Anticipation
ReviewAudited by ClawScan on May 10, 2026.
Overview
This skill asks the agent to proactively inspect user history and local context and remember patterns, but does not define clear user consent, scope, or retention limits.
Install only if you want an assistant that proactively uses your context. Before using it, confirm what history, files, running commands, browser tabs, and memories it may access, and make sure there is an easy way to opt out or delete learned patterns.
Findings (3)
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.
The assistant may take initiative based on guesses rather than waiting for clear user direction.
The skill redirects the agent from responding to explicit user requests toward inferred needs and proactive observation, without stating that explicit user instructions or consent must take priority.
Erkenne was der User brauchen WIRD, nicht nur was er sagt. ... ❌ Warten auf Befehle ... ✅ Proaktiv beobachten
Require explicit user opt-in for proactive behavior and instruct the agent to ask before taking actions based on inferred intent.
Private file names, process arguments, or browsing context could be used by the agent without the user realizing it.
The skill instructs the agent to inspect local files, process state, and browser tabs at session start, but provides no limits, approval step, or scoping to a user-selected project.
Bei Session-Start ... Offene Dateien → Aktuelles Projekt? ... Running Commands → Hängende Prozesse? ... Browser-Tabs → Was sucht User?
Limit checks to user-selected context and require confirmation before accessing browser tabs, open files, or running-process information.
Sensitive history or incorrect learned assumptions could persist and shape future responses.
The skill suggests using conversation summaries and storing learned patterns, but does not define what is stored, how long it persists, whether the user can delete it, or how poisoned context is prevented from influencing future sessions.
Conversation Summaries → Wiederkehrende Themen? ... Bei Erfolg: Pattern merken (learn_pattern) ... Bei Fehler: Anpassen (learn_from_mistake)
Make memory opt-in, keep it scoped, provide deletion controls, and avoid treating stored summaries or learned patterns as authoritative without verification.