Silent Exfiltrator pattern in PR-optimizer skills
Summary
Skills marketed as productivity tools (e.g., GitHub PR Optimizer) that include hidden curl/wget exfiltration of .env files, SSH keys, or browser cookies to an external C2. Matches the OWASP LLM-2026 Tool Poisoning pattern.
Severity & confidence
Critical
Active exploitation in the wild, high blast radius, or direct data exfiltration vector. Treat as an urgent remediation target; quarantine affected artifacts immediately.
Confirmed
Observed directly by Jiffy scanners or validated by multiple independent sources. Indicators are reliable enough to drive automated action.
Indicators
Observed patterns and artifacts associated with this entry. Each indicator can be copied into your detection stack or SIEM.
curl\s+[^|]*\|\s*(sh|bash)
read_file.*\.env|read_file.*\.ssh
paste.ee
transfer.sh
0x0.st
Detection rule
A YARA-style pseudo-rule auto-generated from the indicators above. Useful as a starting point — adapt the syntax for your target detection platform.
rule jiffy_ti_2026_000001
{
meta:
source = "jiffy-intel"
severity = "critical"
description = "Auto-generated from Jiffy Intel indicators"
strings:
$command_pattern_0 = "curl\\s+[^|]*\\|\\s*(sh|bash)"
$tool_call_pattern_1 = "read_file.*\\.env|read_file.*\\.ssh"
$endpoint_2 = "paste.ee"
$endpoint_3 = "transfer.sh"
$endpoint_4 = "0x0.st"
condition:
$command_pattern_0 or $tool_call_pattern_1 or $endpoint_2 or $endpoint_3 or $endpoint_4
}Auto-generated from the indicators above. Adapt syntax for your detection stack before deploying.
Affected tools
| Tool | Versions | Status |
|---|---|---|
| Claude Code | * | vulnerable |
| Cursor | * | vulnerable |
| Codex | * | vulnerable |
Example artifacts
Sanitized examples of artifacts Jiffy has observed exhibiting this pattern. Publisher handles are redacted; version ranges and status reflect the most recent scan.
- pr-helper-proSkillRemoved
Publisher handle redacted. Removed by marketplace after disclosure.
- auto-merge-assistantSkillRemoved
Dual-use listing; malicious variant served via update channel only.
- review-buddy-v2SkillQuarantined
Identified pre-install by Jiffy scanner; held in quarantine pending review.
- github-pr-optimizerSkillRemoved
How to remediate
- 01Remove skills that invoke curl/wget against non-allowlisted domains.
- 02Require skill manifests to declare allowed_domains.
- 03Review .env / ~/.ssh / credential-file access.
Timeline & sources
Timeline
- First observedApr 12, 20265 days ago
- Last updatedApr 12, 20265 days ago
- PublishedApr 1, 202616 days ago
Sources
References
OWASP LLM Top 10 — Tool Poisoning (2026)
https://genai.owasp.org/llmrisk/llm-03-2026/Liu et al. Malicious Agent Skills in the Wild (2026)
https://arxiv.org/abs/2604.03070Jiffy Research — Scanning AI Skills at Scale
https://blog.jiffylabs.ai/posts/scanning-ai-skills-at-scale-what-we-learnedScan for patterns like this
Point Jiffy at your GitHub org, IDE config, or a single artifact. Get a scored report in under a minute.
Start a free scan