ralph-loop
Verdict safe, but 2 findings (2 medium), mainly unauthorized tool use, resource abuse. Most recent scan 2026-02-07.
MODERATE RISK - This skill generates bash scripts for automated AI coding loops. Primary concerns: (1) Encourages dangerous auto-approval flags without adequate security warnings, (2) Potential for unbounded loop execution and resource exhaustion, (3) Incomplete instructions that could lead to malformed script generation. The skill itself does not contain malicious code or data exfiltration, but the automation patterns it promotes (auto-approval + loops + code execution) create security risks if users don't understand the implications. The truncated instruction body is also concerning as it may generate incomplete scripts.
Scanner's overall assessment (LLM analyzer)
Second scanner: NVIDIA SkillSpector
Independent static run, no language model, over the same files as the Cisco scan above. How the two scanners compare across the whole corpus: Two scanners, one corpus.
What this skill does
Generate copy-paste bash scripts for Ralph Wiggum/AI agent loops (Codex, Claude Code, OpenCode, Goose). Use when asked for a “Ralph loop”, “Ralph Wiggum loop”, or an AI loop to plan/build code via PROMPT.md + AGENTS.md, SPECS, and IMPLEMENTATION_PLAN.md, including PLANNING vs BUILDING modes, backpressure, sandboxing, and completion conditions.
- Package
- 1 file, 5.4 KB
- SKILL.md
- 734 words, 11 sections
- Code
- None. Instructions only.
SKILL.md covers: Ralph Loop · Overview · Workflow · 1) Collect inputs (ask if missing) · 2) Phase 1 — Requirements → specs · 3) Phase 2/3 — PROMPT.md + AGENTS.md · 4) Two prompt templates (PLANNING vs BUILDING) · 5) Build the per‑iteration command · 6) Output a copy‑paste script · Safety/Sandbox Guidance (must mention) · Guardrails
Findings
MEDIUM Encourages Dangerous Auto-Approval Flags Without Adequate Warning Unauthorized Tool Usellm
The skill explicitly instructs users to enable dangerous auto-approval flags (--full-auto for Codex, --dangerously-skip-permissions for Claude Code) that bypass security confirmations. While it asks 'explicitly', there is no prominent warning about the security implications of disabling permission checks in automated loops that execute arbitrary code.
SKILL.md
Auto‑approve flags (ask explicitly)
- Codex: `--full-auto`
- Claude Code: `--dangerously-skip-permissions`MEDIUM Unbounded Loop Execution Risk Resource Abusellm
The skill generates bash scripts that run AI coding CLIs in loops with only optional completion conditions. While max iterations are mentioned, the implementation details are incomplete (instruction body cuts off). Without proper safeguards, loops could run indefinitely consuming resources, especially if completion conditions fail or are misconfigured.
SKILL.md
The loop persists context via `PROMPT.md` + `AGENTS.md` (loaded every iteration)... Max iterations... Build the per‑iteration command [instruction cuts off]What is in the package
| File | Kind | Size |
|---|---|---|
SKILL.md | doc | 5.4 KB |
Package contents recorded 2026-10-22. The scan above ran against the version current at scan time; the two can differ if the skill has since been updated.
What we checked
- Static analyzer. Pattern rules over SKILL.md and bundled files: known-bad commands, exfiltration URLs, secrets, obfuscation.
- Llm analyzer. A language model reads the skill and reports intent-level issues the rules cannot express.
- Meta analyzer. A second model pass that validates, deduplicates and re-scores the other analyzers' findings.
What we did not check
- Runtime behaviour. The skill was not executed; findings come from reading its files.
- Anything fetched at install or run time, including remote scripts, package dependencies, and URLs the skill tells the agent to open.
- Content the analyzers cannot read: images, binaries, and some non-English text.
- Known scanner bypass classes. See the methodology and limitations page.
Sources
- Repository: https://clawhub.ai/skills/ralph-loop
- Raw scan result: ralph-loop-scan.json
- Badge for your README:
- Request a re-scan