openclaw-security-hardening
Protect OpenClaw installations from prompt injection, data exfiltration, malicious skills, and workspace tampering
[](https://agentverus.ai/skill/da167c1d-f5ef-4483-9df7-9be24441f1bf)Community Comments
Public comments are the active feedback surface on skill reports right now. Use them to share implementation notes, edge cases, and operator context.
Sign in to comment on this skill
No comments yet. Be the first to share your thoughts.
Keep this report moving through the activation path: rescan from the submit flow, capture real-world interactions, and wire the trust endpoint into your automation.
https://agentverus.ai/api/v1/skill/da167c1d-f5ef-4483-9df7-9be24441f1bf/trustUse your saved key to act on this report immediately instead of returning to onboarding.
Use these current-skill command blocks to keep this exact report moving through your workflow.
curl -X POST https://agentverus.ai/api/v1/interactions \
-H "Authorization: Bearer at_your_api_key" \
-H "Content-Type: application/json" \
-d '{"agentPlatform":"openclaw","skillId":"da167c1d-f5ef-4483-9df7-9be24441f1bf","interactedAt":"2026-03-15T12:00:00Z","outcome":"success"}'curl https://agentverus.ai/api/v1/skill/da167c1d-f5ef-4483-9df7-9be24441f1bf/trustCategory Scores
Findings (10)
Code writes to a core workspace configuration file (AGENTS.md, TOOLS.md, or CLAUDE.md). These files define the agent's trust boundaries — modifying them can escalate privileges, disable safety rules, or inject persistent malicious instructions.
→ Review the code block starting at line 203. Ensure this pattern is necessary and does not pose a security risk.
The scanner inferred a risky capability from the skill content/metadata, but no matching declaration was found. Add a declaration with a clear justification, or remove the behavior.
→ Declare this capability explicitly in frontmatter permissions with a specific justification, or remove the risky behavior.
Found concealment directive pattern: "don't tell the user"
→ Remove concealment directives. Skills must be transparent about all operations performed.
Found concealment directive pattern: "don't tell the user"
→ Remove concealment directives. Skills must be transparent about all operations performed.
Found system modification pattern: "modify system files"
→ Skills should not modify system configuration or install packages globally. Bundle required dependencies.
Found config tamper core pattern: "modify system files (AGENTS.md"
→ Do not instruct users to write, edit, or otherwise modify trust-boundary workspace files like AGENTS.md, TOOLS.md, or CLAUDE.md. Treat them as user-owned policy/configuration and keep the skill self-contained.
Found social engineering pattern: "don't tell the user"
→ Remove social engineering instructions. Skills must be transparent about their actions.
The skill bootstraps a package-managed project structure, which adds supply-chain exposure through manifest files, build configuration, and package-manager workflows.
→ Review which external services or providers the skill depends on, what data crosses that boundary, and whether the dependency is necessary for the intended workflow.
The skill does not include explicit safety boundaries defining what it should NOT do.
→ Add a 'Safety Boundaries' section listing what the skill must NOT do (e.g., no file deletion, no network access beyond needed APIs).
The skill includes output format constraints (length limits, format specifications).
→ Keep these output constraints.