npx agentspec-cli install e681b1d0-7e83-468a-8569-e8c0d74760db --file-type agents-md→ .agents/
# Agent Instructions
## General
- **Confirm First:** Never implement without user confirmation.
- **Java First:** Prefer Java for scripting/agent tasks over Python/Bash, unless standard Unix tooling fits perfectly.
- **Workflow:** Check `specifications/openspec/changes/` for active changes and delta specs before implementing. Use `/opsx-*` workflows.
- **Secrets & Security:** You must never directly store any secrets, such as API keys, in source code, configuration files (e.g. `neodymium.properties`, `ai.properties`), or templates. All API keys and credentials must be injected dynamically via environment variables (such as `GEMINI_API_KEY`) or passed at run-time as JVM arguments (e.g., `-Dneodymium.ai.apiKey=...`). Before you commit anything, verify your staged changes do not contain keys by running `git diff --cached | grep -E "(apiKey|API_KEY|AQ\.[a-zA-Z0-9_\-]{10,})"`. If a secret is accidentally committed, notify the user immediately.
## Coding Standards
- **TDD:** Write unit/integration tests before implementing new functionality. Ensure full coverage.
- **Style:** Allman code style (new line braces), document non-obvious logic and all public API, JDK 21 features.
- **Headers:** Add GNU AGPLv3 license header to all AI-related source files, but MIT license header for the rest.
- **Attribution:** Do not use inline comments like `// AI-generated: <Model>`. Use `@author` tags on class/interface Javadoc level instead (e.g., `@author AI-generated: <Model Name>`), and always add a second author tag for `Xceptance GmbH 2026`.
- **Strict Java:** Aggressive `final` modifiers (variables, args, methods, fields). NO inline FQCNs; use explicit top imports. Unused imports and variables are strictly prohibited.
- **Documentation:** Write proper method comments, inline comments, and class comments where needed. Make it as useful for a human as possible. Also keep in mind, that this is also documentation for AI to reconstruct decisions.
- **Imports:** Always import fully qualified. Use static imports such as `Assertions.assertEquals`, only when it increases readability and the import is used multiple times. Remove unused imports.
## Dependencies & Git
- **Dependencies:** ALWAYS ask permission before adding. Document in `NOTICE.md` and `doc/3rd-party-licenses/`.
- **Git:** No fast-forward merges. Ask before stashing. Branch naming: `(feat|fix|chore|docs)/kebab-case`.
## Testing & Specifications
- **Maintenance:** Tests MUST be created/updated for any logic or UI changes (reinforces TDD).
## Aura AI Test Sandbox
- **Aura Glance Sandbox:** To test any Neodymium Aura AI features (such as visual audits, dHash baselines, parameterizations, multi-port, offline replays, or dynamic visual defects), utilize our self-contained **Aura Test Suite Hub** under `src/test/resources/ai-test-pages/AuraGlanceTest/`.
- **Server Ports:** Served dynamically on random free ports (HTTP + HTTPS self-signed cert `keystore.p12`) via `EmbeddedHtmlServer.java` (no external running app required!).
- **How to Use:** Write tests in `AuraGlanceTest.java` and `AuraFeatureMatrixTest.java` extending `BaseAiTest`. Inject parameterized data from `AuraFeatureMatrixTest.json`.
- **Extensive Documentation:** For detailed guidelines, directory layouts, and step replays instructions, refer to [doc/aura-visual-defect-sandbox.md](doc/aura-visual-defect-sandbox.md).AGENTS.md
AI coding agent instructions for project conventions and guidelines.
by AgentHub Crawler
Agent Instructions & Guidelines
You are an expert autonomous AI developer tasked with building a local Code Intelligence Engine and Model Context Protocol (MCP) server. Your goal is to parse a massive, legacy Python codebase into an Abstract Syntax Tree (AST), convert it into an in-memory Knowledge Graph, and expose tools for AI agents to query.
by AgentHub Crawler
Agent Rules
AI coding agent instructions for project conventions and guidelines.
by AgentHub Crawler
Agent Guidelines for .agents
Sign in to leave a comment.
This repository contains agent definitions, command workflows, skill definitions, and coding rules for AI-powered coding tools. These files configure how specialized AI agents behave when assisting with development tasks.
by AgentHub Crawler