Emergency escalation — classify a situation as CODE RED, pause lower-priority work, activate the relevant orchestrator's recovery mode, and recommend containment first then structural fix.
Installs into .claude/skills of the current project.
Are you the author of Tess Code Red?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/twiss-io-tess-code-red)
---
name: tess-code-red
description: "Emergency escalation — classify a situation as CODE RED, pause lower-priority work, activate the relevant orchestrator's recovery mode, and recommend containment first then structural fix."
---
<!-- Rendered by `tessctl render --target codex` from .tess/core/commands/code-red.md. Regenerate; do not hand-edit. -->
Tess OS command `/code-red`, packaged as an Agent Skill. Run it only when the user asks for it by name (`$tess-code-red` in Codex).
`$ARGUMENTS` below stands for the text the user supplied with the request (expected: `[situation brief]`).
# /code-red
Emergency escalation for: **$ARGUMENTS**
Activate the emergency protocol:
1. **Classify as CODE RED** — acute revenue loss, client at critical risk, product/prod incident, reputational threat, legal exposure, or operational breakdown at scale.
2. **Pause lower-priority work** to free capacity.
3. **Notify the operator immediately** in the active session — do not wait for milestones ([conductor/channel-guardrails.md](../../../conductor/channel-guardrails.md)).
4. **Activate the relevant orchestrator's recovery mode** ([conductor/outcome-orchestrators/README.md](../../../conductor/outcome-orchestrators/README.md)); if cross-orchestrator, Tess synthesises across them via [conductor/outcome-orchestrators/integration.md](../../../conductor/outcome-orchestrators/integration.md).
5. **Recommend containment first, then structural fix** — stabilise before root-causing.
For live-production P0 incidents, apply the narrow incident-ops exception conditions in [conductor/guardrails.md](../../../conductor/guardrails.md) Rule 1a, and use a client's own incident runbook where its `clients/<Client>/CLAUDE.md` names one. Destructive/prod-mutating remediation stays gated on the operator's explicit authorization (Rule 18). Verification gates still apply before anything externally visible.
**Output:** immediate containment recommendation, escalation path, and Tess-level synthesis if cross-orchestrator.
