Skip to main content
Glama

Headline Vibes Analysis MCP Server

by fred-em
progress.md2.73 kB
# Progress — Headline Vibes Last updated: 2025-10-21 ## Current Status - MCP server upgraded to MCP SDK 1.20 with stdio + HTTP transports - EventRegistry (newsapi.ai) client integrated with source URI resolution - Tools available: - analyze_headlines (daily, natural language or YYYY-MM-DD) - analyze_monthly_headlines (range of YYYY-MM months) - Structured outputs returned via MCP `structuredContent` with Zod schemas - Dual sentiment scores and investor relevance filtering in production - Token budgeting diagnostics and sampling telemetry included in responses - Memory Bank refreshed with updated docs ## What Works - Natural language date parsing (chrono-node) with strict YYYY-MM-DD fallback - EventRegistry fetching with pagination + source URI resolution cache - Investor relevance filtering before scoring - Political categorization (left/center/right) via static source mapping - Synopsis generation for general market sentiment and investor climate - Structured text + JSON responses validated via Zod - Graceful error handling (McpError) and shutdown (SIGINT) ## What’s Left To Build / Improve - Expand automated test coverage (event resolver mocks, analysis orchestration) - Evaluate caching/memoization for repeated inputs - Consider persisting token usage to shared store (Redis/Postgres) when multi-instance - Explore advanced diagnostics (anomaly detection, trend deltas) - Optional: integrate structured logging sink for production observability ## Known Issues / Risks - EventRegistry pricing/token consumption: large ranges still consume tokens quickly - Source URI resolution relies on suggestSourcesFast; handle API hiccups gracefully - Relevance lexicon may still over/under-filter edge-case topics - Vitest coverage limited; more mocks/fixtures needed ## Decision Log (Evolution) - Migrated from NewsAPI.org to EventRegistry (newsapi.ai) with source resolver caching - Adopted MCP SDK 1.20 for structured outputs and streamable HTTP transport - Introduced token budgeting diagnostics to respect EventRegistry allowances - Standardized synopsis helpers and relevance utilities for reuse - Added HTTP deployment path (Railway) with health checks and allowlists ## Next Actions 1) Backfill targeted unit tests (analysis orchestration, token budgeting edge cases, source resolver) 2) Evaluate caching/memo strategies for repeated requests 3) Plan deployment hardening: structured logging sink, rate-limit telemetry dashboards 4) Explore anomaly detection / trend comparisons for future product milestones ## References - Env: NEWS_API_KEY, TRANSPORT, HOST, PORT, LOG_LEVEL, NEWS_API_BASE_URL - Build: npm run build (outputs to build/index.mjs) - Run: npm run start (stdio or HTTP via env)

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/fred-em/headline-vibes'

If you have feedback or need assistance with the MCP directory API, please join our Discord server