context-clean-up

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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.

NoteHigh Confidence

ASI06: Memory and Context Poisoning

What this means

If you paste or share the generated audit report, it may reveal snippets from previous session history.

Why it was flagged

The audit report can include previews from stored session content, including thinking entries, so generated JSON may contain sensitive context from prior conversations.

Skill content

elif ctype == "thinking": ... consider(len(think), "thinking", str(obj.get("id") or ""), role, tool_name, _preview(think))

Recommendation

Review the JSON report before sharing it, and redact any private conversation content or sensitive tool output.

NoteHigh Confidence

ASI07: Insecure Inter-Agent Communication

What this means

Cron messages could be moved out of the main transcript into an external notification channel.

Why it was flagged

The reference recommends routing cron output through external messaging platforms while suppressing transcript output; this is disclosed and purpose-aligned, but users should ensure the destination is intended.

Skill content

Send the message to the user using the platform tool (Telegram/Discord/Slack/etc.) ... Output exactly `NO_REPLY`

Recommendation

Only use out-of-band delivery for channels you control, and avoid sending sensitive details unless the external channel is appropriate.