Jev Attack — Development skill for Claude Code
Attack your own gate with 300 prompt injections and report what got through.
How to install Jev Attack
Installs to ~/.claude/skills/eugeniughelbur-jev-engineering-jev-attack/SKILL.md
mkdir -p ~/.claude/skills/eugeniughelbur-jev-engineering-jev-attack && curl -fsSL https://raw.githubusercontent.com/eugeniughelbur/jev-engineering/HEAD/commands/jev-attack.md -o ~/.claude/skills/eugeniughelbur-jev-engineering-jev-attack/SKILL.md Restart Claude Code, or start a new session, for it to be picked up.
What Jev Attack does
description: Attack your own gate with 300 prompt injections and report what got through
Find out whether the gate holds before trusting it. This spends real money, about half a cent, so confirm first.
Tell the user what it will do and what it costs: roughly 300 calls against `typesafe/jev-1.13`, about $0.006, taking two to three minutes. Ask them to confirm.
Offer the quick version if they want a smaller bill:
uv run "${CLAUDE_PLUGIN_ROOT}/attack_test.py" --quick
Alternatives in Development
- Quick Plan — Create a detailed implementation plan based on the user's requirements provided through the USER_PROMPT variab 3.4k ★
- Verify Theme — Run the full Graphite quality gate (drift, tokens, contrast, YAML) and report pass/fail 452 ★
- Brainstorm Requirements — Use this prompt to systematically brainstorm and refine initial requirements through structured discovery and 85 ★
Full documentation available on GitHub
View Source RepositoryRelated Skills
Open Jev Approvals
Fail-closed binary (allow/deny) approval gate for Codex and Claude Code hooks — every intercepted tool call is
Validate Agent
Gate a single agent finding through the FP gate + Negative-Result gate before writing any report. Kills findin
Betterbest
You've already got good. A Claude Code skill for making anything as good as it can be through iterative experi
Jev Permission Gate
A Claude Code mod that uses TypeSafe's Jev to decide auto mode tool calls. 2x faster than the built-in classif
Claude Code Jev
Experimental Jev permission gate for Claude Code via OpenRouter, with reproducible latency and cost benchmarks
Jev Skill Gate
Cut Claude Code's skill manifest by ~75% with TypeSafe Jev. Scores every installed skill for relevance and hid
Related Agents
Typesafe Adversary
Red-teams TypeSafe integrations before users do. Probes the nine documented Jev failure modes, tests prompt in
Deep Security Auditor
Adversarial red-team and security compliance specialist. Inspects ASTs for prompt injections, credential leaks
Qc Engineer
Use this agent when any change has been through the engineering-lead's integration gate and needs to be tested