Rank live signals by freshness, provenance, and temporal relevance before they reach an agent, feed, or retrieval workflow.
Use FreshContext through MCP, APIs, feeds, and freshness-aware infrastructure tools.
Core Platform
FreshContext gives agents a temporal view of retrieved information: where it came from, when it was published, when it was retrieved, and how quickly it should lose ranking weight.
MCP is one interface layer. The core primitive is freshness-aware context: live adapters, temporal scoring, provenance, structured envelopes, and cache-aware delivery.
[FRESHCONTEXT]
Source: https://example.com/source
Published: 2026-05-12T09:00:00Z
Retrieved: 2026-05-12T10:00:00Z
Confidence: high
Score: 95/100 (current)
---
Fresh, source-grounded content...
[/FRESHCONTEXT]
[FRESHCONTEXT_JSON]
{ "freshcontext": { "freshness_score": 95 } }
[/FRESHCONTEXT_JSON]
Ecosystem
The current ecosystem spans a core MCP server, two vertical signal feeds, and an operations intelligence system for the Cloudflare infrastructure behind them.
Temporal scoring and MCP infrastructure for agent retrieval workflows.
Freshness-ranked Hacker News intelligence for tracking live technical signals.
Freshness-ranked jobs intelligence with AI/ML as the default category.
Plain-English Cloudflare operations diagnosis for Workers, D1, cron systems, logs, and runtime analysis.
How Freshness Works
FreshContext helps retrieval workflows preserve time as a first-class signal. It does not claim to solve hallucinations; it helps agents avoid stale context and rank live information more correctly.
R(t) = R0 · e-λt
Older information loses ranking weight unless reinforced by newer signals. Different sources decay at different speeds because a market quote, job listing, release note, repository, and paper do not age the same way.
Install / Developers
The hosted Worker endpoint is the fastest setup. The npm package is available for local stdio transport when you want the server on your machine.
Claude Desktop → Settings → Developer → Edit Config.
This uses the deployed FreshContext Worker. No local Node install required.
{
"mcpServers": {
"freshcontext": {
"command": "npx",
"args": [
"mcp-remote",
"https://freshcontext-mcp.gimmanuel73.workers.dev/mcp"
]
}
}
}
All 21 tools appear automatically after Claude reconnects to the MCP server.
npm install -g freshcontext-mcp
Edit ~/Library/Application Support/Claude/claude_desktop_config.json on macOS or your Claude config path on Linux.
{
"mcpServers": {
"freshcontext": {
"command": "freshcontext-mcp"
}
}
}
npm install -g freshcontext-mcp
where freshcontext-mcp
Use the path returned by where freshcontext-mcp in %APPDATA%\Claude\claude_desktop_config.json.
{
"mcpServers": {
"freshcontext": {
"command": "C:\\Users\\YOUR_NAME\\AppData\\Roaming\\npm\\freshcontext-mcp.cmd"
}
}
}
MCP Tool Examples
These examples route through MCP tools and return provenance, content dates, confidence, freshness_score, and structured JSON.
Use extract_hackernews to find recent discussion around browser agents.Use extract_company_landscape for Palantir with ticker PLTR.Use search_jobs to find current AI engineer roles in Europe.Use extract_idea_landscape for "AI operations monitoring".API / Live Examples
The feeds are intentionally narrow and inspectable: useful for dashboards, scheduled agents, Apify actors, and direct API calls.
curl "https://fresh-hn-feed.gimmanuel73.workers.dev/v1/hn/feed?limit=10"
curl "https://fresh-jobs-feed.gimmanuel73.workers.dev/v1/jobs/feed?category=ai&limit=10"
curl "https://freshcontext-mcp.gimmanuel73.workers.dev/health"
Marketplace
Freshness-ranked Hacker News feed actor for scheduled runs, datasets, and workflow automation.
Freshness-ranked jobs feed actor for AI/ML role monitoring and marketplace workflows.
Roadmap
Expose source-specific decay estimates for downstream retrieval systems.
Detect papers and topics becoming newly relevant again.
Push compact digests when important signals cross freshness thresholds.
Rank release notes by recency, velocity, and operational relevance.