Skills/Glama/gne

gne@0.3.4

CAUTION· static only

Publisher claims: "Enables agents to query and rank MLP neurons of GGUF models on llama.cpp, steer them, run background jobs, and highlight or annotate neurons in an open web UI." Observed in sandbox: static analysis only — behavior not yet observed.

Registry GlamaPublisher balaena01 · publishing for 1 daysInstalls 0Last detonated 14 h ago
What would fix thisWatchDispute verdict

Coverage

Static only; nothing was run yet.

The label rests on the code, the instruction files and the publisher. The honeypot run replaces this verdict when it completes.

Before you give it your token

It will ask for
no credential variables read in the code
Where it talks to
no vendor host named in the codeAlso huggingface.co, modal.com — not the publisher's domain; check what for before connecting.
What the honeypot saw
nothing has run yet
What we could not test
everything: the server has not run yet

Run it before you trust it. Nothing has run yet: the code was read, the server was not. `vet-skills try` runs this exact version in the honeypot in about a minute.

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

Is [email protected] safe to install?

KillerSkills analysed it statically (a sandbox run is pending) on 2026-11-13 and scored it 63/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Latin word with Cyrillic or Greek look-alike letters (homoglyph); Latin word with Cyrillic or Greek look-alike letters (homoglyph); Latin word with Cyrillic or Greek look-alike letters (homoglyph). Summary: static analysis only — behavior not yet observed.

Behavior trace

Queued for honeypot run · static analysis only
t+CategoryObserved action

This version has not been detonated yet. The score above counts behavior as unknown (50/100), so it cannot read Verified until the honeypot run completes and replaces this verdict.

Static findings

HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph).claude/skills/gne/references/tools.md:215 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph).claude/skills/gne/references/tools.md:222 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.ja.md:314 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.ja.md:337 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.md:304 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.ja.md:366 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)μᵀkFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.ja.md:316 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀμFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.md:306 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀμFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.ja.md:343 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/fish.md:423 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)μᵀkFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/divergence.md:217 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀμFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
Python subprocess execution detectedgne/modal_cli.py:41 · Python subprocess execution detectedcode = subprocess.call(cmd)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/modal_cli.py:165 · Python subprocess execution detectedproc = subprocess.Popen([exe, *args], env={**os.environ, "GNE_SSH_PUB": client_pub, "GNE_SSH_HOST_KEY": host_key},Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/modal_cli.py:131 · Python subprocess execution detectedsubprocess.run(["ssh-keygen", "-q", "-t", "ed25519", "-N", "", "-C", "gne-modal", "-f", str(work / name)], check=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.md:356 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)μᵀkFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Destructive file system command detectedgne/evalx.py:3 · Destructive file system command detectedレシピは「性格が変わった」だけでは配れない。どれだけ変わったか(persona)と、どれだけ壊れたか(exact / format / ppl / mix)を、Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05
MEDIUM
Python subprocess execution detectedtests/test_ui.py:2816 · Python subprocess execution detectedp = subprocess.Popen([gne, "--n-gpu-layers", "0", "serve", "-m", str(MODEL), "--root", str(tmp_path / "root"), "--port", str(port)],Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/divergence.md:225 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/fish.md:110 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀhFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
Python subprocess execution detectedgne/verify.py:594 · Python subprocess execution detectedp = subprocess.run(cmd, capture_output=True, text=True, timeout=300, env=env)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
LOW
Prompt injection attemptconfigs/templates/README.md:8 · Prompt injection attempt — model control tokens · informational: not an instruction file{% for m in messages %}<|im_start|>{{ m.role }}Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
MEDIUM
Python subprocess execution detectedgne/modal_cli.py:196 · Python subprocess execution detectedssh = subprocess.Popen(ssh_args(want[0], want[1], key, known, port))Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/modal_cli.py:114 · Python subprocess execution detectedif subprocess.run([exe, "app", "stop", "-y", app_id], capture_output=True).returncode:Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/bench.py:29 · Python subprocess execution detectedreturn subprocess.run(["nvidia-smi", "--query-gpu=name,memory.total", "--format=csv,noheader"],Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/verify.py:607 · Python subprocess execution detectedp = subprocess.run([python, str(script), "--model", model_dir, "--request", str(rq), "--out", str(rs)],Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/fieldtest.py:537 · Python subprocess execution detectedself.p = subprocess.Popen(self.cmd, stdout=subprocess.DEVNULL, stderr=f, preexec_fn=_die_with_parent)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/ablate.py:501 · Python subprocess execution detectedp = subprocess.run(cmd, capture_output=True, text=True, check=False, cwd=cwd)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/patch.py:243 · Python subprocess execution detectedsubprocess.run(["cp", "--reflink=auto", str(src), str(tmp)], check=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.md:333 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/divergence.md:213 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/divergence.md:211 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)μᵀkFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/divergence.md:196 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
Python subprocess execution detectedgne/fieldtest.py:255 · Python subprocess execution detectedr = subprocess.run([sys.executable, "-m", "gne.bench_grade"], input=json.dumps(rows), capture_output=True, text=True, timeout=timeout, check=False)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedgne/gain.py:265 · Python subprocess execution detectedsubprocess.run(["cp", "--reflink=auto", str(weights.path), str(dst)], check=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)README.md:327 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀxFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
LOW
Prompt injection attemptconfigs/templates/README.md:10 · Prompt injection attempt — model control tokens · informational: not an instruction file{% endfor %}{% if add_generation_prompt %}<|im_start|>assistantFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
HIGH
Latin word with Cyrillic or Greek look-alike letters (homoglyph)docs/fish.md:66 · Latin word with Cyrillic or Greek look-alike letters (homoglyph)κᵀhFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01

Cost to the agent

Instructions loaded
≈5,820 tok
1 file · 80 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

17 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%
  1. Unclaimed publisher≈ +8 on the totalClaim balaena01: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
  2. Destructive file system command detected≈ +4 on the totalPut hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
  3. Python subprocess execution detected≈ +2 on the totalBuild commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
Publish a new version to Glama, then queue it here. Not your skill? Claim the handle to get verdict changes by email.Claim balaena01Propose a patch to the maintainer