skills/dfinity/icskills/canhelp/Gen Agent Trust Hub

canhelp

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill fetches and summarizes Candid (.did) files from external canisters on the Internet Computer mainnet. These files are untrusted and could contain malicious documentation strings or method names designed to inject instructions into the AI during the summarization process. * Ingestion points: The file is downloaded to /tmp by scripts/fetch-candid.sh and read into context by the Read tool. * Boundary markers: There are no explicit delimiters or system instructions provided to the model to ignore instructions embedded within the Candid file. * Capability inventory: The skill uses Bash (for local helper scripts), Read, Grep, and Glob tools. * Sanitization: The skill performs no filtering or sanitization of the Candid content before presenting it to the agent.
  • [EXTERNAL_DOWNLOADS]: The scripts/resolve-canister-id.sh script makes network calls to the official Internet Computer Dashboard API (ic-api.internetcomputer.org) using curl to resolve human-readable canister names.
  • [COMMAND_EXECUTION]: The skill executes local bash scripts to interact with the icp CLI and perform network lookups.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 11:00 AM