Back to skill
Skillv0.1.0
ClawScan security
Spark Store Skill · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
SuspiciousMar 9, 2026, 7:15 AM
- Verdict
- suspicious
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill's instructions match a package-manager helper, but the SKILL.md references local scripts and system tools that are not declared or included, which is an incoherence you should resolve before trusting it.
- Guidance
- This skill's goal (manage Linux apps via Spark Store / APM) is plausible, but the SKILL.md references local Python helper scripts (scripts/detect_os.py, spark_store_api.py, spark_apm_api.py) and system binaries (curl, aptss, apm) that are not declared in the registry metadata and are not included in the bundle. Before installing or enabling this skill: 1) ask the publisher to provide the missing scripts or an install spec so you can review code; 2) verify the external endpoints (d.spark-app.store) are trustworthy; 3) ensure aptss/apm/curl presence is declared and available on your systems; 4) be cautious about allowing autonomous invocation because the skill can run sudo package-manager commands — run it first in a safe/non-production environment and require explicit user confirmation for any privileged operation.
Review Dimensions
- Purpose & Capability
- concernThe skill's stated purpose (search/install/uninstall Linux apps via Spark Store/APM) is reasonable, but the registry metadata declares no required binaries or files while the SKILL.md explicitly depends on curl, aptss, apm and local Python modules under scripts/ (detect_os.py, spark_store_api.py, spark_apm_api.py). Requiring package manager binaries and helper scripts is expected for this purpose, but the manifest not listing them and the fact that those local scripts are not present in the file manifest is an inconsistency.
- Instruction Scope
- concernInstructions are narrowly scoped to searching remote JSON endpoints and running package commands (aptss/apm), which fits the purpose. However the runtime examples import and call local Python modules (scripts.*) that are not included in the skill bundle. That gap could lead an agent to (a) fail, (b) attempt to fetch or execute missing code, or (c) run shell commands directly. The SKILL.md also instructs running sudo aptss/apm commands — these require root and are powerful; the doc does not show safeguards (e.g., dry-run, explicit confirmation) beyond simple prompts.
- Install Mechanism
- okThere is no install spec and no code files are present, so nothing is written to disk by an install step — this lowers installer risk. That said, because SKILL.md references local scripts that are missing, the absence of an install step is itself a coherence problem (the skill appears to expect bundled code that isn't provided).
- Credentials
- okThe skill declares no environment variables or credentials, which matches the SKILL.md (no API keys required). It does require system-level package manager binaries and root to install/uninstall packages, which is appropriate for the stated task but should be explicitly declared in metadata.
- Persistence & Privilege
- notealways is false (normal). The skill can be invoked autonomously by the agent (default), which combined with instructions to run sudo package-manager commands increases risk because an autonomous agent could attempt privileged operations. The skill does not request persistent presence or modify other skills' configs.