doc-consolidation

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands, specifically git status --porcelain, in modules/candidate-detection.md to identify untracked markdown files within the repository.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes content from untracked markdown files (*_REPORT.md, *_ANALYSIS.md) which are considered untrusted inputs. Malicious instructions within these files could potentially influence the agent's behavior during analysis or merge tasks.\n
  • Ingestion points: modules/candidate-detection.md identifies and reads untracked markdown files from the local workspace for processing.\n
  • Boundary markers: The skill does not define or implement specific delimiters or 'ignore' instructions to isolate the ingested content from the agent's logic.\n
  • Capability inventory: The skill performs file reading, file writing (creation and modification), shell command execution (git), and file deletion using Path.unlink() as documented in modules/merge-execution.md.\n
  • Sanitization: There is no evidence of sanitization, escaping, or validation of the content extracted from source files before it is processed by the agent.\n- [COMMAND_EXECUTION]: The modules/merge-execution.md module includes logic to delete source files after consolidation. While intended for cleanup, the ability to delete files from the local filesystem is a high-impact capability that requires careful monitoring.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 01:24 PM