Auto Context

PassAudited by ClawScan on May 1, 2026.

Overview

This instruction-only skill is coherent and low-risk, but it will proactively read local project context files and its README includes unpinned/manual install examples.

Install this if you want the agent to automatically read project planning files before work. Keep secrets out of TODOs, roadmaps, handoffs, and session notes, review the context summary before changes, and use a trusted or pinned installation source.

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.

What this means

The agent may inspect project planning and handoff files automatically when starting or resuming work.

Why it was flagged

The skill directs autonomous local file reads before major actions. This is disclosed and central to the purpose, but users should know the agent may read project context without a separate prompt each time.

Skill content
Do not wait for the user to ask — proactively load context... Read TODO.md... Glob .cursor/handoffs/*.md
Recommendation

Use this skill only in projects where you are comfortable with the agent reading the listed context files; keep secrets out of those files.

What this means

Outdated or manipulated project notes could steer the agent toward the wrong task or implementation, though the skill does require a visible summary and staleness warnings.

Why it was flagged

Project documents are intentionally brought into the agent’s context and used to shape subsequent work. If those files are stale or contain misleading instructions, they could influence the agent’s behavior.

Skill content
After loading, produce a context summary... Current Task... Recent Changes... Relevant Findings...
Recommendation

Review the generated context summary before the agent makes changes, and treat project notes as data rather than unquestionable instructions.

What this means

Installing from a moving branch could fetch different content later than the reviewed artifact.

Why it was flagged

The README includes a user-run install command from an unpinned GitHub branch. This is an installation/provenance consideration rather than evidence of unsafe runtime behavior.

Skill content
npx add https://github.com/wpank/ai/tree/main/skills/meta/auto-context
Recommendation

Prefer the trusted registry install path or pin and review the source revision before manual installation.