Works with the AI Agents you already use
bot triage
Diagnose web server load spikes, identify abusive bots, and generate safe firewall or CDN block rules.
Free
bot triage
Example session with this skill installed
My server load is at 40 and the site is timing out. Can you find out what is hitting us?
- Read your context and instructions
- Compiled the bot triage
I have analyzed the current server load and access logs. The load is 42.15, primarily driven by php-fpm processes. I found a distributed crawl from a specific user agent targeting expensive search facets.
Triage Report
| Category | Detail | Impact |
|---|---|---|
| Top IP | 192.0.2.45 | 4,200 requests in 5m |
| Subnet | 198.51.100.0/24 | 12,000 requests in 5m |
| User Agent | DataScraper/1.0 | 85% of total traffic |
| Target Path | /search?filter=* | 92% of CPU time |
| Signature | FACET_CRAWL | High |
The IP 192.0.2.45 is not a verified search engine. The subnet 198.51.100.0/24 belongs to a generic cloud hosting provider.
Proposed Blocks
I recommend blocking the abusive subnet at the firewall and the specific User Agent at the Nginx level to reduce PHP overhead.
# Backup nginx config
cp /etc/nginx/sites-available/default /etc/nginx/sites-available/default.bak.$(date +%F)
# Firewall block for the primary offending subnet
csf -d 198.51.100.0/24 "Aggressive scraper: DataScraper"
# Nginx block for the User Agent
# Add to server block:
# if ($http_user_agent ~* (DataScraper)) { return 403; }
Next steps
- Approve the
csfsubnet block to drop traffic at the network layer. - Apply the Nginx User Agent rule to catch any rotated IPs from the same bot.
- Run
triage.sh -m 5in three minutes to verify the load has dropped below the core count.
Connects securely to your tools. The creator never sees your data.
About this skill
High web server load often stems from aggressive scrapers, vulnerability scanners, or misconfigured crawlers. This skill identifies the specific IPs and patterns causing the spike and provides the exact commands to block them safely.
What it does
- Environment discovery identifies the OS, control panel, web server, and active firewall automatically.
- Traffic triage analyzes access logs to surface top IPs, subnets, user agents, and suspicious URL patterns.
- Search engine verification performs forward-confirmed reverse DNS lookups to prevent blocking legitimate bots like Googlebot or Bingbot.
- Rule generation creates targeted block commands for Cloudflare WAF, firewalls, or web server configurations.
- Safety checks verifies host load and process states to ensure the issue is actually web traffic before recommending action.
How it works
- Analyze load using
triage.shto confirm if CPU spikes are driven by web workers or unrelated I/O bottlenecks. - Drill down on suspect IPs with
ip-detail.shto see request timelines and verify bot identity. - Select layer for the block based on whether the traffic is behind a CDN or hitting the origin directly.
- Review rules generated by
gen-blocks.shfor the specific firewall or web server detected. - Verify impact by monitoring load and testing critical user paths to ensure no false positives.
Frameworks & tools
Works on Linux servers using Nginx, Apache, LiteSpeed, or Caddy. Supports firewalls like CSF, ufw, firewalld, and iptables, as well as Cloudflare WAF. Compatible with cPanel, Plesk, and DirectAdmin environments.
Why this beats prompting it yourself
Generic prompts often fail to account for CDN headers, leading to the accidental blocking of your own CDN's edge nodes. This skill uses a specialized playbook to verify bot identities and ensures blocks are applied at the cheapest layer possible without breaking legitimate site integrations.
Use cases
- Identify and block a distributed brute-force attack on a login endpoint.
- Stop an aggressive crawler that is ignoring robots.txt and maxing out PHP workers.
- Mitigate a spike in vulnerability probes looking for
.envor.gitfiles. - Detect when high load is actually caused by a storage failure rather than traffic.
Known limitations
Requires root or sudo access to read system logs and manage firewalls. Log analysis depends on NCSA combined or Apache vhost_combined formats.
Requires root access for logs and firewall. Only supports NCSA or vhost_combined log formats. Does not automatically apply changes without explicit user approval for each rule.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 16 days ago
- Free with an Agensi account