ccppccpp home

Profiles

Profile catalog

Guardrails block what an agent must never do. Quality and stack profiles hold generated code to your conventions. Set a profile by its id.

ccpp cpo profile set <id>

Security

Secrets, credentials and code from the internet stay out of the agent's reach.

  • Secrets baseline

    5 policies

    Secrets stay out of the model's context and out of files, whatever the tool. Code from the internet is never piped into a shell.

    secrets-baseline

  • Base security

    39 policies

    A Bash and Python workstation: you confirm what destroys, leaves the project, reaches the network or needs privileges.

    base-security

Governance

History rewrites and infrastructure changes need a human.

  • Git safety

    3 policies

    No force-push to protected branches and no skipped hooks.

    git-safety

  • Infrastructure changes

    2 policies

    You confirm every Terraform, Pulumi, Kubernetes, Helm or CloudFormation change. State files are never edited by hand.

    infra-changes

  • Enterprise baseline

    bundle

    Organisation-wide guardrails for every repository, built from the security and governance profiles.

    enterprise-baseline

Cost control

GPU jobs and cloud machines start once you confirm.

  • Compute spend

    2 policies

    You confirm training runs, cluster jobs and new cloud machines before they start billing.

    compute-spend

Code quality

Type safety and file size, so generated code stays maintainable.

  • TypeScript strict

    6 policies

    No new any, loose types, unsafe assertions or suppressions.

    typescript-strict

  • Python strict

    2 policies

    Type checker and linter findings are fixed, not silenced, and modules stay small.

    python-strict

  • Dart strict

    3 policies

    Analyzer findings are fixed and analysis_options.yaml is never weakened.

    dart-strict

Frontend

Web interfaces: markup, styles, copy and build.

  • Next.js 16

    123 policies

    Next.js 16.4 apps without pre-16 APIs, with explicit server, client and cache boundaries.

    nextjs-16

  • Tailwind CSS v4

    39 policies

    Utility-first Tailwind v4. Blocks the v3 patterns agents still write.

    tailwind-v4

  • Anti-AI writing

    13 policies

    Interface copy without stock phrases, filler transitions or templated rhetoric.

    anti-ai-writing

Backend

Services, APIs and the databases behind them.

  • Backend TypeScript

    48 policies

    TypeScript services with trustworthy contracts, bounded work and inspectable failures.

    backend-typescript

  • Fastify 5

    110 policies

    Fastify 5 services without Fastify 3/4 APIs, with explicit encapsulation and trusted schemas.

    fastify-5

  • TypeScript backend

    3 policies

    Node.js, Bun or Deno services in strict TypeScript with a SQL database and migrations.

    typescript-backend

  • Backend Python

    63 policies

    Python services with checked types, explicit failures and owned async work.

    backend-python

  • Backend Go

    63 policies

    Go services with checked errors, owned goroutines and bounded work.

    backend-go

  • Backend Rust

    63 policies

    Rust services (Tokio, axum, SQLx) with explicit failures and bounded resources.

    backend-rust

  • Backend Java

    65 policies

    Java services (Spring, JPA, JDBC) with declared boundaries and consistent transactions.

    backend-java

  • Backend Kotlin

    78 policies

    Kotlin services (Ktor, Spring) with owned coroutine lifetimes and preserved cancellation.

    backend-kotlin

  • Backend C#

    80 policies

    ASP.NET Core and EF Core services with explicit DI lifetimes and safe queries.

    backend-csharp

  • Backend PHP

    78 policies

    Symfony, Laravel and Doctrine services with safe input, SQL and shell calls.

    backend-php

Mobile

Native and cross-platform mobile applications.

  • Flutter app

    5 policies

    Flutter and Dart apps for iOS and Android, with code generation and store signing.

    flutter-app

Data & ML

Data pipelines, model training and notebooks.

  • Python ML

    6 policies

    PyTorch, Hugging Face and scikit-learn projects with raw data, notebooks and checkpoints.

    python-ml