Loading the registry…
For OpenMOSS Mechanistic Interpretability Team's Sparse Autoencoder (SAE) research. Open-sourced and constantly updated.
Cursor Rules are .cursorrules / .cursor/rules files that steer Cursor's AI; scanned with the same pipeline.
We found this by crawling public sources and nobody has attested to it. That is how most of the registry gets here, and it is not a mark against the artifact — it only means no authenticated act of publication is on record. An artifact can be the vendor's own and still appear here.
41 critical findings reported by cisco-skill-scanner, 16 of them in the installed artifact. The composite score alone does not reflect this — read the per-scanner reports below.
Cursor Rules are .cursorrules / .cursor/rules files that steer Cursor's AI. Review the scanned source before copying them into your own project.
nerlo install -- lm-saes
4 of 11 scanners examined this package. The rest do not apply to it — language, ecosystem and packaging decide which analyzers can say anything, and a scanner that cannot examine an artifact reports nothing rather than passing it.
v2.0.11
155 findings · 81.7s
v0.3.74
18 findings · 45.5s
v0.1.0
3 findings · 29.1s
v0.1.0
0 findings · 29.1s
v1.4.0
None of the entry points this scanner reads were found in this artifact. What each scanner covers
0 findings · 0.3s
vv0.3.2
No Go packages were found in this artifact. What each scanner covers
0 findings · 0.2s
v0.1.0
This scanner found nothing in this artifact that it assesses. What each scanner covers
0 findings · 2.0s
No report in this scan: govulncheck, osv-scanner, trivy, trivy_image.
1 scan on record. Every scan's full results are retained immutably for 24 months.
| Completed | Composite | Change | Scanners | Status |
|---|---|---|---|---|
| Aug 22, 2026 | Scan Halted | — | agentshield ·n/a: not applicablecisco-skill-scanner: completeagent-audit-kit: completenerlo-behavioral: completenerlo-install-instruction: completecapslock ·n/a: not applicablenerlo-multi-source ·n/a: not applicable | rejected |
Findings in files this artifact installs stand as reported; a model re-read the rest and said which ones it believes are false positives. This is a second opinion published beside the evidence, not a correction to it: the per-scanner reports above are unchanged, every dismissed finding is still listed there at its original severity, and the score and the Flagged badge are computed from those raw severities alone. Nothing below moved them.
Out of 176 findings across every scanner that examined this package. 25 of 41 findings rated critical by their scanner were among the dismissals.
Not reviewed is not the same as clean. Those 2 findings stand exactly as their scanners reported them, at full severity, as they would have without this review.
Two keys had to turn. The model had to name a reason from a fixed list of six, and a deterministic check of ours — over which files this artifact actually installs — had to independently agree. Anything it could not corroborate is in “not reviewed” above, at full severity.
30 Documentation or example — The match is in documentation or sample code, confirmed by our own check to be outside the installed artifact.
Rules: AAK-SECRET-004, TOOL_ABUSE_UNTRUSTED_PACKAGE_SOURCE, RESOURCE_ABUSE_INFINITE_LOOP and 5 more · Files: examples/test_janus_img_infer.py, examples/test_janus_img_gen.py, examples/test_janus.py and 15 more · reported by agent-audit-kit, cisco-skill-scanner
25 Test or fixture file — The match is in a test or fixture. Our own check confirmed the file is not part of what gets installed.
Rules: COMMAND_INJECTION_EVAL · Files: /repo/JanusCircuit/core/test_vq.py, /repo/JanusCircuit/test_full_pipeline.py, /repo/TransformerLens/tests/acceptance/test_hooked_encoder_decoder.py and 16 more · reported by cisco-skill-scanner
7 CI or editor configuration — The match is in build, CI or editor configuration, confirmed by our own check to be outside the installed artifact.
Rules: YARA_system_manipulation_generic, AAK-GHA-IMMUTABLE-001 · Files: /repo/TransformerLens/.devcontainer/Dockerfile, .github/workflows/publish.yml, .github/workflows/checks.yml · reported by cisco-skill-scanner, agent-audit-kit
These findings are in files this artifact installs and are not placeholder values. No false-positive basis can be corroborated for such a finding, so this review never dismisses one; they were not sent to the model and stand at their scanner's severity.
56 DATA_EXFIL_JS_NETWORK
Files: /repo/ui/public/circuits/util.js, /repo/ui/src/components/app/feature-preview.tsx, /repo/ui/src/components/circuits/static_js/d3.js and 10 more · reported by cisco-skill-scanner
14 COMMAND_INJECTION_EVAL
Files: /repo/JanusCircuit/core/janus_wrapper.py, /repo/JanusCircuit/sae_training/train_sae_on_language_model.py, /repo/JanusCircuit/transcoder_old/activation_collector.py and 7 more · reported by cisco-skill-scanner
7 HIDDEN_DATA_FILE
Files: /repo/TransformerLens/.gitconfig, /repo/ui/.eslintrc.cjs, /repo/.pre-commit-config.yaml and 1 more · reported by cisco-skill-scanner
5 MANIFEST_MISSING_LICENSE
Files: /repo/SKILL.md, /repo/TransformerLens/SKILL.md, /repo/ui/SKILL.md and 2 more · reported by cisco-skill-scanner
5 SOCIAL_ENG_VAGUE_DESCRIPTION
Files: /repo/SKILL.md, /repo/TransformerLens/SKILL.md, /repo/ui/SKILL.md and 2 more · reported by cisco-skill-scanner
2 The model did not answer for these findings.
Rules: AAK-TASKS-003 · Files: examples/test/data_processor.py, examples/train/data_types.py · reported by agent-audit-kit
Critical command injection and secret exfiltration risks identified.
The artifact contains critical command injection vulnerabilities in Python files such as `/repo/JanusCircuit/core/janus_wrapper.py` and `/repo/TransformerLens/transformer_lens/HookedTransformer.py`, and in JavaScript/TypeScript files like `/repo/ui/src/components/ui/multiple-selector.tsx`, allowing for shell execution. High-severity findings indicate potential secret exfiltration via network egress from `/repo/ui/src/components/model/model-card.tsx` and dynamic code execution in `/repo/TransformerLens/transformer_lens/head_detector.py`. Additionally, a resource abuse vulnerability, specifically an infinite loop, is present in `/repo/src/lm_saes/trainer.py`.
Concern level assessed by the review: Critical. This is the review's scale, not the registry badge.
Reviewed by gemini-2.5-flash. The model sees the findings, not your code, and cannot change a score.
3 AAK-TASKS-003
Files: JanusCircuit/data/sampler.py, JanusCircuit/data/check_preprocessed_data.py, JanusCircuit/data/data_type.py · reported by agent-audit-kit
3 EXCESSIVE_FILE_COUNT
Files: /repo/TransformerLens/., /repo/ui/., /repo/. · reported by cisco-skill-scanner
3 MANIFEST_INVALID_NAME
Files: /repo/JanusCircuit/sae_training/geom_median/SKILL.md, /repo/JanusCircuit/saes/geom_median/SKILL.md, /repo/TransformerLens/SKILL.md · reported by cisco-skill-scanner
2 BINARY_FILE_DETECTED
Files: /repo/ui/bun.lockb · reported by cisco-skill-scanner
2 COMMAND_INJECTION_JS_FUNCTION_CONSTRUCTOR
Files: /repo/ui/src/components/circuits/static_js/d3.js · reported by cisco-skill-scanner
2 LOW_ANALYZABILITY
Files: /repo/ui, /repo · reported by cisco-skill-scanner
2 opt.nerlo-rules.nerlo-js-secret-to-network-egress
Files: /repo/ui/src/components/model/model-card.tsx · reported by nerlo-behavioral
2 UNANALYZABLE_BINARY
Files: /repo/ui/bun.lockb · reported by cisco-skill-scanner
1 AAK-LEGAL-002
Files: ui/package.json · reported by agent-audit-kit
1 AAK-SUPPLY-004
Files: pyproject.toml · reported by agent-audit-kit
1 opt.nerlo-rules.nerlo-py-dynamic-exec
Files: /repo/TransformerLens/transformer_lens/head_detector.py · reported by nerlo-behavioral
1 RESOURCE_ABUSE_INFINITE_LOOP
Files: /repo/src/lm_saes/trainer.py · reported by cisco-skill-scanner
1 SECRET_PASSWORD_VAR
Files: /repo/JanusCircuit/data/understanding.py · reported by cisco-skill-scanner
1 TOOL_ABUSE_UNDECLARED_NETWORK
Files: /repo/repo/README.md · reported by cisco-skill-scanner