Decepticon is an autonomous red-team agent that coordinates AI agents, security tools, sandboxes, and supporting services for authorized cybersecurity assessments. Security researchers and red teams can run it through its Docker stack, cloud service, command-line interface, or Python SDK, with the catalogue entries representing its available skills.
Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/PurpleAILAB/Decepticonnpx agentmods add skills/purpleailab/decepticon/webWrote 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/purpleailab/decepticon/web)<a href="https://agentmods.dev/skills/purpleailab/decepticon/web"><img src="https://agentmods.dev/badge/skills/purpleailab/decepticon/web/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/purpleailab/decepticon/web"><img src="https://agentmods.dev/badge/skills/purpleailab/decepticon/web.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Privilege Escalation · line 78 Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.Fix: Remove references to credential paths. Use environment variables or secrets managers. For docs, use placeholder paths (e.g., /path/to/config). Never load .env or token files in production code paths.
- medium Data Exfiltration · line 84 Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.Fix: Verify the destination URL is trusted and necessary. Remove or replace with documented APIs. Ensure no secrets, tokens, or PII are transmitted.
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 | $0.00060 | $0.03773 |
| Opus 5.5 | $0.00024 | $0.01509 |
| Sonnet 5.5 | $0.00012 | $0.00755 |
| Haiku 4.5 | $0.00006 | $0.00377 |
Grade A, and why
web scanned grade A 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 1mo 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.
Sends data to an external URLlowData exfiltration
A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.
curl -s 'https://<TARGET>/api' -H 'Content-Type: application/xml' -d '<?xml version="1.0"?><test>hello</test>' Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
required_tools: [sandbox_web, curl] Copies of this mod
1 near-identical copy found in the catalogue:
- web — 94% identical, 26 lines differ
How it starts
The opening of the file, as written. The whole thing — 142 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Web Application Exploitation — Category Overview
This is a routing skill. It helps you identify the correct attack technique, then directs you to the specialized sub-skill with full exploitation procedures.
Attack Technique Routing
Match the target's characteristics to the right sub-skill:
| Sub-Skill | Covers | When to Load | Path |
|---|---|---|---|
| sqli | Union/Error/Blind/Time-based SQL injection, sqlmap | SQL database, query parameters, login forms, search, filtering | load_skill("/skills/standard/exploit/web/sqli/SKILL.md") |
| blind-sqli | Manual WAF-bypass companion to sqli — token-fingerprinting probe loops, arithmetic-multiplication boolean evaluation, hex-encoded literals, exponential-probe binary search | Load AFTER sqli when sqlmap with --tamper cannot pass the WAF but a binary oracle (two distinct page states) exists; challenge tag is blind_sqli with active filtering |
load_skill("/skills/standard/exploit/web/blind-sqli/SKILL.md") |
| xss | Reflected/stored/DOM XSS, bot exfiltration, CSP bypass | Client-side JS injection, bot/report URL, cookie stealing | load_skill("/skills/standard/exploit/web/xss/SKILL.md") |
| ssti | Jinja2, Twig, Freemarker, ERB, Razor template injection | Template rendering, {{}} or ${} in output, Flask/Symfony/Java |
load_skill("/skills/standard/exploit/web/ssti/SKILL.md") |
| ssrf | Cloud metadata, internal service access, Gopher smuggling | URL fetch parameter, redirect, internal network access | load_skill("/skills/standard/exploit/web/ssrf/SKILL.md") |
| xxe | XML entity injection, SOAP/WSDL, blind OOB | XML processing, SOAP endpoints, XML file uploads | load_skill("/skills/standard/exploit/web/xxe/SKILL.md") |
| lfi | Path traversal, PHP wrappers, log poisoning | File path parameters, ../, include/require, file download |
load_skill("/skills/standard/exploit/web/lfi/SKILL.md") |
| command-injection | OS command injection, blind/OOB, filter bypass | System commands, ping/traceroute, exec, subprocess | load_skill("/skills/standard/exploit/web/command-injection/SKILL.md") |
| deserialization | Java/PHP/.NET/Python deserialization RCE | Serialized objects, base64 blobs, ViewState, pickle | load_skill("/skills/standard/exploit/web/deserialization/SKILL.md") |
| idor | Authorization bypass, ID enumeration, privilege escalation | Object references, sequential IDs, UUIDs, access control | load_skill("/skills/standard/exploit/web/idor/SKILL.md") |
| file-upload | Webshell upload, extension/content-type bypass | File upload forms, unrestricted upload | load_skill("/skills/standard/exploit/web/file-upload/SKILL.md") |
| graphql | Introspection, SQLi via resolvers, auth bypass | GraphQL API, /graphql endpoint, GQL queries | load_skill("/skills/standard/exploit/web/graphql/SKILL.md") |
| race-condition | TOCTOU, parallel POST/GET races, session-write-before-verdict, quota/balance/coupon double-spend | bcrypt/Argon2 auth, check-then-act, slow-op widening race window, challenge tag includes race_condition/toctou/concurrent | load_skill("/skills/standard/exploit/web/race-condition/SKILL.md") |
| smuggling | HTTP request smuggling (HRS) — CL.TE/TE.CL/TE.TE, CL.0, HTTP/2 downgrade (h2.cl, h2.te, CR/LF injection), pipelining, connection-state pinning | Multi-proxy/CDN frontend, differential 4xx/5xx on duplicate or obfuscated TE/CL headers, two Server: strings, h2 frontend with h1 backend, challenge tag includes smuggling_desync/request_smuggling/hrs/desync |
load_skill("/skills/standard/exploit/web/smuggling/SKILL.md") |
| crypto | Padding oracle (Vaudenay), AES-CBC bit-flipping, ECB block substitution, JWT alg confusion, hash-length extension | Base64 cookie/token w/ length %16 or %8, distinct invalid-pad vs auth-fail responses, JWT, repeated 16-byte ciphertext blocks, challenge tag includes crypto/cipher/oracle/captcha |
load_skill("/skills/standard/exploit/web/crypto/SKILL.md") |
| business-logic | POST-body privilege fields, 2FA bypass, predictable TOTP codes, hidden auth headers, multi-step workflow tampering | Challenge tag includes business_logic, privilege_escalation, 2fa_bypass, auth_bypass (not pure IDOR/JWT) |
load_skill("/skills/standard/exploit/web/business-logic/SKILL.md") |
| cve | Known CVE exploitation — fingerprint → cve_lookup → cve_poc_lookup → adapt PoC → flag sweep; CMS/plugin/framework version-specific vulnerabilities |
Challenge tag includes cve, recon fingerprinted a versioned CMS/framework/plugin, challenge name hints at specific software (WordPress, Joomla, Struts, Spring4Shell, Log4j) |
load_skill("/skills/standard/exploit/web/cve/SKILL.md") |
| jwt | JSON Web Token attacks — alg=none, RS256↔HS256 confusion, kid header injection, JWKS spoofing, weak HMAC cracking, signature stripping | Authorization: Bearer eyJ..., id_token= / access_token=, .well-known/jwks.json, .well-known/openid-configuration |
load_skill("/skills/standard/exploit/web/jwt/SKILL.md") |
| oauth | OAuth 2.0 / OIDC abuse — redirect_uri smuggling, state CSRF, code/token leak via referer, PKCE downgrade, scope escalation | /oauth/authorize, /oauth/callback, response_type=code|token, client_id= parameters |
load_skill("/skills/standard/exploit/web/oauth/SKILL.md") |
| saml | SAML SSO abuse — XSW (signature wrapping), XML signature stripping, ACS URL substitution, IdP confusion, comment-in-NameID truncation | SAMLRequest= / SAMLResponse= POST bodies, /Shibboleth.sso/, /saml/acs, ADFS endpoints |
load_skill("/skills/standard/exploit/web/saml/SKILL.md") |
| ato-methodology | Account Takeover end-to-end — credential stuffing, password reset poisoning, email/phone change race, MFA fatigue, session fixation | /login, /reset-password, /account/email, /account/2fa endpoints, MFA enrollment flows |
load_skill("/skills/standard/exploit/web/ato-methodology/SKILL.md") |
| nosqli | NoSQL injection — MongoDB $ne/$gt/$where, Redis CRLF, Cassandra CQL, Couch view injection, DynamoDB filter abuse |
MongoDB/Redis/Cassandra/Couch/Dynamo backends, JSON request bodies with operator-like keys | load_skill("/skills/standard/exploit/web/nosqli/SKILL.md") |
| ldapi | LDAP injection — anonymous bind, wildcard filters, attribute exfil, blind boolean LDAP via attribute presence | LDAP-backed login, cn=/uid=-shaped queries, AD-joined web apps with form auth |
load_skill("/skills/standard/exploit/web/ldapi/SKILL.md") |
| xpath-xslt | XPath / XSLT injection — auth bypass, blind boolean extraction, XSLT RCE via document() / php:function, XEE |
XML-backed search/login, XSLT-transformed responses, ?xsl= parameters |
load_skill("/skills/standard/exploit/web/xpath-xslt/SKILL.md") |
| mass-assignment | Mass-assignment / over-posting — privilege fields (isAdmin, role, verified), nested object injection via JSON, GraphQL input-object abuse |
REST/GraphQL POST/PATCH with JSON bodies, ORM-backed APIs (Rails, Django, Express+Mongoose, Spring) | load_skill("/skills/standard/exploit/web/mass-assignment/SKILL.md") |
| open-redirect | Open redirect — auth callback, SSO relay state, OAuth redirect_uri, header-injection-based, JS-based location overrides |
?url=, ?next=, ?returnTo=, ?redirect=, ?continue= parameters; SSO RelayState |
load_skill("/skills/standard/exploit/web/open-redirect/SKILL.md") |
| cache-deception | Web cache deception — path confusion (/account.css, /account/index.css), Cloudflare/Varnish/CDN cache key abuse, header smuggling |
CDN-fronted apps (Cloudflare, Fastly, Akamai), Cache-Control permissive on static suffixes, per-user pages cacheable |
load_skill("/skills/standard/exploit/web/cache-deception/SKILL.md") |
| dom-clobbering | DOM clobbering — <form id=config> / <a name=cfg> to shadow JS globals, bypass JS validation, defeat client-side sanitizers |
Sites accepting user HTML, postMessage handlers using window.config.*, jQuery-extend patterns |
load_skill("/skills/standard/exploit/web/dom-clobbering/SKILL.md") |
| xs-leaks | Cross-Site Leaks — frame counting, COOP/COEP probing, performance-timing oracles, navigation-side-effects, error-event leaks | Authenticated cross-origin reads, OAuth response inference, SSO state leak, account-existence oracles | load_skill("/skills/standard/exploit/web/xs-leaks/SKILL.md") |
| proxy-misconfig | Reverse-proxy misconfig — origin bypass via Host/X-Forwarded-Host, ALB/ELB stripping, header-trust abuse, SSRF via internal proxy | Cloudflare / nginx / HAProxy / Envoy frontends, IP-allow-list endpoints (/admin), trusted-header auth |
load_skill("/skills/standard/exploit/web/proxy-misconfig/SKILL.md") |
What ships with it
42 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- ato-methodology/SKILL.md 4.2 KB
- bfla/SKILL.md 11 KB
- blind-sqli/SKILL.md 10 KB
- business-logic/SKILL.md 7.3 KB
- cache-deception/SKILL.md 4.1 KB
- clickjacking/SKILL.md 5.4 KB
- command-injection/SKILL.md 5.3 KB
- cors/SKILL.md 4.7 KB
- crypto/SKILL.md 19 KB
- csrf/SKILL.md 6.1 KB
- cve/SKILL.md 11 KB
- deserialization/SKILL.md 20 KB
- dns-rebinding/SKILL.md 8.5 KB
- dom-clobbering/SKILL.md 4.1 KB
- file-upload/SKILL.md 4.2 KB
- graphql/SKILL.md 5.2 KB
- header-injection/SKILL.md 11 KB
- hpp/SKILL.md 8.3 KB
- idor/SKILL.md 16 KB
- jwt/SKILL.md 4.8 KB
- ldapi/SKILL.md 2.3 KB
- lfi/SKILL.md 20 KB
- mass-assignment/SKILL.md 4.6 KB
- mfa-bypass/SKILL.md 6.3 KB
- nosqli/SKILL.md 3.4 KB
- oauth/SKILL.md 6.8 KB
- open-redirect/SKILL.md 4.0 KB
- php-type-juggling/SKILL.md 8.1 KB
- proxy-misconfig/SKILL.md 4.7 KB
- race-condition/SKILL.md 7.1 KB
- saml/SKILL.md 5.3 KB
- smuggling/SKILL.md 38 KB
- sqli/SKILL.md 11 KB
- ssrf/SKILL.md 7.2 KB
- ssti/SKILL.md 17 KB
- verb-tampering/SKILL.md 8.9 KB
- waf-bypass/SKILL.md 13 KB
- web-cache-poisoning/SKILL.md 5.6 KB
- xpath-xslt/SKILL.md 2.6 KB
- xs-leaks/SKILL.md 4.1 KB
- xss/SKILL.md 15 KB
- xxe/SKILL.md 5.4 KB
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.
- 1mo ago First seen · 142 lines · 60 tokens per session scan A 98b7a455bedf
web is a skill published in the GitHub repository PurpleAILAB/Decepticon (6,286 stars, last pushed 7d ago), licensed Apache-2.0. It adds 60 tokens to every session and 3,773 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 2 findings (sends data to an external url, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
interactive-dashboard
Interactive web dashboards: stock trackers, sector heatmaps, portfolio monitors, served via preview URL.
company-profile
One-page company profile slide, four quadrants of overview, financial summary, share price chart and key facts. Also the compact profile paragraph when it has to sit inside a memo, a deck or a chat reply. Triggers on one-page profile, tear sheet, company snapshot, profile slide, quick profile of [company].
idea-generation
Find long and short candidates across a universe when no name is on the table yet: mandate, universe validation, archetype screens, thematic sweep, triage, idea cards, idea log. Triggers on idea generation, stock screen, find ideas, what looks interesting, screen for, new ideas, pitch me something.
onboarding
First run after the user saves their investing profile: import holdings and watchlists from the brokerage they connected or the screenshots they attached, then propose three first tasks (one on their portfolio or watchlist, one from their profile, one on today's market) and do the one they pick.
python-lib-analyzer
Analyze any Python library structure, explore modules, classes, and functions with signatures and documentation.
research-report
A method for writing a research report, topic review, or comparison from supplied material. It separates evidence, source authors' views, and the writer's inferences while keeping sources identifiable.