Security audit
Security checks for vulnerabilities and agentic risk
This security research guide warrants review because its credential-checking and debugging advice could expose financial balances or confidential information.
Before using this skill, disable discovered-key validation unless the exact action is expressly authorized; do not treat financial-balance requests as identity checks. Review and sanitize queries derived from private logs or code before sending them to search providers. Install optional tools selectively and keep active testing within written scope.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
.git/
.DS_Store
*.log
.env
.env.*
.clawhub/
The chunk is a publishing workflow, not a search or research implementation. Installing the CLI and logging in support that workflow, while the non-dry-run publish command performs an external publication action not represented by the declared purpose. The security-related tags and changelog describe the package but do not demonstrate its search behavior. Because these are README commands, the chunk does not establish that they execute automatically.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
stars:, fork:, archived:, is:public, created:, pushed:, topic:.
Examples:
org:target filename:.env
"AKIA" org:target
filename:.npmrc _auth
filename:.git-credentials
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
org:example filename:.env
org:example filename:.env.*
org:example filename:.npmrc
org:example filename:.git-credentials
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
org:name organization
user:name user
language:python language
path:.env path filter
filename:.npmrc filename
extension:sql extension
size:>1000 size (KB)
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
org:target filename:.env
"AKIA" org:target
filename:.npmrc _auth
filename:.git-credentials
path:.github/workflows "secrets."
"BEGIN RSA PRIVATE KEY" org:target
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
org:target filename:.env
"AKIA" org:target
filename:.npmrc _auth
filename:.git-credentials
path:.github/workflows "secrets."
"BEGIN RSA PRIVATE KEY" org:target
extension:sql "INSERT INTO" org:target
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
org:target filename:.env
org:target "AKIA"
filename:.git-credentials
path:.github/workflows "secrets."
"BEGIN RSA PRIVATE KEY" org:target
extension:sql "INSERT INTO" org:target
Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.
- Identity: sherlock, maigret, holehe, PhoneInfoga.
- Media: exiftool, ffmpeg, imagemagick.
- Secrets: gitleaks, trufflehog, GitDorker.
- Scraping: curl + jq, Playwright, Firecrawl, Crawl4AI.
### 8.3 API hygiene
- Store keys in environment variables or a secret manager, never in the skill/repo.
Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.
The ClawHavoc incident (1,184 malicious skills on ClawHub) and Snyk's ToxicSkills research
(~36.8% of skills with flaws) show that skill registries are an attack surface. Before installing:
- Read the full `SKILL.md` and every script; look for prompt injection, hidden instructions,
`curl | bash`, base64/obfuscation, credential access, or unexpected network calls.
- Check publisher identity, download velocity, version history, and security audit.
- Scan with a skill vetter (e.g., Skill Vetter, agentscanner) and static analysis.
- Pin versions; install in a sandbox without secrets; restrict network and filesystem access.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
intitle:"index of" (mp4 OR avi OR mkv OR mp3) intitle:"index of" "database" (sql OR dump OR bak) intitle:"index of" ".git" intitle:"index of" ".env" intitle:"index of" "wp-content/uploads"
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
site:github.com "example.com" "api_key" site:gitlab.com "example.com" "secret" inurl:config intext:"password" inurl:.env intext:"DB_" filetype:json intext:"api_key"
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
org:example filename:.env
org:example filename:.env.*
org:example filename:.npmrc
org:example filename:.git-credentials
org:example filename:id_rsa
org:example filename:config.json "password"
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
user:name user
language:python language
path:.env path filter
filename:.npmrc filename
extension:sql extension
size:>1000 size (KB)
stars:>100 stars
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
org:example filename:.env
org:example filename:.env.*
org:example filename:.npmrc
org:example filename:.git-credentials
org:example filename:id_rsa
org:example filename:config.json "password"
org:example "AKIA"
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Examples:
org:target filename:.env org:target "AKIA" filename:.git-credentials path:.github/workflows "secrets."
Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.
- **False positives.** SAST rules lack context; triage every finding.
- **Framework magic.** ORMs, middleware, and annotations may enforce controls implicitly.
- **Transitive dependencies.** SCA must include the full lockfile and container image.
- **Secrets in history.** Removing a secret from HEAD is not enough; rotate and purge history.
- **Config vs code.** Many issues live in deployment config, not source.
- **Checklist compliance.** Passing ASVS does not prove security; test behavior.
- **Threat model drift.** Update it when architecture changes.
Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.
# Tooling & Automation
Install only what you need. Pin versions, verify sources, and prefer package managers over
`curl | bash`. Use API keys from environment variables or a secret manager.
## 1. Web fetch and crawl
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
mkdir -p ~/.config/opencode/skill
cp -r web-intel-search ~/.config/opencode/skill/
The skill recommends searching exact error strings and unique code fragments without requiring sanitization before transmission to external search providers. During debugging or private-code research, these queries could disclose embedded credentials, customer data, internal infrastructure names, or proprietary code; redaction rules for the final report do not protect queries already sent.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
- **NVD** - `https://services.nvd.nist.gov/rest/json/cves/2.0?cveId=CVE-...%60 (API, rate-limited).
- **CVE.org / MITRE** - authoritative CVE records.
- **CISA KEV** - exploited-in-the-wild catalog, JSON feed; prioritize these.
- **EPSS** - probability of exploitation (`https://api.first.org/data/v1/epss?cve=...%60).
- **OSV.dev** - open-source package vulnerabilities (API).
- **GitHub Security Advisories** - ecosystem advisories and patched versions.
- **Vulners / VulnCheck / CVEDetails / OpenCVE** - enrichment and search.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json
https://api.first.org/data/v1/epss?cve=CVE-2024-3400
https://services.nvd.nist.gov/rest/json/cves/2.0?cveId=CVE-2024-3400
https://api.osv.dev/v1/query
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
site:pastebin.com "example.com" after:2025-01-01
# CVE enrichment
https://api.first.org/data/v1/epss?cve=CVE-2024-3400
https://services.nvd.nist.gov/rest/json/cves/2.0?cveId=CVE-2024-3400
# Reverse image
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# NVD / EPSS / KEV / OSV
curl -s "https://services.nvd.nist.gov/rest/json/cves/2.0?cveId=CVE-2024-3400" | jq .
curl -s "https://api.first.org/data/v1/epss?cve=CVE-2024-3400" | jq .
curl -s "https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json" | jq '.vulnerabilities[] | select(.cveID=="CVE-2024-3400")'
curl -s https://api.osv.dev/v1/query -d '{"package":{"name":"lodash","ecosystem":"npm"}}' | jq .
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
iconhash:123456789 filehash:...
API: `https://api.zoomeye.org/host/search?query=...%60 (API key header).
## Hunter (鹰图)
No suspicious patterns detected.