pixel-capi

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill contains several shell scripts (scripts/capi-send.sh, scripts/capi-test.sh, scripts/emq-check.sh, scripts/pixel-audit.sh, scripts/pixel-setup.sh) that use standard system utilities like curl, jq, and sha256sum to interact with Meta's developer APIs. These scripts are used for legitimate administrative and testing purposes.
  • [EXTERNAL_DOWNLOADS]: The skill communicates with well-known Meta domains, including graph.facebook.com, connect.facebook.net, and facebook.com. These interactions are documented as safe as they are necessary for the skill's primary function of managing marketing signals through an official service provider.
  • [DATA_EXFILTRATION]: While the skill transmits data to an external service, it follows best security practices by hashing all Personal Identifiable Information (PII)—including email addresses, phone numbers, and names—using SHA-256 locally before sending the data to the Meta Conversions API. Access tokens are managed via standard environment variables (META_TOKEN, ACCESS_TOKEN).
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local workspace files (workspace/brand/stack.md) to retrieve pixel and account identifiers. This ingestion surface is used strictly for technical configuration values that are inserted into API calls, posing no meaningful risk of prompt injection or behavioral override.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 09:11 AM
Security Audit — agent-trust-hub — pixel-capi