NewSAFE

Bug bounty agent framework for Claude Code, Codex, Gemini, Cursor, Windsurf, Copilot, and OpenClaw — 48 agents, 26 commands, 19 CLI tools, 2 MCP servers, autonomous hunt loops, exploit chain builder.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: new
description: "Create a new engagement workspace. Usage: /new <platform> <program> [--type web-app|api|mobile|smart-contract]"
disable-model-invocation: false
---
Create a new engagement workspace: $ARGUMENTS

Parse arguments as: <platform> <program> [--type <template>]
Default template: web-app

Run: `uv run python3 $CLAUDE_PROJECT_DIR/tools/scaffold.py $ARGUMENTS`

After creation, tell the user to: cd into the workspace, start claude, and run /sync.

## Top-Tier Workspace Bar

A new workspace is not ready until it can support evidence-grade hunting.

After scaffold, verify or create:
- `scope.yaml` placeholder with platform, program, in-scope and out-of-scope sections
- `policy.md` placeholder for required headers, rate limits, account rules, and prohibited actions
- `recon/`, `findings/`, `evidence/`, `poc/`, `reports/drafts/`, and `scans/`
- brain initialized or clear instruction to run `/brain init`
- next commands in order: `/sync`, `/pipeline`, `/analyze`, then `/hunt` or `/autopilot`

Warn if the program type needs special setup: mobile test device, API tokens, sandbox tenant, OAuth app, webhook receiver, cloud account, or smart-contract fork.

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (1)

Gates applied: no_behavioural_pass, no_license.