently expanding authority.

Shell commands use the same fixed-width result panels. Each result has a clear title, semantic state, and a contextual next action where useful, so repeated commands remain visually consistent without changing the read-only security boundary. /audit and /init preview reuse the existing non-destructive audit engine. /check, /validate, and /report summarize baseline readiness. /classify and /pr-checklist prepare governance decisions from a described change. /hook-status, /action, and /approval expose the approval boundary without creating receipts, installing hooks, committing, pushing, or publishing.

Initialization

riscala init [target]
riscala init [target] --dry-run
riscala init [target] --feature <name>

Governance

riscala classify "<change description>" [--json] [--file <path>] [--files <path,path>] [--target <path>] [--config <path>]
riscala enforce "<change description>" [--json] [--max-level "Level 2"] [--file <path>] [--files <path,path>] [--target <path>] [--config <path>]
riscala pr-checklist "<change description>" [--json] [--file <path>] [--files <path,path>] [--target <path>] [--config <path>]

Architecture

riscala adr "<decision title>" [--json] [--target <path>] [--date YYYY-MM-DD] [--status Proposed]

Quick Start

Inside any greenfield or existing project:

# 1. Install the beta.
npm install -g @ptechsolution/psdm-framework@beta

# 2. Define the active boundary for the task.
riscala work init "add Google OAuth while preserving password login" \
  --mode implement \
  --files src/auth/login.ts,tests/auth/login.test.ts

# 3. Connect Codex to the boundary.
riscala adapters init codex

# 4. Open or continue Codex in this project.
riscala work show

A fresh Codex chat should recover the same Active Work by running riscala work show, then continue only when the request remains inside the recorded repository, objective, mode, scope, and authority.

Optional Judgment Review

Inside an authorized Active Work boundary, impact and review remain available as advisory, read-only checks:

# Think before implementation.
riscala impact "add Google OAuth while preserving password login" \
  --files src/auth/login.ts,tests/auth/login.test.ts

# After implementation, stage only the files you expect to review.
git add src/auth/login.ts tests/auth/login.test.ts

# Compare staged evidence with intended scope.
riscala review "add Google OAuth while preserving password login" --staged \
  --files src/auth/login.ts,tests/auth/login.test.ts

impact and review do not implement code, run tests, approve changes, or claim human authority. A scope-aligned review still reports validation evidence as unverified until you run and assess the relevant checks.

Use --guidance learn for more teaching, the default balanced mode for everyday work, or --guidance concise for compact senior/staff review. JSON output is available for automation.

Adopt the broader PSDM baseline only when durable project governance adds value:

riscala audit       # Preview adoption without writing files.
riscala init        # Create the selected governance baseline.
riscala validate    # Validate the adopted baseline.

See docs/GETTING_STARTED.md for greenfield, legacy, shell, and AI-agent examples.

Use riscala audit before initializing PSDM in an existing project. It does not modify files; it shows current state, what riscala init would create or skip, pros, cons, and recommendations.

If the repository already has AGENTS.md, Copilot, Cursor, Claude, Codex, skills, prompts, or AI instruction files, riscala audit reports adoption mode integrate and recommends preserving those files. During riscala init, PSDM creates docs/PSDM_ADOPTION.md so the integration plan is explicit.

riscala audit --json also emits aiReadiness, a stable contract for AI runtime readiness signals. It reports detected AI surfaces, governance artifact groups, gaps, and recommendations for guardrails, data classification, cost, latency, evals, prompt injection, PII, and tool security. Current detection covers common AI folders and manifest dependencies such as OpenAI, Anthropic, LangChain, LlamaIndex, vector stores, embeddings, and n8n. The contract is documented in docs/AI_READINESS_AUDIT.md.

riscala init also creates psdm.config.json. Existing files are skipped.

Model And Tool Independence

PSDM is independent from any specific model, provider, coding assistant, or init command. It does not replace claude init, Cursor rules, Copilot instructions, Codex instructions, custom skills, prompts, or agent runtimes. It gives them a shared governance layer.

Teams customize PSDM through psdm.config.json, AGENTS.md, docs/CHANGE_GOVERNANCE.md, docs/TOOL_REGISTRY.md, and AI guardrail docs. Tool-specific files should adapt those rules for a given assistant, but must not weaken PSDM change-level, security, data, deployment, approval, or release boundaries.

See docs/MODEL_AND_TOOL_INDEPENDENCE.md for customization examples.

Knowledge As Code Layer

PSDM treats knowledge as a first-class artifact. A project should not preserve only source code, but also intent, specifications, architectural decisions, business rules, agent instructions, workflows, prompts, verification criteria, and evolution notes as versioned knowledge assets.

Knowledge as Code is a transversal layer, not a new mandatory phase. Markdown and YAML in Git remain the source of truth. Obsidian can be used as an optional authoring tool; vector databases and graph databases are derived runtime indexes, not the primary record.

RAG retrieves semantically similar text fragments. A Knowledge Graph connects explicit entities and relationships such as rules, workflows, agents, systems, owners, and decisions. GraphRAG is an advanced evolution path, not a starting requirement.

See docs/KNOWLEDGE_AS_CODE.md for structure, tool roles, maturity levels, and an example knowledge note.

Common Workflows

Classify a Change

riscala classify "change Supabase RLS policy for client documents"

Expected result:

Estimated level: Level 3

Machine-readable output:

riscala audit --json
riscala validate --json
riscala classify "change Stripe webhook ownership validation" --json

Classify by description and touched files:

riscala classify "small cleanup" --file backend/auth/session.py

Configured risk paths can raise the level even when the description is vague:

Estimated level: Level 3

Inspect Staged Changes

Inspect the Git index without writing files or requiring a change description:

riscala inspect --staged

The command reports staged file status, applies a Level 1 minimum to any real file change, and raises the result when a configured riskPath matches. It does not inspect unstaged or untracked files.

Use JSON output for automation:

riscala inspect --staged --json

The JSON contract includes decision, git.changes, files, evidence, and classification. NO_STAGED_CHANGES is a successful no-op; NOT_A_GIT_REPOSITORY exits non-zero.

Open The Interactive Shell

Open the Active Work governance console in the current project:

riscala shell

Use /status to refresh project context and /inspect to review staged changes. Mutating slash commands remain blocked until trusted approvers and independent enforcement hooks are configured. See docs/INTERACTIVE_SHELL.md for the interface and safety contract.

Prepare And Verify Approval

Create a machine-readable record for the exact staged diff:

riscala action prepare git.commit --json

After a trusted external signer returns a receipt, verify it against the live Git index:

riscala approval verify git.commit --receipt ./approval-receipt.json --json

Riscala does not sign receipts. Signing must happen through a hardware-backed or separately authenticated channel. See docs/ACTION_RECORDS_AND_APPROVAL_RECEIPTS.md.

Install the local pre-commit enforcement hook after a trusted approver is configured:

riscala hook install pre-commit

The hook is optional. When Active Work declares --files, it blocks staged files outside those paths before evaluating approval requirements. /review reports the same violations without approving or committing anything. Without declared paths, existing lightweight commit behavior remains unchanged.

The hook consumes a valid receipt once and blocks invalid or missing Level 3/4 approval. It preserves existing unmanaged hooks instead of overwriting them. Local hooks can still be bypassed with Git options or filesystem control, so protected branches and remote required checks remain necessary for agent-resistant enforcement.

Solo Maintainer And Team Mode

Riscala separates local governance from repository ownership rules:

solo maintainer mode keeps required CI checks, admin enforcement, conversation resolution, and blocked force pushes/deletions, but does not require an approving reviewer;
team mode should add at least one required approving review from a maintainer with write access;
high-risk Level 3/4 changes still require content-bound approval when project policy says approval is required.

This keeps the product usable for one maintainer without weakening the path to team governance later.

Generate A PR Checklist

Generate a PR checklist:

riscala pr-checklist "change auth session validation" --file backend/auth/session.py

Enforce In CI

Enforce a maximum level in CI:

riscala enforce "small cleanup" --file src/index.mjs --max-level "Level 2"

Create An ADR

Create an ADR:

riscala adr "Adopt CI change level enforcement"

Local Validation

Run local fixtures:

npm test

Run the non-publishing release gate:

npm run release:check

Use npm run release:check -- --allow-dirty only when validating local changes before commit.

Beta release scope and exit criteria are tracked in docs/BETA_RELEASE_NOTES.md.

The public documentation index is docs/INDEX.md.

Contributing And Security

Public contribution expectations are in CONTRIBUTING.md and CODE_OF_CONDUCT.md.

Security reporting expectations are in SECURITY.md. Do not open public issues for secrets, credentials, exploitable behavior, private repository output, or customer data.

What PSDM Is Not

PSDM is not a hosted observability platform, a runtime security product, a secret scanner, or a replacement for owner approval. It provides governance artifacts, local validation, change classification, CI enforcement hooks, and AI-agent guardrail templates.

Examples

examples/nextjs-saas is a lightweight SaaS/AI fixture used by npm test.

It is not a runnable application. It exists to prove that PSDM can audit, initialize, and validate a downstream-like project with package metadata, prompt assets, and an AI-assisted backend surface without requiring external services or dependency installation.

Configuration

psdm.config.json is optional. When it is absent, PSDM uses the default baseline artifacts.

Minimal example:

{
  "profile": "backend-api"
}

Use the README to start. Use docs/CONFIG_SCHEMA.md for the full schema, supported fields, compatibility rules, AI policy fields, and riskPaths examples.

Supported profiles:

standard
framework
backend-api
ai-agent
saas
monorepo

Profiles add sensible default artifacts and risk paths for common repository types. Explicit config still wins for project-specific policy.

The ai-agent profile adds guardrail artifacts for AI runtime governance:

docs/AI_GUARDRAILS.md
docs/DATA_CLASSIFICATION.md
docs/COST_LATENCY_BUDGET.md
docs/PROMPT_INJECTION_TESTS.md
docs/AI_EVALS.md

These artifacts define policy, evidence, owners, gates, and accepted external reports. They do not turn PSDM into a native tracing, observability, or hosted eval platform.

Unsupported profile values fail validation instead of silently falling back to standard. The validation JSON still reports config.profile.name and config.profile.recognized so automation can surface the exact policy problem.

Schema stability rules are documented in docs/CONFIG_SCHEMA.md.

The optional ai block declares repository-level AI policy for PII, redaction, cost budgets, latency SLOs, tool registry expectations, eval requirements, prompt-injection testing, and human approval. null means the policy is intentionally not declared yet; invalid field types fail validation.

Use a non-default config path:

riscala validate --config ./governance/psdm.config.json

Feature Artifacts

For product changes that should not require rewriting the whole project baseline, create scoped artifacts:

riscala init --feature billing
riscala validate --feature billing

Default feature paths:

docs/features/<feature>/PROJECT_BRIEF.md
docs/features/<feature>/SPEC.md
docs/features/<feature>/ARCHITECTURE.md
docs/features/<feature>/SECURITY.md
docs/features/<feature>/TESTING.md

Backend And Platform Governance

PSDM governs the whole repository, but its strongest controls usually apply to backend and platform surfaces:

authentication and authorization;
payments and billing;
database migrations;
infrastructure and deployment;
AI agents and RAG pipelines;
CI/CD workflows.

These controls live in riskPaths. A matching path raises the minimum change level even when the textual change description looks low risk.

riscala inspect --staged obtains touched paths directly from the Git index, so developers do not need to repeat them through --file or --files before review.

riscala validate fails when riskPaths contains malformed rules. Invalid risk path rules are ignored by classification so a broken local policy does not crash the CLI.

Change Levels

Level	Meaning	Governance
Level 0	Safe trivial change	Diff review only.
Level 1	Local low-risk code change	Scope note, relevant docs, narrow validation.
Level 2	Product behavior change	Product spec, tasks, testing, architecture review if needed.
Level 3	Security / data / payment / AI change	Spec, architecture, security, testing, owner approval.
Level 4	Deployment / operations / infrastructure change	Deployment, operations, rollback plan, explicit production confirmation.

Production Gate

Production execution is never implied.

Commands that mutate production require exact owner confirmation:

CONFIRM PRODUCTION DEPLOY

GitHub Action

This repository includes a composite GitHub Action:

Replace <owner> with the public GitHub owner for the repository.

steps:
  - uses: actions/checkout@v4
  - uses: <owner>/psdm-framework@main
    with:
      target: .

The action writes psdm-report.json and fails when validation has blocking failures.

Enable change-level enforcement:

steps:
  - uses: actions/checkout@v4
  - uses: <owner>/psdm-framework@main
    with:
      target: .
      enforce-change-level: 'true'
      change-description: ${{ github.event.pull_request.title }}
      files: src/index.mjs,docs/SPEC.md
      max-level: Level 2

The action writes psdm-enforcement.json and fails when the classified change exceeds max-level.

Design Principles

Risk-scaled governance.
Specification before significant implementation.
Knowledge as a versioned project asset.
Explicit AI-agent boundaries.
Agents justify meaningful mutations but cannot authorize their own high-risk actions.
Security-sensitive work requires security context.
Deployment-sensitive work requires rollback context.
Documentation must support delivery, not replace it.

Current Limitations

This beta does not yet provide:

Full AI agent security runtime guardrail enforcement.
Tool registry enforcement.
Deep code-level semantic AI readiness detection.
SBOM or supply-chain scanning.
Deep semantic validation of specs.
Remote approval service, remote replay persistence, or hardware signing ceremony integration.

Freshly initialized templates intentionally contain placeholders. riscala validate reports them as warnings and returns METHOD_BASELINE_REVIEW_REQUIRED until the artifacts are filled with project-specific content.

See docs/ROADMAP.md.

Built by Ptech AI Applied Lab

PSDM is the open governance framework behind the PTECH SPEC-DRIVEN METHOD.

Specification-first AI Engineering.

https://ptechsolution.net

Name		Name	Last commit message	Last commit date
Latest commit History 165 Commits
.github		.github
ADRs		ADRs
assets		assets
bin		bin
docs		docs
examples/nextjs-saas		examples/nextjs-saas
governance/keys		governance/keys
scripts		scripts
src		src
templates		templates
tests		tests
.gitignore		.gitignore
.npmignore		.npmignore
AGENTS.md		AGENTS.md
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
CONTRIBUTING.md		CONTRIBUTING.md
HANDOFF.md		HANDOFF.md
LICENSE		LICENSE
README.md		README.md
ROADMAP.md		ROADMAP.md
SECURITY.md		SECURITY.md
TODO.md		TODO.md
action.yml		action.yml
package.json		package.json
psdm.config.json		psdm.config.json

Folders and files

Latest commit

History

Repository files navigation

Table Of Contents