agentleFS
apacheapache/streampipes/AGENTS.md

streampipes

AGENTS.md756 starsChanged 52 days agoApache-2.0View on GitHub

Applies to the whole repository. A deeper AGENTS.md adds to this file; it never replaces it. Before editing inside a directory that has its own guide, read that guide. Not every agent harness loads nested guides automatically, so the index at the end of this file is the map. - Java: JDK 25 (Temurin) everywhere: CI, the dev container, the Docker images and the compiler (<release>25</release> in pom.xml). The enforcer plugin rejects older JDKs, so the language level is Java…

  • Installs packages

We split this AGENTS.md into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    Not every agent harness loads nested guides automatically, so the index at the end of this file is the map.

    line 7
  2. Very specific:

    Before opening a pull request, run what CI runs (.github/workflows/pr-validation.yml):

    line 30
  3. Distinctive:

    Prefer constructor-injected, typed APIs over loosely typed maps when extending domain logic.

    line 64
  4. Distinctive:

    Keep backward compatibility for public APIs, serialized models and storage contracts unless the task explicitly changes them.

    line 65
  5. Very specific:

    Commit messages: follow the history: feat(#1234): ..., fix(#1234): ..., test(#1234): ..., chore(deps): ..., ci: ..., docs: ....

    line 67
  6. Distinctive:

    Keep changes scoped to the requested modules.

    line 73
  7. Very specific:

    Do not edit generated code by hand (ui/projects/streampipes/platform-services/src/lib/model/gen/, anything under target/).

    line 74

≈ “Common” is an estimate: at least 1 in 5 of the 36,368 AGENTS.md files we read have a rule of the same kind, not the same words. “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
AGENTS.md

AGENTS Guide (Repository Root)

Scope

Applies to the whole repository. A deeper AGENTS.md adds to this file; it never replaces it. Before editing inside a directory that has its own guide, read that guide. Not every agent harness loads nested guides automatically, so the index at the end of this file is the map.

Toolchain (read this before running anything)

  • Java: JDK 25 (Temurin) everywhere: CI, the dev container, the Docker images and the compiler (<release>25</release> in pom.xml). The enforcer plugin rejects older JDKs, so the language level is Java 25. .java-version is gitignored; if you use jenv or similar, point it at a local JDK 25.
  • Maven 3.8+. Node 22 with npm for ui/. Docker + Compose for the service stack.
  • Python client: Poetry — see streampipes-client-python/AGENTS.md.
  • Go client: Go 1.25 in CI, module targets go 1.21 — see streampipes-client-go/AGENTS.md.
  • Main branch is dev. Pull requests target dev.

Build and validate

Fast loop for a backend change:

mvn -pl <module> -am test        # -am also builds upstream modules; add -o when offline
mvn -pl <module> -am -DskipTests package   # compile + checkstyle only

Before opening a pull request, run what CI runs (.github/workflows/pr-validation.yml):

mvn clean verify                 # from the repository root
cd ui && npm run format && npm run lint && npm run i18n:check

Why the root verify matters: checkstyle runs in the validate phase and is therefore covered by -pl <module> test, but the Apache RAT license-header check runs only in the verify phase and only from the root pom (inherited=false). A module-level test will pass on a file that is missing its license header; CI will not.

Full backend build: mvn clean package. Full UI build: cd ui && npm install && npm run build.

Run the stack locally

  • .devcontainer/ starts CouchDB, the message broker, the time-series store and the other third-party services with Docker Compose (.devcontainer/README.md). It is the reference for required services and environment variables even outside VS Code.
  • .vscode/launch.json has working run configurations: Debug Core, Debug UI (UI on port 8082 with proxy), Debug IIoT Extensions, Open Cypress. .vscode/tasks.json has the matching dev: ... build tasks.
  • docker-compose.yml at the root starts a complete stack from images (no volumes).

Code standards

  • License header on every source file (Java, TypeScript, SCSS, HTML, Markdown, YAML, shell). Markdown uses an HTML comment. Exempt by RAT configuration: **/AGENTS.md, **/CLAUDE.md, .claude/**, *.json, *.txt (full list in pom.xml, apache-rat-plugin).
  • Java style and import order: enforced by tools/maven/checkstyle.xml with the header template tools/maven/checkstyle-header.txt. Eclipse formatter settings that match checkstyle: tools/maven/streampipes-code-formatter.xml. Import order is org.apache.streampipes → third-party → jakarta → javax → java → scala → static. (Spotless is configured in pom.xml but skipped; checkstyle is the gate.)
  • Prefer constructor-injected, typed APIs over loosely typed maps when extending domain logic.
  • Keep backward compatibility for public APIs, serialized models and storage contracts unless the task explicitly changes them.
  • Commit messages follow the history: feat(#1234): ..., fix(#1234): ..., test(#1234): ..., chore(deps): ..., ci: ..., docs: .... Reference the issue when one exists. The PR template is .github/PULL_REQUEST_TEMPLATE.md.

Monorepo rules

  • Keep changes scoped to the requested modules; no opportunistic cross-module refactors.
  • Do not edit generated code by hand (ui/projects/streampipes/platform-services/src/lib/model/gen/, anything under target/). Regenerate it — see the owning module's guide.

Architecture boundaries

  • streampipes-service-core: bootstrapping, security configuration, migrations, scheduling. Orchestration only — no domain logic.
  • streampipes-rest: HTTP resource layer. Thin: parse, authorize, delegate, map the response.
  • streampipes-*-management (pipeline, resource, user, connect, storage, extensions, data-explorer): business and domain logic. New behaviour goes here.
  • streampipes-storage-api / streampipes-storage-couchdb: persistence contracts and the CouchDB implementation.
  • streampipes-model, streampipes-model-client, streampipes-model-shared: the domain model. It is persisted to CouchDB and generated into TypeScript for the UI, so field changes are compatibility-sensitive on both sides.
  • streampipes-extensions/: adapters, processors and sinks, built on streampipes-sdk, streampipes-extensions-api and the streampipes-wrapper-* runtimes.
  • ui/: Angular application plus the @streampipes/platform-services and @streampipes/shared-ui libraries.

Security

THREAT_MODEL.md defines the trust boundaries (REST front door, adapter ingestion, extension runtime) and what is the operator's responsibility rather than the engine's. Read it when a change touches authentication, permissions, adapters or extension loading. SECURITY.md says how vulnerabilities are reported. The full procedure for triaging a suspected vulnerability lives in .claude/skills/security-triage/SKILL.md; it is deliberately not inlined here so that it does not load on unrelated tasks.

Guide index

DirectoryGuide
streampipes-service-core/AGENTS.md — startup, security config, adding a migration
streampipes-rest/AGENTS.md — adding an endpoint, authorization annotations
streampipes-pipeline-management/AGENTS.md — matching, verification, graph, execution
streampipes-resource-management/AGENTS.md — resource lifecycle, permissions coupling
streampipes-user-management/AGENTS.md — tokens, passwords, roles
streampipes-storage-couchdb/AGENTS.md — DAOs, views, serializers
streampipes-extensions/AGENTS.md — adding an adapter, processor or sink
streampipes-client-python/AGENTS.md — Poetry toolchain and gates
streampipes-client-go/AGENTS.md — Go toolchain and gates
ui/AGENTS.md — design sources of truth, build, page composition
ui/cypress/AGENTS.md — E2E authoring rules
ui/projects/streampipes/platform-services/AGENTS.md — API client library, generated models
ui/projects/streampipes/shared-ui/AGENTS.md — shared component library

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.