agentleFS
terrene-foundationterrene-foundation/pact/.claude/skills/13-architecture-decisions/SKILL.md

architecture-decisions

Skill0 starsChanged 7 months agoApache-2.0View on GitHub

Architecture decision guides for Kailash SDK including framework selection (Core SDK vs DataFlow vs Nexus vs Kaizen), runtime selection (Async vs Sync), database selection (PostgreSQL vs SQLite), node selection, and test tier selection. Use when asking about 'which framework', 'choose framework', 'which runtime', 'which database', 'which node', 'architecture decision', 'when to use', 'Core SDK vs DataFlow', 'PostgreSQL vs SQLite', 'AsyncLocalRuntime vs LocalRuntime', or 'test tier selection'.

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    NEVER use ORM when DataFlow can generate nodes

    line 72
  2. Very specific:

    NEVER build API/CLI/MCP manually when Nexus can do it

    line 73
  3. Distinctive:

    NEVER use LocalRuntime in Docker (causes hangs)

    line 78
  4. Distinctive:

    NEVER mix runtimes in same application

    line 79
  5. Distinctive:

    NEVER skip connection pooling config

    line 84

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Kailash Architecture Decisions

Decision guides for selecting the right frameworks, runtimes, databases, nodes, and testing strategies.

Reference Documentation

DecisionFileQuick Answer
Frameworkdecide-frameworkCore SDK (custom), DataFlow (DB), Nexus (multi-channel), Kaizen (AI agents)
Runtimedecide-runtimeDocker/async -> AsyncLocalRuntime; CLI/Scripts -> LocalRuntime
Databasedecide-database-postgresql-sqliteProduction -> PostgreSQL; Dev/Test -> SQLite
Nodedecide-node-for-taskSee node selection flow below
Test Tierdecide-test-tierUnit (fast), Integration (real infra), E2E (full system)

Framework Selection Matrix

NeedFrameworkWhy
Custom workflowsCore SDKFull control, 140+ nodes
Database CRUDDataFlowAuto-generated nodes
Multi-channel APINexusAPI + CLI + MCP instantly
AI agentsKaizenSignature-based agents
All of aboveCombine themThey work together

Runtime Selection Flow

Deploying to Docker/async/Kubernetes?
  YES -> AsyncLocalRuntime (async-first, no threads)
  NO  -> CLI/script?
    YES -> LocalRuntime (sync execution)
    NO  -> Use get_runtime() for auto-detection

Database Selection Flow

Production deployment?     -> PostgreSQL (scalable, enterprise)
Development/testing?       -> SQLite (simple, fast setup)
High concurrency?          -> PostgreSQL (better concurrency)

Node Selection Flow

Custom Python logic        -> PythonCodeNode
LLM/AI tasks               -> LLMNode, OpenAINode, AnthropicNode
Database operations        -> DataFlow auto-generated nodes
HTTP API calls             -> HTTPRequestNode
File reading               -> FileReaderNode
Conditional routing        -> SwitchNode
Not sure?                  -> Check nodes-quick-index

Test Tier Flow

Individual function        -> Tier 1 (Unit)
Workflow execution         -> Tier 2 (Integration, real infrastructure)
Complete user flow         -> Tier 3 (E2E)

Critical Decision Rules

Framework

  • NEVER use ORM when DataFlow can generate nodes
  • NEVER build API/CLI/MCP manually when Nexus can do it

Runtime

  • Docker/async -> AsyncLocalRuntime (mandatory)
  • NEVER use LocalRuntime in Docker (causes hangs)
  • NEVER mix runtimes in same application

Database

  • Production -> PostgreSQL; NEVER SQLite for production high-concurrency
  • NEVER skip connection pooling config
  • Multi-instance -> One DataFlow per database

Related Skills

  • 01-core-sdk - Core SDK fundamentals
  • 02-dataflow - DataFlow framework
  • 03-nexus - Nexus framework
  • 04-kaizen - Kaizen framework
  • 08-nodes-reference - Node reference
  • 12-testing-strategies - Testing strategies

Support

  • decide-framework skill - Framework selection and architecture
  • analyst - Deep analysis for complex decisions
  • pattern-expert - Pattern recommendations

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.