ReactAgentBuilder should support behavior based on AGENTS.md standard format

[ans-4175]

Summary

Enhance ReactAgentBuilder to detect and adapt its configuration or behavior by reading an AGENTS.md (or similar) documentation file, following a standard format for agent definitions and capabilities.

Motivation

Currently, agent configuration is manual via code and environment variables. If ReactAgentBuilder could automatically recognize configuration or prompts from a standard markdown specification (e.g., AGENTS.md), it would:

• Enable declarative, human-readable documentation-driven agent setup
• Reduce configuration duplication between docs and code
• Allow easier onboarding and clarity for users

Proposed Approach

• Define or adopt a standard format for describing agent roles, capabilities, prompt templates, and configuration in AGENTS.md.
• Add logic to ReactAgentBuilder to optionally parse and apply configuration and prompt settings from this file at build/init time.
• Ensure fallback to existing config when the file is absent or incomplete.
• Add documentation and examples for the AGENTS.md format and usage.

References

• See README.md and /docs/contents/ReactAgentBuilder-Guide.md for current config patterns.
• Example configuration snippet:

const agent = new ReactAgentBuilder({
  geminiKey: process.env.GEMINI_KEY,
  openaiKey: process.env.OPENAI_KEY,
  openrouterKey: process.env.OPENROUTER_KEY,
  braveApiKey: process.env.BRAVE_API_KEY,
  useEnhancedPrompt: true,
  memory: "in-memory",
  enableToolSummary: true
})
.init({
  selectedProvider: "gemini",
  model: "gemini-2.5-flash"
})
.build();

Acceptance Criteria

• ReactAgentBuilder can read agent and prompt configuration from a markdown file in a standard format.
• Existing manual config via code is still supported and takes precedence if both exist.
• AGENTS.md usage is documented in the repo.
• Example AGENTS.md is provided.

Additional Notes

• This would facilitate config-as-docs and better onboarding for agent builders.
• Consider extensibility for future agent types, prompt schemas, and multi-agent setups.