Updated model from gpt-oss-120b (no function calling support) to
anthropic/claude-haiku-4.5 which has excellent function calling support
for enabling Morgan's agent creation tool.
Changes:
- Update OPENROUTER_MODEL to anthropic/claude-haiku-4.5
- Add debug logging to show available tools for each agent
- Claude Haiku 4.5 provides fast, cost-effective function calling
This enables tool calling for Morgan's create_agent_package functionality.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Re-enabled the create_agent_package tool with proper parameter typing using
z.infer for runtime type safety. The tool now works with Vercel AI SDK's
tool calling mechanism.
Changes:
- Update create-agent-package.ts with z.infer type annotations
- Update qdrant-rag.ts with consistent z.infer typing pattern
- Re-enable tool import in definitions.ts
- Activate create_agent_package in Morgan's agent configuration
- Add explanatory comments about Vercel AI SDK typing workarounds
This enables Morgan to create custom AI agents through the native Vercel AI SDK
tool calling interface. The workaround is necessary due to TypeScript
overload resolution limitations in the tool() function, but the tool executes
correctly at runtime.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Replaced n8n-specific JSON output instructions with proper Vercel AI SDK
tool calling format. Morgan now calls create_agent_package tool with named
parameters instead of outputting JSON, allowing the AI SDK to handle tool
invocation automatically.
Changes:
- Replace JSON output instructions with parameter-based tool calling
- Clarify required and optional parameters for create_agent_package
- Update example flow to show tool calling instead of JSON output
- Remove references to messageType and toolCall JSON structure
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Morgan's system prompt is now generated at build time and embedded directly
in the code, making it available in Cloudflare Worker environments where
file system access isn't available.
Changes:
- Add scripts/generate-morgan-prompt.js to generate TypeScript constant from markdown
- Generate src/lib/agents/morgan-system-prompt.ts with full Fortura Agent Bundle
- Update agent definitions to import and use the embedded constant
- Update package.json build scripts to generate prompt before building
- Remove runtime file system access (readFileSync) that failed on Cloudflare
This ensures Morgan agent has full system prompt capabilities on all deployments.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Update `wrangler.jsonc` to correct the `CUSTOM_AGENT_WEBHOOK` URL.
- Enhance `page.tsx` to load custom agents from localStorage and merge them with predefined agents.
- Modify `route.ts` to validate `systemPrompt` for custom agents and include it in the webhook payload.
- Adjust `chat-interface.tsx` to handle custom agents more effectively, including system prompt integration and event dispatching for pinned agents.
- Remove obsolete `CUSTOM_AGENT_EXECUTION_PRD.md` and `DIFF_TOOL_USAGE.md` files as part of cleanup.
- Move new chat button to left side, bookmark button stays on right
- Add max-width constraint (75%) to user messages with proper text wrapping
- Remove right-align text from user message frames (keep bubbles on right)
- Add overflow handling for code blocks in messages
- Change scrollbar color from orange to gray in light and dark modes
- Fix pill loading animation flicker by initializing pinnedAgents from localStorage
- Add 0.2s base delay to pill animations for staggered reveal
- Improve Create new button animation: longer duration (0.6s), bouncy scale sequence, easeInOut easing
Key features:
- Multi-agent support with dynamic agent configuration via environment variables
- Floating shadcn Dialog for elegant agent selection
- Per-agent chat sessions and message history with localStorage persistence
- Image upload support with base64 encoding
- Dynamic webhook routing based on selected agent
- Refactored components: ChatInterface, Header, and new AgentSelector
- New API endpoints: GET /api/agents and updated POST /api/chat
- Centralized TypeScript types in src/lib/types.ts
- Comprehensive CLAUDE.md documentation
Agent Configuration:
- Set environment variables: AGENT_N_URL, AGENT_N_NAME, AGENT_N_DESCRIPTION
- Agent selector automatically discovers agents on first visit
- Seamless switching between agents without page reload
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
