docs: add AGENTS.md and gitignore AI agent local files

Add AGENTS.md as a cross-tool standard for AI coding agents with
pointers to in-tree docs, build verification commands, safety-critical
directory table, coding rules, and an agent decision framework.

Update docs/en/contribute/code.md with coding conventions enforced
during review (return early, no get/is prefixes, default member init,
no std:: in embedded code, no dynamic alloc after init).

Update docs/en/simulation/index.md with a "Choosing a Simulator"
section explaining when to use SIH vs Gazebo.

Gitignore tool-specific AI agent configs and local state for Claude
Code, Cursor, Windsurf, Aider, Cline, Continue.dev, Codex CLI,
Amazon Q, Augment, Cody, Junie, Tabnine, and Gemini.

Signed-off-by: Ramon Roche <mrpollo@gmail.com>

docs/en/contribute/code.md

@@ -86,6 +86,20 @@ private:
 };
 ```
 
+## Coding Conventions
+
+The following conventions are enforced during code review:
+
+- **Return early** to reduce nesting depth.
+- **No `get`/`is` prefixes** on boolean methods: prefer `enabled()` over `isEnabled()`.
+- **Default member initialization** in class headers, not constructors (e.g. `float _value{0.f};`).
+- **Avoid `goto`.**
+- **Prefer `++i`** over `i++`.
+- **No `std::` containers or algorithms** in embedded (NuttX) code.
+- **No dynamic memory allocation after init.**
+
+Formatting source of truth: `Tools/astyle/astylerc`. Run `make format` to fix formatting project-wide, or `Tools/astyle/fix_code_style.sh <file>` for a single file.
+
 ## In-Source Documentation
 
 PX4 developers are encouraged to create appropriate in-source documentation.

docs/en/simulation/index.md

@@ -24,6 +24,13 @@ The following simulators are supported by the PX4 core development team.
 
 There are also a number of [Community Supported Simulators](../simulation/community_supported_simulators.md).
 
+### Choosing a Simulator
+
+- **SIH (Simulation In Hardware):** Quick verification, no GPU needed, headless CI, basic flight logic testing. Runs as a PX4 module (no external process). Ideal for controller and estimator validation.
+- **Gazebo:** Sensor simulation (camera, lidar, depth), multi-vehicle, visual testing, environment interaction. Requires more resources.
+
+Use SIH when you need fast, lightweight testing of control logic or CI pipelines. Use Gazebo when you need realistic sensor data or visual environment interaction.
+
 ---
 
 The remainder of this topic is a "somewhat generic" description of how the simulation infrastructure works.