Menu
Enforce agent isolation with code guards, not documentation. Prevent cross-contamination between agent contexts.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | zero-trust-agents |
| description | Enforce agent isolation with code guards, not documentation. Prevent cross-contamination between agent contexts. |
Enforce with code, not documentation. Agents and humans don't reliably follow written instructions.
SCITEX_OROCHI_DISABLE=true -- env var kill switchCLAUDE_AGENT_ROLE check -- role-based blocking (e.g., telegram must not load Orochi)TELEGRAM_BOT_TOKEN detection -- context-based blocking/tmp/, not shared workdirAccept: true, 1, yes, enable, enabled (case-insensitive)
Reject: false, 0, no, disable, disabled, empty, unset
exit(0) for intentional disable, exit(1) for error/forbiddendef _is_truthy(val: str | None) -> bool:
return (val or "").lower() in ("true", "1", "yes", "enable", "enabled")
# Layer 1: kill switch
if _is_truthy(os.environ.get("SCITEX_OROCHI_DISABLE")):
log.info("Skipping — SCITEX_OROCHI_DISABLE is set")
return []
# Layer 2: role-based
role = os.environ.get("CLAUDE_AGENT_ROLE", "")
if role.lower() == "telegram":
log.info("Skipping — telegram agent must not load this")
return []
exit(1) for intentional opt-out (use exit(0))Master index of all available skills for development workflows. Use when needing guidance on which skill to apply or when starting work on any project.
Master index of all available skills for development workflows. Use when needing guidance on which skill to apply or when starting work on any project.
Autonomous work mode — never ask user to run commands, never stand by, always take next action. Use when working independently on tasks.
Claude Code setup and configuration reference. Skills, commands, hooks, MCP servers, CLI, and permissions.
Code audit, cleanup, refactoring, API minimization, test review, examples, and project evaluation. Use when reviewing code quality or preparing for release.
Core development philosophy including development cycles, naming conventions, architectural decisions, and multi-agent workflows. Use when establishing project foundations or making architectural decisions.