skills/agentscope-ai/hiclaw/file-sync/Gen Agent Trust Hub

file-sync

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/copaw-sync.py performs process replacement using os.execv. This is used as a bootstrapping mechanism to re-execute the sync logic within a specific virtual environment (/opt/venv/lite or /opt/venv/standard) where the copaw-worker package is installed.
  • [DATA_EXFILTRATION]: The scripts/push-shared.sh script facilitates the upload of local files and directories to a remote MinIO server using the mc (MinIO Client) tool. This is the intended purpose of the 'push' functionality for sharing task results.
  • [CREDENTIALS_UNSAFE]: The synchronization scripts access sensitive filesystem credentials and connection endpoints via environment variables (e.g., HICLAW_FS_ACCESS_KEY, HICLAW_FS_SECRET_KEY). This follows standard practices for managed service authentication.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it downloads and instructs the agent to read data from external sources.
  • Ingestion points: Files including configuration (openclaw.json), personality definitions (SOUL.md), and task specifications (spec.md) are pulled from a remote MinIO bucket.
  • Boundary markers: None present; the agent is directed to directly read the content of synced files.
  • Capability inventory: The skill can write to the filesystem, execute shell commands, and perform network operations via the MinIO client.
  • Sanitization: There is no verification of the integrity or content of the downloaded files before the agent processes them.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 08:12 AM