skills/amit-t/skills/session-handoff/Gen Agent Trust Hub

session-handoff

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard git commands (git rev-parse --show-toplevel, git status --porcelain) to determine the project root and the state of the working directory. These are routine operations for context-aware development tools.
  • [DATA_EXPOSURE]: The skill summarizes the current conversation history and local environment metadata. This summary is saved locally within the project at .claude/handoffs/. To protect against accidental exposure, the skill automatically ensures this directory is listed in .gitignore, preventing sensitive session data from being committed to a repository.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes the conversation transcript, which is an external input source.
  • Ingestion points: Processes the full conversation transcript in SKILL.md (Step 6) to extract goals and decisions.
  • Boundary markers: Employs a structured YAML/Markdown template (DOCUMENT-TEMPLATE.md) to isolate metadata from human-readable summaries.
  • Capability inventory: Includes file system write access and execution of git commands.
  • Sanitization: The instructions specifically mandate referencing external artifacts (PRDs, issues, logs) by path or URL rather than duplicating their content, which limits the potential for malicious payload replication.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 04:28 PM