freshprobe
Data freshness verification for AI agents. Probes endpoints for HTTP cache staleness, latency percentiles, content fingerprinting, TLS certificate health, DNS timing, and redirect chains. Returns deterministic FRESH/STALE/UNKNOWN JSON verdicts with policy evaluation and NIST AI RMF mapping.
README
<p align="center"> <h1 align="center">freshprobe</h1> <p align="center"> <strong>Data freshness verification for AI agents.</strong> <br /> Stop your agents from acting on stale data. </p> <p align="center"> <a href="https://github.com/Sudhan30/freshprobe/actions/workflows/ci.yml"><img src="https://github.com/Sudhan30/freshprobe/actions/workflows/ci.yml/badge.svg" alt="CI" /></a> <a href="https://goreportcard.com/report/github.com/Sudhan30/freshprobe"><img src="https://goreportcard.com/badge/github.com/Sudhan30/freshprobe" alt="Go Report Card" /></a> <a href="https://github.com/Sudhan30/freshprobe/blob/main/LICENSE"><img src="https://img.shields.io/github/license/Sudhan30/freshprobe" alt="License" /></a> <a href="https://github.com/Sudhan30/freshprobe/releases"><img src="https://img.shields.io/github/v/release/Sudhan30/freshprobe" alt="Release" /></a> <img src="https://img.shields.io/badge/go-1.25-blue" alt="Go Version" /> </p> </p>
"I asked my agent to check flight prices. It gave me options. I booked one. The fare had changed 3 hours ago."
AI agents routinely act on stale data without knowing it. A financial agent queries cached quotes from 47 minutes ago. A support bot tells a customer their order doesn't exist because the CRM hasn't synced. An RAG pipeline confidently answers with yesterday's docs.
freshprobe sits between your agent and the external world. Before the agent acts, it asks: is this data fresh enough? The answer is always a deterministic JSON verdict: FRESH, STALE, or UNKNOWN.
$ freshprobe check https://api.example.com/v2/quotes
{
"verdict": "STALE",
"confidence": 0.94,
"endpoint": "https://api.example.com/v2/quotes",
"freshness": {
"data_age_seconds": 2847,
"freshness_score": 0.12,
"cache_control": "max-age=3600"
},
"liveness": {
"status": "DEGRADED",
"latency_p50_ms": 342,
"latency_p95_ms": 1847,
"body_size_bytes": 4096,
"error_rate": 0.03
},
"redirects": {
"total_hops": 1,
"final_url": "https://api-v2.example.com/quotes",
"has_redirect": true
},
"nist_mapping": {
"ai_rmf_function": "MEASURE",
"control": "MS-2.6-001"
}
}
Single Go binary. No dependencies. Runs as CLI, MCP server, or HTTP microservice.
Why this matters
| Problem | Cost |
|---|---|
| E-commerce agent used 6-month-old product data | $5M+ revenue loss |
| Enterprise RAG with overlapping refresh infrastructure | $340K/year wasted |
| AI project failures from data quality issues | 60%+ of failures (Gartner) |
Unlike crashes that trigger alerts, stale data produces confident, well-formatted, completely wrong responses. Chain a few of those in a multi-agent pipeline and every component reports green while the output is catastrophically wrong.
Install
Go install (recommended):
go install github.com/Sudhan30/freshprobe/cmd/freshprobe@latest
Docker:
docker run --rm ghcr.io/sudhan30/freshprobe:latest check https://example.com
From source:
git clone https://github.com/Sudhan30/freshprobe.git && cd freshprobe && make build
./bin/freshprobe --version
GitHub Releases: Download pre-built binaries for Linux, macOS, and Windows from Releases.
Quick start
# Basic freshness check
freshprobe check https://api.example.com/data
# Human-readable output
freshprobe check https://api.example.com/data --output text
# Content fingerprinting: detect if data actually changes
freshprobe check https://api.example.com/data --repeat 3 --interval 2s
# Check against a freshness policy
freshprobe check https://api.example.com/data --policy-dir ./policies --policy financial-data
# Batch check multiple endpoints
freshprobe batch https://api1.example.com https://api2.example.com https://cdn.example.com
# Continuous monitoring (Ctrl+C to stop)
freshprobe watch https://api.example.com/data --interval 30s --output text
# Only alert on verdict changes (FRESH -> STALE)
freshprobe watch https://api.example.com/data --interval 1m --on-change --output text
# View probe history for an endpoint
freshprobe history https://api.example.com/data --limit 20 --output text
Six verification signals
| Signal | What it checks |
|---|---|
| HTTP cache headers | Parses Last-Modified, Cache-Control, Age, ETag, Date, Expires. Computes 0.0 to 1.0 freshness score |
| Endpoint liveness | Measures response latency (P50/P95/P99), status codes, body size, degradation patterns |
| Content fingerprinting | SHA-256 hashes response bodies across repeated probes to detect stale caches |
| TLS certificate health | Certificate validity, days remaining, OCSP stapling status |
| DNS resolution timing | DNS lookup latency as infrastructure health signal |
| Redirect chain analysis | Tracks 301/302/307/308 hops, detects stale CDN configs |
Three deployment modes
CLI
freshprobe check <url> [flags]
freshprobe batch <urls...> [flags]
freshprobe watch <url> --interval 30s [flags]
freshprobe history <url> --limit 20
MCP server (for AI agents)
Add to your AI tool config:
<details> <summary><strong>Claude Desktop / Claude Code</strong></summary>
{
"freshprobe": {
"type": "stdio",
"command": "freshprobe",
"args": ["serve", "--mode", "mcp", "--policy-dir", "/path/to/policies", "--stateless"]
}
}
</details>
<details> <summary><strong>Cursor</strong></summary>
In .cursor/mcp.json:
{
"mcpServers": {
"freshprobe": {
"command": "freshprobe",
"args": ["serve", "--mode", "mcp", "--stateless"]
}
}
}
</details>
<details> <summary><strong>VS Code (Copilot)</strong></summary>
In .vscode/mcp.json:
{
"servers": {
"freshprobe": {
"type": "stdio",
"command": "freshprobe",
"args": ["serve", "--mode", "mcp", "--stateless"]
}
}
}
</details>
This exposes three tools to AI agents:
| Tool | Description |
|---|---|
freshprobe_check |
Probe a single endpoint. Returns JSON verdict |
freshprobe_batch |
Probe multiple endpoints concurrently |
freshprobe_policy |
Check an endpoint against a named freshness policy |
HTTP server
freshprobe serve --mode http --addr :8080
POST /api/v1/check {"url": "https://..."}
POST /api/v1/batch {"urls": ["https://...", "https://..."]}
POST /api/v1/policy {"url": "https://...", "policy_name": "api-realtime"}
GET /healthz
GET /metrics # Prometheus-compatible metrics
Policies (freshness-as-code)
Define freshness thresholds per domain in YAML:
version: "1"
policies:
financial-data:
name: "Financial Data"
domains: ["*.market.*", "*.trading.*"]
max_staleness: "30s"
min_freshness_score: 0.9
max_latency_p95_ms: 200
require_tls: true
min_tls_days_left: 30
require_changing: true
api-standard:
name: "Standard API"
domains: ["api.*"]
max_staleness: "5m"
min_freshness_score: 0.6
max_latency_p95_ms: 2000
require_tls: true
When a probe violates a policy:
{
"policy_result": {
"policy_name": "Financial Data",
"passed": false,
"violations": [
{"check": "max_staleness", "expected": "<= 30s", "actual": "2m15s"},
{"check": "max_latency_p95", "expected": "<= 200 ms", "actual": "847 ms"}
]
}
}
Four built-in policies included: api-realtime, api-standard, static-content, financial-data.
Continuous monitoring
# Watch an endpoint, print every probe
freshprobe watch https://api.example.com/quotes --interval 30s --output text
# Only print when verdict changes (FRESH -> STALE transitions)
freshprobe watch https://api.example.com/quotes --interval 1m --on-change --output text
# Run 10 probes and exit
freshprobe watch https://api.example.com/quotes --count 10 --interval 5s
Example output:
Watching https://api.example.com/quotes every 30s
[14:22:01] FRESH conf=0.90 score=0.87 p95=142ms
[14:22:31] FRESH conf=0.90 score=0.85 p95=156ms
[14:23:01] STALE conf=0.85 score=0.22 p95=1847ms [FRESH -> STALE]
Prometheus metrics
The HTTP server exposes /metrics with Prometheus-compatible text format:
freshprobe_probes_total 142
freshprobe_verdict_total{verdict="FRESH"} 98
freshprobe_verdict_total{verdict="STALE"} 31
freshprobe_verdict_total{verdict="UNKNOWN"} 13
freshprobe_latency_p95_seconds 0.234000
freshprobe_freshness_score 0.7200
How it compares
| Feature | freshprobe | Uptime Kuma | Gatus | freshcontext-mcp |
|---|---|---|---|---|
| Purpose | Data freshness for AI agents | Uptime monitoring | Health dashboards | Web extraction timestamps |
| Knows data is stale | Yes (cache headers + fingerprinting) | No (only checks HTTP status) | No (only checks response assertions) | Partial (timestamps, no verification) |
| MCP server | Yes (3 tools) | No | No | Yes |
| Policy engine | Yes (YAML, per-domain) | No | Yes (YAML conditions) | No |
| Continuous monitoring | Yes (watch command) |
Yes (dashboard) | Yes (dashboard) | No |
| Prometheus metrics | Yes | No (push-based) | Yes | No |
| Deployment | Single binary | Docker + DB | Single binary | npm package |
Architecture
+------------------+
| freshprobe |
| single binary |
+--------+---------+
|
+--------------+--------------+
| | |
+----+----+ +----+----+ +-----+-----+
| CLI | | MCP | | HTTP |
| (cobra) | | (stdio) | | (net/http)|
+---------+ +---------+ +-----------+
| | |
+--------------+--------------+
|
+--------+---------+
| Probe Engine |
| |
| HTTP headers |
| Latency P50/95/99|
| Content SHA-256 |
| TLS/OCSP |
| DNS timing |
| Redirect chains |
+--------+---------+
|
+--------------+--------------+
| | |
+----+----+ +----+----+ +-----+-----+
| Verdict | | Policy | | Store |
| Engine | | Engine | | SQLite / |
| | | (YAML) | | Stateless |
+---------+ +---------+ +-----------+
Kubernetes deployment
apiVersion: apps/v1
kind: Deployment
metadata:
name: freshprobe
spec:
replicas: 1
selector:
matchLabels: { app: freshprobe }
template:
metadata:
labels: { app: freshprobe }
spec:
containers:
- name: freshprobe
image: ghcr.io/sudhan30/freshprobe:latest
args: ["serve", "--mode", "http", "--addr", ":8080",
"--policy-dir", "/etc/freshprobe/policies", "--stateless"]
ports:
- containerPort: 8080
resources:
requests: { cpu: 50m, memory: 64Mi }
limits: { cpu: 200m, memory: 128Mi }
readinessProbe:
httpGet: { path: /healthz, port: 8080 }
livenessProbe:
httpGet: { path: /healthz, port: 8080 }
Claude Code plugin
/plugin install github:Sudhan30/freshprobe
After installing, ask Claude:
- "Is the trading API returning fresh data?"
- "Check all our endpoints before running the batch job"
- "Does this API meet our real-time SLA?"
Development
make build # Build binary
make test # Run tests with race detector
make lint # go vet
make cross # Cross-compile (linux, macOS, Windows)
make docker # Docker build
Contributing
See CONTRIBUTING.md. High-value areas:
- Policy packs for specific domains (healthcare, weather, finance)
- WebSocket/gRPC/GraphQL probe signals
- OpenTelemetry integration
- Homebrew formula
License
MIT. See LICENSE.
<p align="center"> If freshprobe helps your agents make better decisions, <a href="https://github.com/Sudhan30/freshprobe">give it a star</a>. </p>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.