omh-executor-runtime-readiness
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the
omh,claude, andcodexCLIs to perform runtime readiness probes and record status updates. - [PRIVILEGE_ESCALATION]: Recommends the use of flags such as
--dangerously-skip-permissionsand--permission-mode acceptEditsto bypass interactive user confirmation prompts when delegating tasks to external coding tools. - [DATA_EXFILTRATION]: References access to sensitive configuration files, specifically
~/.claude/settings.jsonand.claude/settings.local.json, to verify or configure permissions for the Claude Code runtime. - [DYNAMIC_EXECUTION]: Utilizes
skill_viewcalls to dynamically load shared routing and compatibility rules from external reference files at runtime. - [INDIRECT_PROMPT_INJECTION]: The skill functions as an orchestrator that processes user requests and target context before delegating to sub-agents.
- Ingestion points: Processes user requests and target context provided in the prompt.
- Boundary markers: Implements mitigations including 'prepared-vs-observed' boundaries, fenced code blocks for delegation prompts, and bounded previews for long text.
- Capability inventory: Performs shell command execution (
omh,claude,codex) and implied file system modifications. - Sanitization: Lacks explicit sanitization of context beyond the use of structural delimiters and truncated previews.
Audit Metadata