Skip to content
LloydCoderPublic

About

Production-grade AI engineering and FDE platform for enterprise automation, AI infrastructure, security, and intelligent systems.

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Repository files navigation

Tinlance

Tinlance

AI engineering and Forward-Deployed Engineering for teams that need to ship secure, production-oriented AI systems around real business workflows.

CI Secret Scan Production Smoke

Important

Tinlance is publicly visible for engineering transparency and verification, but the repository is proprietary. Public visibility does not grant an open-source reuse license. See LICENSE.

Visual proof

Tinlance platform visual

The visual above is generated from the current application code. It is a product/architecture visual, not live telemetry or simulated operational data. The live public engineering surface is tinlance.com/engineering.

Why Tinlance

Tinlance connects commercial discovery, customer delivery, engineering execution, security controls, evaluation, and productization without creating parallel authorities for identity, authorization, audit, or execution.

Differentiator What it means in this repository
Evidence-first Public claims are classified as IMPLEMENTED, TESTED, VALIDATED, EXPERIMENTAL, or PLANNED.
Governed execution Agents and automation operate through explicit identity, authorization, policy, approval, and audit boundaries.
FDE boundary Tinlance owns the customer-facing platform; the authenticated FDE API separates the web application from FDE Mastery execution.
Security in the loop Dependency audit, secret scanning, SAST, DAST, SBOM, container validation, and AI/agent regression gates are part of CI.
Productization path M1/M3/M4/M5/M6/M7/M8/M9/M10/M11/M12/M13/M14 are represented as bounded platform capabilities rather than unrelated feature silos.

Quick start

The fastest way to inspect the web application locally:

git clone https://github.com/LloydCoder/Tinlance.git
cd Tinlance
corepack enable
pnpm install --frozen-lockfile
pnpm --filter @tinlance/web dev

Open http://localhost:3000.

The public pages are designed to run without production credentials. Authenticated, database-backed, billing, email, FDE, and customer-workspace flows require the corresponding environment and service configuration.

Installation

Prerequisites

Component Current repository requirement
Node.js 22.x in CI
pnpm 10.14.0
Python 3.12 for apps/fde-api
PostgreSQL 16 for application CI/migrations
Docker Required for container, SBOM, Trivy, and ZAP validation

Web application

corepack enable
pnpm install --frozen-lockfile
pnpm --filter @tinlance/web dev

FDE API

cd apps/fde-api
python -m venv .venv
source .venv/bin/activate
pip install -e '.[test]'
uvicorn app.main:app --host 127.0.0.1 --port 8000

The FDE API exposes liveness at /health and readiness at /ready. Interactive API documentation is disabled by default and can be enabled deliberately with FDE_ENABLE_DOCS=true.

Docker

The repository contains production-oriented Dockerfiles at apps/web/Dockerfile and apps/fde-api/Dockerfile. CI builds both images and validates non-root execution, health checks, vulnerability scans, and SBOM generation.

Usage

Basic web development

pnpm --filter @tinlance/web dev

Use the public routes to inspect the commercial and engineering surface:

  • /assessment
  • /engineering
  • /engineering/fdse
  • /fde-mastery
  • /products
  • /security
  • /work

Run the main validation suite

pnpm typecheck
pnpm lint
pnpm test
pnpm build
pnpm format:check

FDE API validation

cd apps/fde-api
pip install -e '.[test]'
pytest -q
ruff check .
python -m pip check

The FDE API is an authenticated service boundary. Production execution requires a service credential, tenant-context signing, allowed-host configuration, and upstream FDE Mastery authentication. Do not expose those credentials in source or client-side code.

Configuration

The application uses optional development configuration but validates a stricter set in production.

Variable Purpose Default / requirement
NEXT_PUBLIC_APP_URL Canonical application origin Optional in development; required in production
BETTER_AUTH_URL Better Auth origin Optional in development; must match NEXT_PUBLIC_APP_URL in production
BETTER_AUTH_SECRET Session/signing secret Optional in development; at least 32 characters in production
DATABASE_URL Neon/PostgreSQL connection Required for database-backed flows and production
UPSTASH_REDIS_REST_URL Distributed rate limiting Required in production
UPSTASH_REDIS_REST_TOKEN Distributed rate limiting credential Required in production
PAYSTACK_SECRET_KEY Billing integration Required only when billing flows are enabled
RESEND_API_KEY Transactional email Required only when email flows are enabled
RESEND_FROM_EMAIL Transactional sender Required only when email flows are enabled
CRON_SECRET Internal scheduled-worker protection Required when outbox/cron production flows are enabled
FDE_ALLOWED_HOSTS FDE API trusted-host allowlist localhost,127.0.0.1,testserver by default
FDE_SERVICE_TOKEN Tinlance-to-FDE API authentication Required for authenticated execution
FDE_ENV FDE API execution mode production by default
FDE_ENABLE_DOCS Enables FastAPI docs false by default

Never place secrets in NEXT_PUBLIC_* variables.

Features

Area Current capability
Identity Better Auth, sessions, organizations, memberships, invitations, RBAC
Customer workspace Projects → assessments → findings → evidence → reports → remediation
Commercial Assessments, leads, opportunities, proposals, booking, billing
API Stable v1 contract, scoped organization credentials, idempotency, ETags, webhooks
FDE Authenticated FastAPI gateway with tenant/domain validation and upstream OAuth
MCP Registry, policy, security, server, approvals, and governed tool boundary
AI security M7 policy/risk/approval/audit control plane
Evaluation M8 deterministic and regression assurance
Agent runtime M9 identity-bound executions, bounded capabilities, memory, approvals, recovery, evidence
Knowledge M10 tenant-scoped knowledge and retrieval controls
Intelligence M13 permissioned knowledge-moat workflow
Productization M14 observation → pattern → opportunity → playbook → evaluation → productization
SDKs TypeScript and Python clients for the stable API surface
CI/security Tests, typecheck, lint, formatting, dependency audit, secret scan, Semgrep, ZAP, Trivy, SBOM, container validation

Architecture

The repository is a monorepo:

Tinlance
├── apps/web        Next.js application, portal, admin, and API
├── apps/fde-api    Authenticated FastAPI execution gateway
├── packages/       TypeScript and Python SDKs
├── docs/           Architecture, API, security, operations, authority, and evidence
└── .github/        CI, security, dependency automation, issue forms

The central execution relationship is:

Tinlance web
    |
    +--> M5 API / Core
    |       |
    |       +--> M7 Security / authorization
    |       +--> M6 MCP / tools
    |       +--> M8 Evaluation
    |       +--> M9 Agent Runtime
    |       +--> M10 Knowledge
    |
    +--> FDE API --> FDE Mastery

M7 remains the authorization/security authority. M6 remains the MCP/tool boundary. M8 remains evaluation authority. M9 is the controlled execution plane. The arrows are responsibility relationships, not a claim that every capability is a hard runtime dependency.

Security and trust boundaries

Tinlance treats browser input, webhooks, email, AI/model output, MCP messages, tool output, and external API responses as untrusted.

Key rules:

  • Authorization is server-side.
  • Customer resources have explicit organization/tenant boundaries.
  • High-impact operations use idempotency and/or approval controls.
  • Secrets remain server-side.
  • FDE execution is separated behind an authenticated service boundary.
  • Production FDE execution uses upstream OAuth 2.0 client credentials.
  • Auditability and request correlation are preserved across boundaries.
  • Unrestricted AI-to-database access is not permitted.
  • Unrestricted shell, arbitrary filesystem access, arbitrary HTTP, and autonomous red-team execution are not enabled by the M9 runtime.

See SECURITY.md and the security documentation.

Documentation

Start with the documentation index.

Topic Reference
Architecture Canonical architecture
Decisions ADRs
Evidence Evidence and trust
FDE FDE integration
API API v1
OpenAPI v1 contract
Customer workspace M3
Automation M4
M7 AI Security Gateway
M8 Agent Evaluation
M9 Agent Runtime
M10 Knowledge/RAG
M13 Knowledge Moat
M14 Productization
CI Enterprise CI gates
Production Gate E
Scale Gate F

Contributing

See CONTRIBUTING.md. Pull requests should be focused, tested, security-aware, and accompanied by documentation when behavior or architecture changes.

License and acknowledgements

Tinlance is proprietary software owned by Tinlance Limited. See LICENSE.

Third-party libraries remain governed by their own licenses. Dependency metadata and lockfiles are committed to make the software supply chain inspectable.

Acknowledgements include the open-source projects that make the platform possible, including Next.js, React, TypeScript, Prisma, Better Auth, FastAPI, Pydantic, Vitest, pnpm, Turborepo, and the security tooling used in CI.

Roadmap and maintenance posture

Roadmap

The repository maintains a phase ledger in docs/PHASES.md. Current platform work is organized around bounded commercial, customer-workspace, API, automation, MCP, security, evaluation, agent-runtime, knowledge, intelligence, and productization capabilities.

Roadmap entries are not evidence of implementation. Current code, tests, CI, and production verification remain authoritative.

Support and troubleshooting

For installation problems, include the exact command, runtime versions, commit SHA, and relevant non-sensitive logs.

For security problems, use SECURITY.md rather than a public issue.

For contribution and review requirements, see CONTRIBUTING.md.

About

Production-grade AI engineering and FDE platform for enterprise automation, AI infrastructure, security, and intelligent systems.

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages