Skip to main content
AI/MLcivitai

cloudflare

Query Cloudflare HTTP analytics AND manage security rules across zones. Traffic investigation (top IPs, paths, user agents, bot scores, timelines, scrape analysis) plus export and port of Custom Rules + Rate Limit Rules between zones (e.g. civitai.com -> civitai.red).

Stars
7,126
Source
civitai/civitai
Updated
2026-05-31
Slug
civitai--civitai--cloudflare
View on GitHubRaw SKILL.md

// install — copy + paste into any project

mkdir -p .claude/skills && curl -fsSL https://raw.githubusercontent.com/civitai/civitai/HEAD/.claude/skills/cloudflare/SKILL.md -o .claude/skills/cloudflare.md

Drops the SKILL.md into .claude/skills/cloudflare.md. Works with Claude Code, Cursor, and any agent that loads SKILL.md files from .claude/skills/.

Cloudflare Analytics + Rule Management

Query Cloudflare HTTP traffic analytics and manage security rules (Custom Rules, Rate Limit Rules) across zones. The CF_API_TOKEN in .env needs Zone Analytics: Read for analytics, and Zone WAF: Edit for rule management.

Running Commands

node .claude/skills/cloudflare/query.mjs <command> [options]

Commands

Command Description
top-clients Top IPs by request count
top-paths Top request paths
top-agents Top user agents
timeline Requests per minute timeline (bar chart)
ip <addr> Full breakdown for a specific IP (paths, UAs, statuses, geo)
scrape Full scrape analysis — top IPs, paths, UAs + detail on high-volume IPs
bot-scores Bot score distribution
bot-clients Top clients with low bot scores (likely bots)
firewall Recent firewall/WAF events
rate-limits Current rate limit rules
waf-rules Current WAF custom rules
list-zones List all accessible zones (id, name, plan)
export-rules <zone> Dump Custom Rules + Rate Limit Rules from a zone
port-rules Copy rules between zones (dry-run by default)

Flags

Flag Description
--start, --from Start time: relative (-1h, -2d, -30m), time (16:41), or ISO datetime
--end, --to End time (default: now)
--limit Max results (default: 20)
--path Filter by path pattern using SQL LIKE (/api/%, /api/trpc/%)
--ip Filter by client IP
--score Max bot score for bot-clients (default: 10)
--source, --target Zone name or id (used by port-rules)
--phase custom, ratelimit, or all (default) for rule commands
--out Write export-rules output to a JSON file
--skip-disabled When porting, skip rules with enabled: false
--skip-hosts Comma-separated list of hosts; rules referencing them are skipped
--rewrite-host FROM:TO Replace "FROM" with "TO" in rule expressions during port
--only Comma-separated substrings; only port rules whose description matches
--pro-compat Strip Enterprise-only syntax to fit Pro/Free target zones (see below)
--apply Actually write. Without this, port-rules is a dry run

Examples

# Full scrape analysis for a time window
node .claude/skills/cloudflare/query.mjs scrape --start "2026-03-24 16:41" --end "2026-03-24 19:08"

# Top IPs hitting API endpoints in the last 2 hours
node .claude/skills/cloudflare/query.mjs top-clients --start -2h --path "/api/%"

# What paths is a specific IP hitting?
node .claude/skills/cloudflare/query.mjs ip 1.2.3.4 --start -6h

# Traffic timeline for API/trpc routes
node .claude/skills/cloudflare/query.mjs timeline --start -1h --path "/api/trpc/%"

# Find likely bots (low bot score) hitting API
node .claude/skills/cloudflare/query.mjs bot-clients --start -3h --score 5 --path "/api/%"

# Top user agents on search-related paths
node .claude/skills/cloudflare/query.mjs top-agents --start -2h --path "/api/trpc/image%"

# Check current rate limit rules
node .claude/skills/cloudflare/query.mjs rate-limits

# List all zones (id, name, plan)
node .claude/skills/cloudflare/query.mjs list-zones

# Export civitai.com rules (custom + rate limit) to a file
node .claude/skills/cloudflare/query.mjs export-rules civitai.com --out com-rules.json

# Dry-run a port from civitai.com to civitai.red — preview only
node .claude/skills/cloudflare/query.mjs port-rules \
  --source civitai.com --target civitai.red \
  --skip-disabled \
  --skip-hosts api.civitai.com,metrics.civitai.com,education.civitai.com,image.civitai.com,meilisearch-v1-9.civitai.com,meilisearch-v1-6.civitai.com,meilisearch-metrics.civitai.com \
  --rewrite-host civitai.com:civitai.red \
  --pro-compat

# Same thing, but actually write it (REPLACES target rulesets wholesale)
node .claude/skills/cloudflare/query.mjs port-rules \
  --source civitai.com --target civitai.red \
  --skip-disabled \
  --skip-hosts api.civitai.com,metrics.civitai.com,education.civitai.com,image.civitai.com,meilisearch-v1-9.civitai.com,meilisearch-v1-6.civitai.com,meilisearch-metrics.civitai.com \
  --rewrite-host civitai.com:civitai.red \
  --pro-compat \
  --apply

# Port only specific rules (match by description substring)
node .claude/skills/cloudflare/query.mjs port-rules \
  --source civitai.com --target civitai.red --phase ratelimit \
  --only "Global Limit,Rate limit trpc (No Regex)" \
  --rewrite-host civitai.com:civitai.red \
  --pro-compat --apply

Porting Security Rules Between Zones

port-rules replaces the target zone's Custom Rules ruleset and/or Rate Limit Rules entrypoint with a filtered + transformed copy of the source zone's rules. It overwrites the target ruleset — existing rules on the target are wiped.

What gets ported

The Cloudflare "Security Rules" surface maps to two phase entrypoints:

Phase UI name Flag value
http_request_firewall_custom Custom Rules --phase custom
http_ratelimit Rate Limit Rules --phase ratelimit

Managed Rules, IP Access Rules, Zone Lockdown, and User Agent Blocking are not ported.

Filters

Without filters, every rule from the source is copied as-is.

  • --skip-disabled drops rules where enabled: false.
  • --skip-hosts host1,host2 drops any rule whose expression literal-matches one of the listed hosts (useful for source-only subdomains that don't exist on the target zone).
  • --only pat1,pat2 keeps only rules whose description contains one of the given substrings.
  • --rewrite-host FROM:TO rewrites "FROM""TO" inside rule expressions after the skip filter runs. Use this to point apex-host rules at the new zone (civitai.com:civitai.red).

Pro-plan / Free-plan compatibility (--pro-compat)

Cloudflare Pro zones reject several Enterprise-only rule features that are commonly used on Business/Enterprise source zones. When the target zone plan is pro or free, --pro-compat is applied automatically; otherwise pass it explicitly. Transforms:

  • IP list exclusions: and not ip.src in {…} / and ip.src in {…} clauses are stripped. Pro zones don't support ip.src in expressions in rate-limit rules.
  • Bot management checks: and not cf.bot_management.verified_bot / and cf.bot_management.verified_bot clauses are stripped (Bot Management is an Enterprise feature).
  • Rate-limit period: Pro only allows periods in [10, 15, 20, 30, 40, 45, 60] seconds. Rules with longer periods (e.g. 300s) are scaled to 60s and the requests_per_period is scaled proportionally to preserve the same req/s rate.
  • requests_to_origin: stripped (Enterprise-only).

Each transform is reported in the dry-run output as a ~ line so you can review them before applying.

Rule-count limits

Target zone plans cap the number of rules per phase. If the port exceeds the cap, Cloudflare returns exceeded the maximum number of rules in the phase X: N out of M. Common caps:

Plan Custom Rules Rate Limit Rules
Free 5 1
Pro 20 2
Business 100 15
Enterprise 1000+ 1000+

Use --only or manually trim the source rule list if you hit the cap.

Known constraints

  • ref is stripped on write so rules are re-id'd on the target. This means subsequent ports between the same two zones replace wholesale, not diff-and-merge.
  • Managed Rules, IP Access Rules, Zone Lockdown, User Agent Blocking, and legacy Firewall Rules are not ported.
  • Account-scoped custom lists (e.g. $attackers, $bandwidth_theft) work across zones in the same account with no change.

Investigation Workflow

For scraping investigations:

  1. Start broad: scrape --start ... --end ... to see top IPs, paths, and UAs
  2. Identify suspects: Look for IPs with disproportionate request counts
  3. Deep dive: ip <addr> to see what specific IP is doing
  4. Check bots: bot-clients --score 5 to find CF-detected bots
  5. Timeline: timeline --ip <addr> to see traffic pattern over time
  6. Review defenses: rate-limits and firewall to see what's in place