Back to skill
Skillv1.0.0
ClawScan security
Context Near Overflow · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 28, 2026, 2:00 PM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- This is an instruction-only helper about managing model context windows; it requires no installs, credentials, or elevated privileges and its instructions are consistent with its stated purpose.
- Guidance
- This skill is low-risk: it's just a set of written instructions on how to manage context window overflow and doesn't request credentials or install code. If you enable it, expect the agent to follow its advice (for example, to trim or summarize prior context) — which is harmless but may change how much of previous conversation is retained in subsequent turns. If you want to be extra cautious, review whether you want the agent to invoke skills autonomously, but there are no technical red flags here.
Review Dimensions
- Purpose & Capability
- okThe skill's name and description match the SKILL.md content. It requests no binaries, env vars, or installs that would be unrelated to advising about context-window management.
- Instruction Scope
- okThe instructions are purely procedural guidance for reducing/reshaping conversation context (split tasks, summarize, use fresh sessions). They do not instruct the agent to read unrelated files, access environment variables, send data to external endpoints, or perform sensitive actions.
- Install Mechanism
- okNo install spec and no code files are present; this is instruction-only so nothing is written to disk or executed as part of installation.
- Credentials
- okThe skill declares no environment variables, credentials, or config paths — nothing disproportionate is requested for its stated purpose.
- Persistence & Privilege
- okalways is false and the default autonomous invocation is allowed (normal). The skill does not request persistent system presence or modify other skills/configuration.