一键在 Manus 中运行任何 Skill

ai-redteam

星标1

分支0

更新时间2026年5月21日 06:53

AI and LLM red-teaming skill for evaluating prompt injection, jailbreak, data exfiltration, tool abuse, agent hijack, RAG poisoning, model denial of service, and unsafe tool-use chains in Gemini, Claude, GPT, and open-weight models. Use to build evaluation harnesses, attack corpora, defensive guardrails, and red-team reports for AI systems you own or are authorized to test.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

Masriyan

Masriyan/gemini-security-skills

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

AI Red Team

Authorization Boundary

Only target models, agents, and applications the user owns or has written authorization to test.
Do not produce working malware, CSAM, weapons synthesis, or content that bypasses model safety for harmful real-world outcomes.
Treat findings as defensive: every successful attack must ship with a detection and a mitigation.

Attack Taxonomy

Direct injection: instructions in user input override system policy.
Indirect injection: hostile content in retrieved docs, web pages, emails, files, image alt-text, or tool output.
Jailbreak: roleplay, hypothetical, encoding, obfuscation, multi-turn drift, persona pinning.
Tool abuse: forcing browse, shell, code, or MCP tools to act outside scope (SSRF, file read, command exec).
Data exfiltration: leaking system prompt, secrets, embeddings, training data, or per-user memory.
Agent hijack: rewriting plans, looping, escalating permissions, calling unintended sub-agents.
RAG poisoning: index pollution, ranking manipulation, embedding collisions.
Resource exhaustion: token bombs, recursive tool calls, prompt-bomb DoS.

Evaluation Workflow

Map the AI surface: model, system prompt, tools, retrievers, memory, output sinks, and downstream callers.
Build a threat model with abuse cases per surface.
Generate a versioned attack corpus (attacks.jsonl) with {id, technique, payload, expected_block, severity}.
Run the harness; record {pass, fail, partial, refusal_quality, leak_evidence}.
Score with attack success rate per technique, mean time to refuse, and blast radius.

Defensive Output

Input/output classifiers and allow-listed tool schemas.
Provenance tags on retrieved context; quarantine for untrusted sources.
Plan-then-execute pattern with human approval on high-impact tools.
Rate limits, recursion caps, and per-tool budgets.
Red-team report: top techniques, reproductions, fix owners, regression tests.

Deliverables

surface.md, threats.md, attacks.jsonl, harness/, results.csv, report.md, regression-suite/.

同仓库更多 Skills

同仓库

threat-intel-fusion

Masriyan/gemini-security-skills

Threat intelligence fusion skill for collecting, normalizing, deduplicating, enriching, and operationalizing IOCs and TTPs from OTX, MISP, abuse.ch, VirusTotal, Shodan, GreyNoise, Censys, CISA KEV, NVD, vendor blogs, and internal telemetry. Use to build STIX/TAXII feeds, ATT&CK-mapped actor profiles, prioritized blocklists, and detection-ready intel packages.

2026-05-211

api-security-automation

Masriyan/gemini-security-skills

API security automation skill for REST, GraphQL, gRPC, and WebSocket APIs. Covers OpenAPI/AsyncAPI ingestion, authenticated fuzzing, OWASP API Top 10 (BOLA, BFLA, mass assignment, SSRF), schema diffing, GraphQL introspection abuse, JWT and OAuth misuse, rate-limit and replay testing. Use to automate API assessments with safe, scoped, evidence-backed findings.

2026-05-211

bug-bounty-workflow

Masriyan/gemini-security-skills

Bug bounty workflow skill for program scope mapping, recon-to-report automation, deduplication against prior submissions, and high-signal reporting on HackerOne, Bugcrowd, Intigriti, YesWeHack, and self-hosted programs. Use to organize bounty work end to end while staying inside program rules.

2026-05-211

cloud-security-automation

Masriyan/gemini-security-skills

Cloud security posture skill for AWS, Azure, GCP, and multi-cloud. Covers IAM least-privilege, key and secret hygiene, network exposure, data protection, logging coverage, IaC scanning (Terraform, CloudFormation, Bicep, Pulumi), CSPM remediation, and landing-zone hardening. Use for misconfig discovery, IaC review, and automated drift-and-fix workflows on cloud accounts you own.

2026-05-211

detection-engineering

Masriyan/gemini-security-skills

Detection engineering skill for writing, testing, and tuning Sigma, YARA, YARA-L, Suricata, Snort, Zeek, KQL, SPL, EQL, Chronicle, and Elastic detection rules. Covers ATT&CK coverage mapping, telemetry gap analysis, false-positive tuning, detection-as-code pipelines, and validation with Atomic Red Team and Caldera. Use to design durable detections backed by tests.

2026-05-211

forensics-triage

Masriyan/gemini-security-skills

Digital forensics and incident response triage skill for disk, memory, network, cloud, and endpoint artifacts. Covers Volatility 3, Velociraptor, KAPE, Plaso, Autopsy, Chainsaw, Hayabusa, Zeek, Arkime, EVTX, AWS/Azure/GCP audit logs, and timeline reconstruction. Use to extract IOCs, build incident timelines, and produce defensible analyst reports.

2026-05-211

name	ai-redteam
description	AI and LLM red-teaming skill for evaluating prompt injection, jailbreak, data exfiltration, tool abuse, agent hijack, RAG poisoning, model denial of service, and unsafe tool-use chains in Gemini, Claude, GPT, and open-weight models. Use to build evaluation harnesses, attack corpora, defensive guardrails, and red-team reports for AI systems you own or are authorized to test.

AI Red Team

Authorization Boundary

Only target models, agents, and applications the user owns or has written authorization to test.
Do not produce working malware, CSAM, weapons synthesis, or content that bypasses model safety for harmful real-world outcomes.
Treat findings as defensive: every successful attack must ship with a detection and a mitigation.

Attack Taxonomy

Direct injection: instructions in user input override system policy.
Indirect injection: hostile content in retrieved docs, web pages, emails, files, image alt-text, or tool output.
Jailbreak: roleplay, hypothetical, encoding, obfuscation, multi-turn drift, persona pinning.
Tool abuse: forcing browse, shell, code, or MCP tools to act outside scope (SSRF, file read, command exec).
Data exfiltration: leaking system prompt, secrets, embeddings, training data, or per-user memory.
Agent hijack: rewriting plans, looping, escalating permissions, calling unintended sub-agents.
RAG poisoning: index pollution, ranking manipulation, embedding collisions.
Resource exhaustion: token bombs, recursive tool calls, prompt-bomb DoS.

Evaluation Workflow

Map the AI surface: model, system prompt, tools, retrievers, memory, output sinks, and downstream callers.
Build a threat model with abuse cases per surface.
Generate a versioned attack corpus (attacks.jsonl) with {id, technique, payload, expected_block, severity}.
Run the harness; record {pass, fail, partial, refusal_quality, leak_evidence}.
Score with attack success rate per technique, mean time to refuse, and blast radius.

Defensive Output

Input/output classifiers and allow-listed tool schemas.
Provenance tags on retrieved context; quarantine for untrusted sources.
Plan-then-execute pattern with human approval on high-impact tools.
Rate limits, recursion caps, and per-tool budgets.
Red-team report: top techniques, reproductions, fix owners, regression tests.

Deliverables

surface.md, threats.md, attacks.jsonl, harness/, results.csv, report.md, regression-suite/.