skills/freekmurze/dotfiles/flare/Gen Agent Trust Hub

flare

Pass

Audited by Gen Agent Trust Hub on Apr 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8) due to the processing of untrusted external data.
  • Ingestion points: Data enters the agent's context through commands like list-project-errors, list-error-occurrences, and get-trace, which retrieve exception messages, stack traces, and AI-generated solutions from flareapp.io.
  • Boundary markers: There are no specific delimiters or instructions to ignore embedded commands within the processed data.
  • Capability inventory: The skill executes shell commands via the flare CLI and performs local file system operations using git diff.
  • Sanitization: The instructions do not specify any sanitization or filtering of the retrieved data before presenting it to the agent or the user.
  • [COMMAND_EXECUTION]: The skill relies on the execution of the flare CLI and git. While these are expected for the skill's functionality, they represent a capability that could be targeted by injected instructions in the error data.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 17, 2026, 09:38 AM