cursor-agent-safety
Audits and hardens Cursor agent run mode settings for Cursor 3.6+. Reviews the current run mode, writes project-appropriate classifier instructions (agent-policy.md), builds a terminal allowlist from the project's actual command usage, and tests the configuration against common prompt-injection and scope-creep patterns. Use when setting up a new Cursor project, after adding MCP servers or new tools, or when agents are producing too many approval interrupts.
v1.0.0New
Signing
SignedSLSA L2- Signed by
- skills-hub.ai distributor
- Method
- Distributor-signed by skills-hub.aiCryptographically signed by the skills-hub.ai distributor key at publish time.
- Signed
Install this skill
Run this command in your terminal. No account required — it auto-detects your AI tool and installs the skill file.
npx @skills-hub-ai/cli install cursor-agent-safetyOr download directly:
Browse all CLI commands →Setup by platform
Install
One-click setup for your editorRun in your project root
npx @skills-hub-ai/cli install cursor-agent-safety --target claude-codeInstructions
Security
Loading security scan...
Reviews (0)
Related skills
Browse all →skill-creatorCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.8 installsrecallReconstructs the development cycle from git history, distills sequential/parallel patterns, and produces actionable insights for improving future iterations. Triggers: recall, retrospective, development analysis, what happened, dev cycle analysis.7 installsskillifyCreate a reusable, self-healing, self-evolving global skill — either from the current conversation context OR from a plain-language description of a workflow. Triggers on: 'turn this into a skill', 'make a skill from this', 'skillify this', 'capture this as a skill', 'save this workflow as a skill', 'create a skill for this', 'make this repeatable', 'skill this up', 'build me a skill that does X', 'I want a skill for Y', or any time a workflow just played out and the user wants to preserve it. Invoke even if the user only hints at wanting to reuse what just happened.7 installsweb-gameScaffolds a browser-based game with Phaser 3, PixiJS, or Three.js including game loop, asset pipeline, responsive canvas, input handling, audio, save/load, and deployment to itch.io or web. Triggers on: "web game", "browser game", "html5 game", "phaser game", "pixi game", "three.js game", "javascript game", "typescript game", "make a game for the browser", "build a web game", "2d browser game", "3d web game", "itch.io game", "canvas game", "webgl game".6 installsapi-reviewReview API design against REST best practices and internal consistency. Audits naming conventions, HTTP method semantics, status code correctness, pagination and filtering patterns, error response format, versioning strategy, rate limiting, idempotency keys, HATEOAS links, and content negotiation. Works with Express, Fastify, Hono, Flask, Django REST, Spring, Rails, Gin, and any HTTP API framework. Use when you need to review an API, audit REST endpoints, check API consistency, fix HTTP status codes, add pagination, or prepare an API for public release.6 installsprompt-engineeringLearn effective AI prompting through practice -- good vs bad prompts, context management, the STAR framework, and advanced techniques6 installs
More from Skills Hub
View source →unit-testGenerate comprehensive unit tests with edge cases, error paths, and boundary values. Auto-detects test framework (Vitest, Jest, pytest, go test, flutter_test, RSpec, cargo test, JUnit), scans for untested functions by priority, adopts existing test conventions, generates tests with proper mocks, runs them, and self-heals failures in up to 3 iterations. Measures coverage before and after. Use when you need to add unit tests, increase code coverage, test edge cases, or verify error handling paths.18 installsdesign-to-codeTurn a design into production-quality frontend code: extract a design system with tokens and components, make layouts responsive across breakpoints, add dark mode with WCAG-compliant theming, then run a UX audit to validate everything. Use when implementing UI from Figma, adding theming, making an app responsive, or overhauling frontend design consistency.18 installssecurity-reviewSecurity audit and vulnerability assessment for any codebase. Scans for authentication bypasses, missing auth middleware, broken JWT validation (algorithm confusion, weak secrets, missing expiry), OAuth state and PKCE flaws, IDOR and horizontal privilege escalation, vertical privilege escalation via role manipulation, SQL injection, NoSQL injection, XSS (stored, reflected, DOM), command injection, path traversal, SSRF, CSRF, hardcoded secrets and API keys (sk_live_, AKIA, ghp_), .env and credential file exposure, PII leaking in logs and error responses, overfetching sensitive fields, CORS misconfiguration, session fixation, missing secure/httpOnly/sameSite cookie flags, and Firebase/Firestore rule weaknesses. Produces a severity-ranked findings report with exploit scenarios and fix recommendations. Covers OWASP Top 10.16 installspreflightPre-deploy verification gate. Checks git status, build, tests, migrations, secrets, and commit conventions. Reports READY or NOT READY. Read-only, no changes. Trigger words: preflight, pre-deploy check, ready to deploy, deployment checklist.11 installscleanup-sprintDeep codebase cleanup — kills dead code, fixes all lint/format warnings, removes orphaned files, cleans stale TODOs, strips security hazards, tightens TypeScript strict mode, and organizes imports. Triggers on: clean up, dead code, unused imports, lint, technical debt cleanup, spring cleaning, tidy up the codebase, remove dead code, code hygiene, declutter.10 installsdesign-claudeHTML-first hi-fi design skill: interactive prototypes, slide decks, animation demos, design variant exploration, visual direction advising, and expert critique. Triggers: make a prototype, design demo, interactive prototype, HTML presentation, animation demo, design variants, hi-fi design, UI mockup, prototype, design exploration, make an HTML page, make a visualization, app prototype, iOS prototype, mobile app mockup, export MP4, export GIF, 60fps video, design style, design direction, design philosophy, color scheme, visual style, recommend style, pick a style, make it look good, review, does this look good, review this design. Core capabilities: Junior Designer workflow (show assumptions+reasoning+placeholders first, then iterate), anti-AI-slop checklist, React+Babel best practices, Tweaks variant switching, Speaker Notes presentation, Starter Components (slide shell/variant canvas/animation engine/device frames), App prototype rules (default real images from Wikimedia/Met/Unsplash,8 installs
More Security skills
Browse category →pentestStatic-analysis penetration test that hunts for exploitable vulnerabilities with proof-of-concept payloads and fix code. Covers SQL and NoSQL injection (string concatenation, raw queries, operator injection), XSS (reflected, stored, DOM-based, template injection, dangerouslySetInnerHTML), authentication bypass (missing auth middleware, JWT algorithm confusion, predictable tokens, OAuth state CSRF), authorization flaws (IDOR, mass assignment, horizontal/vertical privilege escalation), path traversal and file inclusion (unsanitized file paths, upload validation, LFI), command injection (exec, system, subprocess with user input), CSRF and SSRF (missing SameSite, user-supplied URLs, open redirects), hardcoded secrets (AWS keys, private keys, JWT secrets, connection strings, .env in git), and insecure deserialization (pickle, yaml.load, XXE, ObjectInputStream). Maps full attack surface with route inventory. Use for pre-release security validation, finding exploitable bugs, or generating pen3 installssecureFull-stack security posture assessment with 0-100 risk scoring. Scans dependency vulnerabilities (npm audit, pip-audit, cargo audit, govulncheck), dangerous code patterns (SQL injection, eval, command injection, ReDoS, innerHTML, XSS vectors), authentication gaps (missing auth middleware, CSRF, hardcoded JWT secrets, insecure session flags), insecure crypto (MD5/SHA1 password hashing, Math.random for tokens, hardcoded encryption keys), configuration issues (exposed .env files, debug mode, permissive CORS, missing security headers CSP/HSTS, Docker root containers, default credentials), and data handling problems (PII in logs, missing input validation, file upload exploits, missing rate limiting). Produces a prioritized risk report and routes to specialized skills (pentest, owasp, gdpr, encryption, soc2). Use as a first-pass security triage before deeper audits or before shipping to production.2 installscheck-vantaFetches dependency vulnerabilities from Vanta, Snyk, Dependabot, or GitHub Security Advisories, creates a tracking issue in Jira/Linear/GitHub Issues, then fixes, commits, pushes, and opens PRs for each affected repo. Trigger on: vulnerabilities, security scan, Vanta, CVE, dependency audit, Snyk, Dependabot.2 installssoc2SOC 2 Type II readiness assessment against all five Trust Service Criteria. Evaluates Security controls (CC6/CC7 -- RBAC, access provisioning/removal, network segmentation, TLS enforcement, input validation, vulnerability management, incident detection and response), Availability controls (A1 -- capacity management, auto-scaling, backup frequency, disaster recovery, RTO/RPO, health checks, uptime monitoring), Processing Integrity (PI1 -- data validation, error handling, transaction logging, idempotency, race condition protection), Confidentiality (C1 -- data classification, encryption at rest and in transit, access logging, secure disposal, key rotation), and Privacy (P1-P8 -- notice, consent, collection limitation, retention/disposal, data access/export, third-party disclosure, data quality, privacy monitoring). Produces a control-by-control PASS/PARTIAL/FAIL matrix with evidence references, remediation roadmap, and evidence collection checklist. Use when preparing for a SOC 2 audit, 1 installsbenefits-fraudAudit government benefits and entitlement systems for fraud prevention, detection, and recovery capabilities. Assesses identity proofing (document verification, SSA cross-match, biometrics, NIST 800-63 IAL levels), synthetic and stolen identity detection, deceased and incarcerated person checks, duplicate applicant matching (fuzzy, probabilistic, Soundex/metaphone), cross-program and cross-jurisdiction benefit matching, income verification (state wages, IRS 1075, new hire reporting), anomaly detection (statistical outliers, behavioral analytics, geographic clustering, ML model bias testing), rule-based fraud scoring, EBT usage pattern analysis, provider and vendor billing fraud, overpayment calculation and recovery (recoupment, Treasury offset, hardship waivers), investigation case management, whistleblower hotline integration, and due process safeguards (notice, hearing rights, demographic bias analysis). Covers cash assistance, SNAP, Medicaid, housing, energy, and childcare programs.1 installsgdprGDPR and CCPA/CPRA privacy compliance audit for codebases. Inventories PII fields (email, phone, SSN, IP, device ID, geolocation, biometrics, behavioral data), maps data collection points (forms, APIs, cookies, analytics, error tracking), audits consent mechanisms (cookie banners, opt-in, pre-checked boxes, consent withdrawal), verifies data subject rights implementation (right to access, erasure, rectification, portability, opt-out, Do Not Sell), traces third-party data sharing (Google Analytics, Facebook Pixel, Stripe, SendGrid, Sentry), and checks data retention policies and automated purging. Use when auditing privacy compliance, building data export or deletion endpoints, reviewing cookie consent, or assessing DSAR readiness.1 installs
Frequently asked questions about cursor-agent-safety
What does the cursor-agent-safety skill do?
Audits and hardens Cursor agent run mode settings for Cursor 3.6+. Reviews the current run mode, writes project-appropriate classifier instructions (agent-policy.md), builds a terminal allowlist from the project's actual command usage, and tests the configuration against common prompt-injection and scope-creep patterns. Use when setting up a new Cursor project, after adding MCP servers or new tools, or when agents are producing too many approval interrupts. It's a reusable SKILL.md instruction set that loads into your AI coding assistant on demand, no prompt engineering, no copy-pasting every session.
How do I install the cursor-agent-safety skill?
Run `npx @skills-hub-ai/cli install cursor-agent-safety` from your terminal. The CLI writes the SKILL.md to the correct location for your AI tool (e.g. ~/.claude/skills/cursor-agent-safety/ for Claude Code or ~/.cursor/skills/ for Cursor with --target cursor) and adds it to your project's .skills.json lockfile.
Which AI tools does cursor-agent-safety work with?
cursor-agent-safety runs in Cursor, Claude Code. It follows the open Agent Skills standard (SKILL.md), so the same skill works in every supported tool without modification.
Is the cursor-agent-safety skill free?
Yes. Every skill on skills-hub.ai is free and open-source. There are no premium tiers, paywalls, or usage limits. You only pay for whatever AI assistant you're already using.
How do I use cursor-agent-safety after installing it?
In Cursor, type `/cursor-agent-safety` (or whatever slash command the skill registers) and the AI follows the skill's instructions immediately. You can also reference it by name in natural language, your AI loads the skill into context when relevant.
Can I share the cursor-agent-safety skill with my team?
Yes. Commit your project's .skills.json lockfile and teammates run `npx @skills-hub-ai/cli install` (no args) to install every skill at the exact version you pinned. Organization-scoped installs work via skills-hub.ai organizations.