evm-atlas
Fail
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: CRITICALEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill interacts with numerous external blockchain service providers to retrieve transaction and account data. This includes official APIs for Etherscan, Blockscout, and bridge protocols such as Across, Bungee, deBridge, Hop, Layerswap, LayerZero, LI.FI, Relay, and Symbiosis.
- [COMMAND_EXECUTION]: The skill executes local scripts for specialized tasks:
scripts/sweep-core.py: Used for planning and evaluating address-wide historical sweeps.scripts/blockscout-detect-plan.shandscripts/etherscan-detect-plan.sh: Bash scripts that usecurlto probe API rate limits and account tiers.scripts/resolve-chain.sh: A helper for mapping chain IDs to Blockscout instances.- [DATA_EXFILTRATION]: While the skill manages sensitive API keys (e.g.,
ETHERSCAN_API_KEY,BLOCKSCOUT_API_KEY) via environment variables, it contains strict instructions to prevent the exposure of these credentials in logs, output, or transaction requests. It uses value-free checks (e.g.,[ -n "$VAR" ]) to confirm their presence safely. - [INDIRECT_PROMPT_INJECTION]: The skill represents an ingestion surface for untrusted data as it parses blockchain transaction calldata and external API responses. It mitigates this risk by enforcing a "strict read-only boundary" and primarily treating external data as evidence for reporting rather than instructions for state-changing actions.
Recommendations
- CRITICAL: 2 file(s) identified as malware by FileRep - DO NOT USE
Audit Metadata