⏳ This skill is pending AI review.
Scores will appear once the review pipeline completes.
vercel-deepsec
Detects broad web application security vulnerabilities using the Vercel DeepSec benchmark prompt. Use when benchmarking security review coverage or running an open-ended appsec scan for auth bypass, missing auth, XSS, RCE, SQL injection, SSRF, path traversal, secrets, weak crypto, unsafe redirects, webhook verification, Next.js Server Actions, Lua/OpenResty, Go, cache poisoning, or header trust bugs.
Choose how to use this skill
You do not need every option. Choose the path your AI client supports. The stable page stays the same; versioned files are immutable.
1. Native installer
This listing has no registered native installer command. Use the complete package or source fallback below, depending on what your client supports.
Do not guess an installer command or replace an existing version without reviewing the diff.
2. Complete package recommended
Download the ZIP when available. It includes SKILL.md plus the references, security notes and version metadata.
No complete ProSkills package is published for this listing yet.3. Prompt-only
Copy the prompt above when the agent can read the stable page or when you want to adopt the workflow without installing a skill.
Need only the instruction file?
Download SKILL.md only if your client requires a single file. The complete ZIP is safer for a full installation because it preserves the references and release context.
No path installs or executes anything by itself. Your agent still needs access to the project files. Before updating, compare the installed version and review the diff.
// RATINGS
// README
vercel-deepsec
This Warden skill adapts the Vercel Labs DeepSec default processor prompt into SKILL.md form.
It is used to benchmark broad security-review behavior against the DeepSec prompt. It is intentionally wider than most production Warden skills, so focused skills such as authz, code execution, data exfiltration, GitHub Actions, or PII should remain preferred when the review target is narrower.
Upstream
- Repository: https://github.com/vercel-labs/deepsec
- Source prompt:
packages/processor/src/index.ts - Retrieved commit:
cf8bae9ad37e22a4f23c675493c4c553c6379162
The upstream Apache 2.0 LICENSE is included intact. The upstream NOTICE is included because DeepSec ships one.
// HOW IT'S BUILT
// FILES
2 files shipped with this skill. Inline text is the catalog copy. The upstream link is the source repository.
skills/vercel-deepsec/SKILL.md7.8 KBupstream ↗
--- name: vercel-deepsec description: Detects broad web application security vulnerabilities using the Vercel DeepSec benchmark prompt. Use when benchmarking security review coverage or running an open-ended appsec scan for auth bypass, missing auth, XSS, RCE, SQL injection, SSRF, path traversal, secrets, weak crypto, unsafe redirects, webhook verification, Next.js Server Actions, Lua/OpenResty, Go, cache poisoning, or header trust bugs. allowed-tools: Read Grep Glob Bash --- You are a senior application security researcher. This skill adapts the Vercel Labs DeepSec default processor prompt into Warden form. It is intentionally broad because it is used for benchmark and comparison runs. For focused production review, prefer a narrower Warden skill when one maps directly to the concern. When this benchmark skill runs, keep the bar high: report only exploitable vulnerabilities with a traced source, sink, missing guard, impact, and fix. Source provenance and benchmark notes live in `SOURCES.md`. ## Benchmark Contract - Use the same criteria across runs. Do not tune the analysis to expected answers. - Treat scanner hits, grep hits, and suspicious filenames as starting points only. - Investigate beyond the flagged pattern when the surrounding file exposes a different bug. - Return no findings for generated, vendored, gitignored, build output, fixture-only, or non-production code unless the benchmark target explicitly includes it. - Do not inflate severity to make a benchmark look better. A noisy high is worse than an empty result. ## Trace. Do Not Skim. The sink tells you what could happen. The source tells you whether it will. - Read each target file fully before reporting. - Identify attacker-controlled input: request body, query string, path params, cookies, headers, webhook payloads, OAuth callback params, uploaded files, user profile fields, database values written by users, third-party callbacks, or caller-controlled workflow/service inputs. - Identify the security boundary: login state, tenant, team, org, project, account, role, OAuth state, webhook signature, internal network, filesystem root, cache namespace, or paid API quota. - Follow imports, wrappers, middleware, guards, serializers, validators, route definitions, shared utilities, and sibling handlers. - Verify mitigations in context. Parameterized queries, exact allowlists, safe URL fetchers, escaping, signature checks, handler-level auth wrappers, and resource-level ownership checks can close the path. - Use `rg` and `git log -p <file>` when needed to compare sibling handlers or see whether a guard was recently removed. - Drop speculation. If the dataflow, boundary, or missing mitigation cannot be proven from available code, do not report it. ## Severity | Level | Criteria | |-------|----------| | high | Remote code execution, authentication bypass with broad account or tenant access, missing auth on sensitive operations, privilege escalation, unrestricted file upload to execution, SQL/NoSQL injection over sensitive data, SSRF to internal services or cloud metadata, unsafe deserialization of attacker data, hardcoded production credentials, or cross-tenant access to sensitive resources. | | medium | XSS with user-controlled payload execution, open redirect that affects auth or token flow, path traversal with bounded but meaningful file access, webhook without signature verification on sensitive side effects, weak JWT or OAuth validation, missing rate limit on sensitive or expensive actions, race condition in auth/payment/state transition, or info disclosure of internal fields, tokens, PII, or stack traces. | | low | Defense-in-depth issue with a plausible but limited path, weak crypto or random generation that does not yet protect sensitive data, incomplete validation around low-impact operations, or a benchmark-relevant hardening gap with concrete exploit preconditions. | Pick the lower severity when impact depends on unproven preconditions. ## Vulnerability Categories Use these slugs when they fit. For novel issues, use `other-<specific-name>`. | Slug | Report when | |------|-------------| | `auth-bypass` | Authentication checks can be skipped, spoofed, confused, or reached only through bypassable client/edge middleware. | | `missing-auth` | A sensitive HTTP endpoint, RPC, server action, admin path, or service handler has no effective authentication. | | `acl-check` | RBAC, permission, role, team, tenant, org, account, or ownership checks are absent, inverted, stale, or checked against the wrong actor. | | `cross-tenant-id` | User-supplied IDs reach lookups or mutations without scoping to the authenticated tenant, account, org, team, project, or owner. | | `server-action` | A Next.js Server Action or equivalent callable server function performs sensitive work without explicit auth and authorization. | | `jwt-handling` | JWT signing or verification accepts weak algorithms, missing audience/issuer/expiry, unpinned algorithms, unsigned tokens, or user-controlled key selection. | | `webhook-handler` | Webhook payloads trigger state changes without signature verification, timestamp freshness, replay protection, or source validation. | | `rce` | Request-controlled data reaches `eval`, dynamic function construction, shell execution, unsafe template compilation, unsafe deserialization, or equivalent code execution. | | `sql-injection` | User input reaches raw SQL/NoSQL query construction without parameterization or strict allowlisting. | | `xss` | User-controlled data reaches HTML, DOM, script, URL, or dangerous framework escape hatches without context-correct escaping or sanitization. | | `dangerous-html` | `innerHTML`, `dangerouslySetInnerHTML`, `unsafeHTML`, template source, Markdown HTML, or inline script receives data that may contain user content. | | `ssrf` | User-controlled URLs, hosts, redirects, webhooks, image proxies, fetchers, or Go/Node/Python HTTP clients can reach internal networks or metadata services. | | `path-traversal` | User-controlled paths, archive entries, filenames, route params, or object keys can escape an intended root. | | `secrets-exposure` | Real credentials, API keys, tokens, private keys, or auth material are hardcoded, sent to clients, logged, or exposed through fallback values. | | `secret-env-var` | Server secrets are read in code that can ship to clients or be exposed to untrusted execution contexts. | | `env-exposure` | Secret or sensitive values are placed under public client prefixes such as `NEXT_PUBLIC_` or equivalent. | | `secret-in-fallback` | A secret environment variable has a hardcoded fallback that would become a production credential or shared secret. | | `secret-in-log` | Credentials, auth headers, cookies, tokens, or signed URLs are logged or returned in errors. | | `insecure-crypto` | Weak hashes, ECB mode, static IVs, timing-unsafe compares, predictable randomness, or custom crypto protect security-sensitive data. | | `open-redirect` | User-controlled redirects affect login, OAuth, SSO, token flows, phishing-resistant flows, or trusted callback destinations. | | `unsafe-redirect` | Redirect validation is substring, prefix, regex, double-encoding, path normalization, or origin-confusion based. | | `public-endpoint` | A public or anonymous endpoint exposes sensitive data or performs sensitive side effects. | | `service-entry-point` | A service handler trusts caller identity, headers, or internal-only assumptions without verifying the boundary. | | `iam-permissions` | Cloud IAM policy, token scope, or resource pattern grants more privilege than the code path needs and is reachable through an exploit path. | | `rate-limit-bypass` | Sensitive operations such as login, token refresh, password reset, MFA, invite, billing, export, or expensive API calls lack abuse controls. | | `expensive-api-abuse` | LLM, AI, billing, email, SMS, search, export, or paid third-party calls are reachable without quota, auth, or dedupe controls. | | `cachREADME.mdupstream ↗
Contents are not inlined in the catalog. Open the upstream file.
Plain-text copy, no login: /skill-bundles/getsentry-warden-skills-skills-vercel-deepsec.md