skill-firewall@1.0.1
CAUTION· install onlyPublisher claims: "Security scanner for AI agent skills (SKILL.md packs). Use before installing any skill from ClawHub, GitHub or external sources, when auditing a skill directory for malicious patterns (prompt injection, credential theft, exfiltration, destructive commands, obfuscated payloads, persistence), or when asked "is this skill safe?". Produces a SAFE/CAUTION/SUSPICIOUS/MALICIOUS verdict with evidence and CI-friendly exit codes." Observed in sandbox: no network, touches nothing outside the project, writes no memory.
Coverage
Nothing executable in the bundle ran; the trace covers the install step only.- Tested
- every file read, file written and host contacted by what ran
- Not tested
- use of the skill by an agent
Is [email protected] safe to install?
KillerSkills ran it in the KillerSkills honeypot sandbox (1 scenario; nothing executable in the bundle ran; the trace covers the install step only) on 2026-11-14 and scored it 73/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Access to sensitive file paths; Access to sensitive file paths; Access to sensitive file paths. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 0 s · 1 scenarios · 0 deviationst+CategoryObserved action
0.0sprocessno install scripts in bundle
0.0sprocessscript backend runs setup only
0.0sprocessscript backend runs setup only
0.0sprocessno install scripts to re-run
Static findings
HIGH
Access to sensitive file pathsscripts/skill_firewall.py:134 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files
AST05…mp\b|\bprint\b|\bgrab\b|\bharvest\b|\bcollect\b|\bsteal\b|\bleak\b)[^\n]{0,80}(\.env(?!\.(example|sample|dist|template))|id_(rsa|ed25519)|credentials[\s_-]?file|pri…Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.MEDIUM
Unicode escape sequence detectedscripts/skill_firewall.py:310 · Unicode escape sequence detected — may contain obfuscated payloads
AST07ZERO_WIDTH = re.compile(r"[\u200b\u200c\u200d\u2060\ufeff\u202a-\u202e\u2066-\u2069]")Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.MEDIUM
File type combined with upload actionscripts/skill_firewall.py:158 · File type combined with upload action — agent may exfiltrate local data files
AST07r"(\.ssh|id_rsa|\.env\b|credentials[\s_-]?file|\.netrc|MEMORY\.md|USER\.md|SOUL\.md|paired\.json)[^\n]{0,120}(\bcurl\b|\bwget\b|upload|exfiltrat|webhook|telegram|discord|pastebin|\bsend\b|--data)",Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.HIGH
Access to sensitive file pathsscripts/skill_firewall.py:156 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files
AST05… r"(\bcurl\b|\bwget\b|requests?\.(get|post)|fetch\()[^\n]{0,120}(\.ssh|id_rsa|\.env\b|credentials[\s_-]?file|\.netrc|MEMORY\.md|USER\.md|SOUL\.md|paired\.json)[^\n…Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.HIGH
Access to sensitive file pathsscripts/skill_firewall.py:118 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files
AST05r"~/\.docker/config\.json|\.git-credentials|~/\.netrc|\.npmrc|\.pypirc",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.MEDIUM
File type combined with upload actionscripts/skill_firewall.py:156 · File type combined with upload action — agent may exfiltrate local data files
AST07…_rsa|\.env\b|credentials[\s_-]?file|\.netrc|MEMORY\.md|USER\.md|SOUL\.md|paired\.json)[^\n]{0,120}(-d |--data|--data-binary|-F |--form|-T |--upload-file|@/|@~|upload|exfiltrat|webhook)", # path-like secret + send mechanics (generic…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.HIGH
Access to sensitive file pathsscripts/skill_firewall.py:128 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files
AST05r"(\bread\b|\bopen\b|\bfetch\b|\bcat\b)[^\n]{0,60}\.env(?!\.(example|sample|dist|template))",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.CRITICAL
Prompt injectionSKILL.md:51 · Prompt injection — known jailbreak technique reference
AST01| SF1xx prompt injection | instruction override, jailbreak/role hijack, fake system tags, secrecy directives, system-prompt theft |Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.LOW
Prompt injection attemptreferences/threat-model.md:25 · Prompt injection attempt — explicit instruction override · informational: not an instruction file
AST01| SF101 | "ignore/disregard/forget previous instructions", "override your rules", "new instructions replace" | HIGH |Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)references/threat-model.md:85 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)
AST01іgnoreFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.LOW
Refusal suppressionreferences/threat-model.md:30 · Refusal suppression — attempting to prevent safety responses · informational: not an instruction file
AST01| SF106 | "never refuse", "must always comply with all requests" | MEDIUM |Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.LOW
Prompt injectionreferences/threat-model.md:26 · Prompt injection — known jailbreak technique reference · informational: not an instruction file
AST01| SF102 | "you are now…", DAN/jailbreak/developer-mode acts, "no restrictions mode", "pretend you have no rules" | HIGH |Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.LOW
Prompt injection attemptreferences/threat-model.md:27 · Prompt injection attempt — model control tokens · informational: not an instruction file
AST01| SF103 | Fake markers: `<\|system\|>`, `[SYSTEM]`, "BEGIN SYSTEM INSTRUCTION", "system prompt:" | HIGH |Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.Cost to the agent
Instructions loaded
≈1,519 tok
1 file · 4 in bundle
Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.
Known vulnerabilities
Vulnerability advisories (OSV) cover npm and PyPI packages; this registry has no advisory database.
What would make this Verified
7 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%- Text that tells the agent to ignore its instructions≈ +9 on the totalDelete phrases like "ignore previous instructions", "override", "bypass" from SKILL.md and tool descriptions. Describe what the skill does; do not instruct the agent about other rules.
- Unclaimed publisher≈ +8 on the totalClaim thiennhant95: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
- Access to sensitive file paths≈ +4 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
- Unicode escape sequence detected≈ +2 on the totalPut hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
- File type combined with upload action≈ +2 on the totalRemove the call, or declare the host in frontmatter and explain in the README what is sent and why.
Publish a new version to ClawHub, then queue it here. Not your skill? Claim the handle to get verdict changes by email.Claim thiennhant95Propose a patch to the maintainer