portfolio-signal-tracker — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited portfolio-signal-tracker (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Website visitor intelligence agent. Monitors your site via GA4 for high-intent behavior patterns — not just traffic, but signals that suggest a serious visitor. Delivers an alert when something worth acting on is detected.
Before running, set in CONFIG.md:
GA4_PROPERTY_ID= # Your GA4 property ID (numeric, found in GA4 Admin)
GA4_ACCESS_TOKEN= # Google OAuth2 access token with GA4 read permissions
YOUR_COMPANY= # Used to contextualize alerts
ANTHROPIC_API_KEY=How to get your GA4 credentials:
# CLI
agent run portfolio-signal-tracker
# Voice (after ACTIVATION.md setup)
"[your wake word] portfolio"
# Scheduled — set your preferred interval
# Recommended: every 2-4 hours during working hours
# See ACTIVATION.md for scheduling instructionsClaude looks for combinations of behavior, not single data points:
| Signal | What it suggests |
|---|---|
| 3+ page views in one session | Genuine interest, not accidental |
| 5+ minutes on site | Reading carefully, not skimming |
| Visited work/case study pages | Evaluating your experience |
| Multiple visits within 48 hours | Returned to look again — strong intent |
| Visited contact page but did not submit | Considered reaching out |
| Direct URL visit (not from search) | Someone shared your link intentionally |
A single signal = low priority. Two or more signals in the same session = escalate.
You are a website visitor intelligence analyst. Review the following GA4 session
data and identify any sessions that show high-intent behavior — not casual browsing.
High-intent signals to look for:
- 3 or more pages visited in one session
- 5 or more minutes on site
- Visited work samples, case studies, or project pages
- Returned within 48 hours for a second visit
- Visited a contact or hire page
- Arrived via direct URL (not organic search or social)
For each high-intent session return:
- session_id or timestamp
- pages_visited (list)
- time_on_site (minutes)
- signal_strength (HIGH / MEDIUM / LOW)
- signals_detected (list of what triggered the alert)
- recommended_action: one specific thing to do in response
- alert_message: one sentence suitable for a push notification
If no high-intent sessions are found, return alert_level: GREEN with a brief summary.
Return valid JSON only.{
"alert_level": "HIGH",
"high_intent_sessions": [
{
"timestamp": "2026-06-16T09:14:00Z",
"pages_visited": ["/", "/work", "/case-studies/project-name", "/contact"],
"time_on_site_minutes": 8,
"signal_strength": "HIGH",
"signals_detected": [
"4 pages visited in one session",
"Visited case study and contact page",
"8 minutes on site"
],
"recommended_action": "Follow up within 24 hours — this visitor reviewed your work and visited the contact page.",
"alert_message": "High-intent visitor: 4 pages, 8 minutes, visited contact page."
}
],
"total_sessions_reviewed": 47,
"report_period": "Last 4 hours",
"next_recommended_run": "2 hours"
}Add UTM tracking — if visitors arrive from a specific campaign, UTM parameters will appear in GA4 and give you the source. Set up UTM on all outbound links to your site.
Add page-specific alerts — modify the Claude prompt to escalate when specific high-value pages are visited (e.g. a page you only share with specific prospects).
Connect to email or Slack — pipe the output to a notification script that sends you an email or Slack message when alert_level is HIGH.
# Alert via email when signal is high
agent run portfolio-signal-tracker | python notify.py --channel email --threshold HIGHlinkedin-post-generator — if a signal suggests a recruiter, prepare a posticp-radar — combine with CRM scoring for full pipeline intelligenceACTIVATION.md for voice and scheduled setup~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.