Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add forefy/.context --skill cdn-peekgit clone --depth 1 https://github.com/forefy/.contextWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/forefy/.context/cdn-peek)<a href="https://agentmods.dev/skills/forefy/.context/cdn-peek"><img src="https://agentmods.dev/badge/skills/forefy/.context/cdn-peek/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/forefy/.context/cdn-peek"><img src="https://agentmods.dev/badge/skills/forefy/.context/cdn-peek.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00039 | $0.02578 |
| Opus 5 | $0.00019 | $0.01289 |
| Sonnet 5 | $0.00008 | $0.00516 |
| Haiku 4.5 | $0.00004 | $0.00258 |
Grade C, and why
cdn-peek scanned grade C with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Downloads and executes remote codehighSupply chain
curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.
curl -s "https://crt.sh/?q=%25.DOMAIN&output=json" | python3 -c "import sys,json;print('\n'.join(sorted({n for x in json.load(sys.stdin) for n in x['name_value'].split(chr(10))})))" Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
This skill is deliberately dependency-free: every step is a plain `dig`, `curl`, `openssl`, `whois`, or `nc` one-liner. It covers origin reachability, not WAF signature evasion (payload encoding / rule bypass) - that is How it starts
The opening of the file, as written. The whole thing — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Scope & authorization
Only run against hosts the user owns or is contractually engaged to test. This is origin discovery for defensive hardening (confirming the edge actually hides the origin), not for attacking third parties. If the target isn't clearly the user's or in-scope, ask before probing.
This skill is deliberately dependency-free: every step is a plain dig, curl, openssl, whois, or nc one-liner. It covers origin reachability, not WAF signature evasion (payload encoding / rule bypass) - that is a separate discipline. Techniques that need API keys or heavy tooling (favicon-hash pivots on Shodan/Censys/ZoomEye/FOFA, historical-DNS via SecurityTrails, passive-source or brute-force subdomain enumeration, app-callback/SSRF leaks) are intentionally out of scope here; note them as unrun if the target warrants a deeper pass.
Step 0 - identify the edge
Match the host's IPs/NS/CNAME against known edge fingerprints, then load that provider's client-IP header. Everything after is provider-agnostic; only the IP ranges and the client-IP header name differ.
| Edge | Fingerprint (IP/NS/CNAME) | Client-IP header |
|---|---|---|
| Cloudflare | 104.16/13, 172.64/13, 2606:4700::/32, *.ns.cloudflare.com | CF-Connecting-IP |
| CloudFront | *.cloudfront.net, AWS CLOUDFRONT ranges | CloudFront-Viewer-Address |
| Akamai | *.akamaiedge.net, *.akamai.net, *.edgekey.net | True-Client-IP |
| Fastly | *.fastly.net, 151.101/16 | Fastly-Client-IP |
| Sucuri | *.sucuri.net, 192.88.134/23, 185.93.228/22 | X-Sucuri-ClientIP |
| Imperva/Incapsula | *.incapdns.net | Incap-Client-IP |
| Azure Front Door | *.azurefd.net, AzureFrontDoor.Backend tag | X-Azure-ClientIP |
| Google Cloud CDN | Google ranges, ghs.googlehosted.com | X-Forwarded-For |
| Gcore / Bunny / Edgio | *.gcdn.co, *.b-cdn.net, *.edgecastcdn.net | X-Forwarded-For |
Methodology (run in order; each step is a check with a done/result state)
- Confirm fronted.
digA/AAAA/NS/CNAME, match IPs against the edge ranges from Step 0. If not fronted, stop - there's nothing to bypass. - DNS record sweep.
digA/AAAA/CNAME/TXT/MX/SOA/SRV/CAA and_dmarc. SPFip4:/include:, MX, and autodiscover records frequently name a non-fronted origin or mail IP. - CT lookup (free, no key, no brute). One
curlto crt.sh and certspotter for*.domain: enumerates subdomains and the host's own cert history. Flag any subdomain resolving to a non-edge IP; a Let's Encrypt / self-issued cert scoped to just the target host (not the edge's shared cert) is a pivotable leak. - Fingerprint every non-edge IP.
dig -x(reverse DNS),whois(ASN/hosting org),curlbanner,openssl s_clientcert CN/SAN/issuer. Reveals the hosting footprint even when the box isn't the origin. - Host-confusion.
curl --resolve target:443:<IP>with the target Host + SNI against each non-edge IP; compare title/body/length to the real fronted response. Same content = origin bypass; different (default vhost) = not the origin. - Response-header & error-page leak.
curl -Ithe target and a bogus path; inspectVia,X-Cache,X-Served-By,X-Backend-Server,Server,X-Powered-By, redirectLocation, and verbose error bodies for internal hostnames or RFC1918/origin IPs. - Client-IP header trust. Send
X-Forwarded-For,X-Real-IP, and the edge's own client-IP header (Step 0) =127.0.0.1; compare responses. Bypasses IP allowlist/geo/rate-limit logic - not the origin. A hardened edge rejects a client-supplied copy of its own header (Cloudflare answers 403error code: 1000); if the app instead trusts it, that is the finding. - Edge header tricks.
X-Forwarded-Host,X-Original-URL, blank/mismatchedHost- test for cache/redirect poisoning, routing bypass, and header reflection in links. - Confirm candidate origin. For any candidate IP from steps 2/3/4:
curl -skI --resolve target:443:<IP>. If it returns the target app and isn't firewalled to the edge ranges, it's a full bypass. - Port check on a confirmed origin.
nc -zthe non-HTTP services the edge never proxies (SSH/22, k8s API/6443, kubelet/10250, DB/admin ports). - Shared-egress re-front (origin firewalled to the edge is still reachable). When step 9's direct fetch is refused because the origin's firewall/SG only admits the edge's IP ranges, that lock is necessary but not sufficient on a shared-egress CDN - those ranges are shared by every tenant of the provider. Stand up your own zone on the SAME provider (e.g. a free Cloudflare account: a proxied A record -> the target's origin IP, SSL mode Full), and send traffic through it: it now exits the shared edge IPs and passes the firewall, skipping the victim's WAF, rate limits, bot rules, and geo - all of which lived on their edge config, not yours. If the origin answers, the edge-range allowlist is bypassed and only origin authentication closes it. Needs a known origin IP (steps 2-4). This is a distinct bug class ("CDN origin-trust bypass"): the control authenticated a network position, not the request.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago Changed · +7 lines 50d9836ffac5
- 11d ago First seen · 76 lines · 39 tokens per session scan C 14825df39525
cdn-peek is a skill published in the GitHub repository forefy/.context (145 stars, last pushed yesterday), licensed MIT. It adds 39 tokens to every session and 2,578 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
eresus-manual-security-audit
Elite manual security code review skill for deep, adversarial vulnerability hunting and exploit-chain discovery. Trigger when the user asks to: "do a deep security audit", "manual code review", "find exploit chains", "hunt for logic bugs", "red-team this codebase", "do an offensive security review", "review this like…
eresus-php-audit
Deep PHP-specific security audit skill covering injection, deserialization, file operations, auth bypass, POP chain discovery, and CMS-specific patterns. Trigger when auditing PHP code: "audit this PHP app", "find PHP security issues", "check Laravel/WordPress for vulnerabilities", "PHP SAST review", "check for PHP…
eresus-sast-scanner
General-purpose Static Application Security Testing (SAST) skill for code vulnerability analysis. Trigger when the user asks to: "analyze code for vulnerabilities", "review code security", "find security bugs", "do a SAST scan", "check for [vulnerability type] in code", "audit source code", or requests a security code…
eresus-python-audit
Deep Python-specific security audit skill with 50+ vulnerability class coverage across 7 categories. Trigger when auditing Python code: "audit this Python app", "find Python security issues", "check Flask/Django for vulnerabilities", "Python SAST review", "check for pickle vulnerabilities", "review this FastAPI code".…
eresus-remediator
Security remediation skill for fixing confirmed or likely SAST findings in source code. Trigger when the user asks to: "fix a vulnerability", "patch this security bug", "remediate SAST findings", "harden this endpoint", "make this auth flow safe", or wants code changes that remove a confirmed security issue while…
eresus-variant-analysis
GHSA/CVE variant analysis workflow for finding similar vulnerability patterns across a codebase. Trigger when the user asks to: "find variants of this CVE", "GHSA variant analysis", "find similar bugs", "hunt for the same pattern", "are there other places with this vulnerability?", or when a known vulnerability is…