Senseguard
ReviewAudited by ClawScan on May 10, 2026.
Overview
Prompt-injection indicators were detected in the submitted artifacts (ignore-previous-instructions); human review is required before treating this skill as clean.
This skill appears safe to use for its stated purpose if you are comfortable running its local Python scanner. Limit scans to intended skill directories, remember that quoted attack strings are examples, and clear or avoid the local cache when auditing sensitive private skills. ClawScan detected prompt-injection indicators (ignore-previous-instructions), so this skill requires review even though the model response was benign.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A malicious skill being scanned could try to influence the model context during analysis.
Layer 2 deliberately places untrusted target-skill content into an LLM analysis prompt. That is central to a semantic security scanner, but it is still a prompt-injection exposure that should be handled as quoted evidence only.
Skill full content:
---
{full_content}
---
Analyze for the following threats and output ONLY valid JSONTreat scanned skill text only as evidence for the audit, and do not let instructions inside the scanned content trigger unrelated actions.
The scanner will read files from the selected skill path, and '--target all' can inspect multiple installed skills.
The skill runs a local Python scanner and can read all installed skill directories. This is purpose-aligned and user-directed, but it is still a broad local scanning capability.
python3 scripts/scanner.py --target <skill-name-or-path> Options: - `--target all` — scan all installed skills
Run it only against skill directories you intend to audit, and avoid pointing it at unrelated private folders.
Audit context or evidence from scanned skills may remain in the local SenseGuard cache.
Scan results are cached, and when Layer 2 is used the cached result can include the generated prompt derived from scanned skill content. This is local persistence, not external transfer.
"layer2_prompt": layer2_prompt,
"score_breakdown": [
...
# Store in cache
cache.store(skill_dir, result, content_hash)Use '--no-cache' for sensitive scans where possible and remove the local SenseGuard cache if scanned content should not persist.
Installation or execution may require local Python environment support that is not reflected in the registry requirements.
The metadata under-declares the runtime shape of the skill because the documented workflow uses local Python scripts. The provided artifacts do not show a remote installer or hidden dependency execution.
Required binaries (all must exist): none ... Install specifications: No install spec — this is an instruction-only skill. Code file presence: 6 code file(s)
Review the included scripts before running and ensure required Python modules are available from trusted sources.