speckit-agent-context-update

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local scripts located within the project's extension directory. Evidence: Uses .specify/extensions/agent-context/scripts/bash/update-agent-context.sh and .specify/extensions/agent-context/scripts/powershell/update-agent-context.ps1 to perform the context updates.
  • [PROMPT_INJECTION]: The skill manages content within files that define agent behavior (Indirect Prompt Injection surface). This is the primary intended function of the skill.
  • Ingestion points: Reads project configuration from .specify/extensions/agent-context/agent-context-config.yml and discovers content from the specs/ directory.
  • Boundary markers: Implements <!-- SPECKIT START --> and <!-- SPECKIT END --> delimiters to isolate the automated updates from the rest of the agent instructions.
  • Capability inventory: Executes local shell and PowerShell scripts with the ability to modify system-level instruction files like CLAUDE.md.
  • Sanitization: The skill explicitly validates that context file paths remain project-relative and rejects absolute paths, Windows drive paths, and path traversal segments (..).
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 09:09 PM
Security Audit — agent-trust-hub — speckit-agent-context-update