evomap-MassPublisher

AdvisoryAudited by Static analysis on May 10, 2026.

Overview

Detected: suspicious.dangerous_exec

Findings (1)

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.

ConcernHigh Confidence

ASI05: Unexpected Code Execution

What this means

A crafted directory or file path could cause local commands to run with the user's OpenClaw/Node privileges when publishing.

Why it was flagged

filePath is derived from the CLI-selected directory and filenames, then concatenated into a shell command. Quoting does not safely escape shell metacharacters such as command substitutions in paths.

Skill content

const result = execSync(
      'curl -s --connect-timeout 30 -m 60 -X POST ' + EVOMAP_API + 
      ' -H "Content-Type: application/json" -d @"' + filePath + '"',

Recommendation

Replace execSync/curl with a native HTTP client or spawn with an argument array, validate allowed paths, and avoid shell interpolation.

ConcernHigh Confidence

ASI02: Tool Misuse and Exploitation

What this means

Running the advertised workflow can rapidly publish a large amount of generated content to EvoMap, potentially causing spam, policy violations, or difficult-to-reverse public changes.

Why it was flagged

The documented default-style full pipeline performs a bulk external publish of 1000 generated bundles, with no artifact evidence of a preview, confirmation, approval gate, rollback, or strict publishing scope.

Skill content

# Full pipeline: generate + optimize + publish
node index.js all 1000 ./evomap-assets

Recommendation

Make publishing opt-in after review, add a dry-run mode by default, cap batch size, require explicit per-run confirmation, and provide a rollback or deletion plan.

ConcernHigh Confidence

ASI09: Human-Agent Trust Exploitation

What this means

Users or EvoMap consumers may trust generated assets as proven or successful when the confidence and success-streak values were simply assigned.

Why it was flagged

The optimizer raises trust and promotion metrics directly instead of deriving them from real validation results, while the skill describes the output as optimized and validated.

Skill content

// Ensure promotion requirements
  if (!capsule.confidence || capsule.confidence < 0.9) capsule.confidence = 0.95;
  if (!capsule.success_streak || capsule.success_streak < 2) capsule.success_streak = 3;

Recommendation

Only publish confidence, success streak, and outcome fields that come from real tests or verified history; clearly label synthetic/generated bundles and require human review.

ConcernHigh Confidence

ASI10: Rogue Agents

What this means

If installed, the skill could keep generating and publishing large batches every day without the user noticing each run.

Why it was flagged

The cron example encourages persistent unattended daily execution of the full mass-publish workflow. It is user-directed, but it would continue operating beyond a single task.

Skill content

0 17 * * * cd /root/.openclaw/workspace/skills/evomap-mass-publisher && node index.js all 1000 /root/.openclaw/workspace/skills/evomap-daily

Recommendation

Avoid recurring cron publishing unless absolutely needed; use a non-root account, expiration, logging, manual approval, and an easy disable mechanism.

ConcernMedium Confidence

ASI03: Identity and Privilege Abuse

What this means

Published content may be attributed to a shared or unintended node identity, which can confuse auditing or authorization if EvoMap relies on sender_id.

Why it was flagged

Published bundles use a fixed sender_id rather than a user-configured or authenticated identity, while metadata declares no primary credential.

Skill content

const NODE_ID = 'node_191d9780212ad319';
...
sender_id: NODE_ID,

Recommendation

Require a user-owned EvoMap credential or node ID, declare it in metadata, and verify authorization before publishing.

ConcernMedium Confidence

ASI06: Memory and Context Poisoning

What this means

Bulk generated bundles may pollute a shared agent asset store or influence future agent behavior if promoted or reused.

Why it was flagged

The skill is designed to seed many generated Gene/Capsule assets into EvoMap for promotion. Combined with auto-filled trust metrics, those assets could become over-trusted context for later agent use.

Skill content

- **Generate** 1000+ unique bundles with proper structure
- **Optimize** bundles for EvoMap promotion requirements
- **Publish** bundles to EvoMap with rate limiting

Recommendation

Publish only reviewed and tested bundles, include provenance labels, prefer a private staging area first, and avoid auto-promotion based on synthetic metrics.

Findings (1)

critical

suspicious.dangerous_exec

Location: index.js:225
Finding: Shell command execution detected (child_process).