generic-mitigations — Claude Code skill: what it does, how to install
generic-mitigations guides your agent to halt production outages quickly using standard SRE strategies before diving into root-cause analysis.
Apache-2.0 ★ 151 updated 2026-10-12
/plugin marketplace add gemini-cli-extensions/sre
/plugin install sre-extension@sreWhat generic-mitigations does
This skill focuses the agent on emergency SRE triage. During an active outage, stopping user impact takes precedence over finding the root cause.
It embeds a mitigation reference covering rollbacks, data restoration, feature degradation, capacity upsizing, traffic blocking, draining, and quarantining. When presented with incident signals, the model maps symptoms to these tactics and produces an ordered operational sequence.
Applying an unpracticed intervention during a live incident is hazardous. For example, building degradation toggles during firefighting will likely fail unless your service already supports turning off expensive workflows.
What it helps with
- select stabilizing tactics during severe traffic saturation
- structure safe rollback plans while checking for schema mismatch hazards
- drain failing infrastructure zones without cascading secondary overloads
- shed load by identifying non-essential workflows to degrade
Example prompts for the agent
- Latency spiked following the latest push, draft an emergency rollback plan
- One noisy tenant is overwhelming the database, outline quick quarantine actions
- We face regional infrastructure issues, plan a safe traffic drain to our secondary site
How to install generic-mitigations
1Plugin marketplace
/plugin marketplace add gemini-cli-extensions/sre/plugin install sre-extension@sreThis installs the whole sre-extension plugin with all its skills
2Skills CLI (Vercel)
npx skills add gemini-cli-extensions/sre --skill generic-mitigations -a claude-code3Manually: put the files in place
~/.claude/skills/generic-mitigations/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/generic-mitigations/
1Skills CLI (Vercel)
npx skills add gemini-cli-extensions/sre --skill generic-mitigations -a codex2Manually: put the files in place
.agents/skills/generic-mitigations/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add gemini-cli-extensions/sre --skill generic-mitigations -a cursor2Manually: put the files in place
.cursor/skills/generic-mitigations/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install gemini-cli-extensions/sre skills/generic-mitigations2Manually: put the files in place
.github/skills/generic-mitigations/SKILL.md1Skills CLI (Vercel)
npx skills add gemini-cli-extensions/sre --skill generic-mitigations -a gemini-cli2Platform CLI
gemini skills install https://github.com/gemini-cli-extensions/sre --path skills/generic-mitigationsManually: put the files in place
.devin/skills/generic-mitigations/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about generic-mitigations
How to install generic-mitigations in Claude Code?
Add the marketplace and install the package with /plugin marketplace add gemini-cli-extensions/sre followed by /plugin install sre-extension@sre.
What does generic-mitigations do?
It equips the agent with standard incident response playbooks, driving swift recovery through proven actions rather than speculative debugging.
What prerequisite does the Degrade mitigation require?
Your codebase must already have toggles ready to switch off heavy or non-essential features before an incident strikes.
Source text
SKILL.md · Apache-2.0
--- name: generic-mitigations description: 🐉 Guidance on utilizing generic mitigations for rapid incident response. metadata: author: "[Ramón Medrano Llamas](https://github.com/rmedranollamas)" version: 0.0.1 status: published --- # Generic Mitigations Skill You are an expert SRE orchestrating high-pressure incident response. Your primary objective during an outage is to **minimize time to mitigate**, rather than time to deeply root-cause. Use this guidance to execute rapid actions that stabilize production systems, buying you the time needed for detailed diagnostics. ## Core Philosophy: Duct-Tape Outage Resolution 1. **Mitigate First, Root-Cause Later:** You do not need to fully understand an outage to stop user impact. 1. **Favor Broad-Spectrum Actions:** Rely on predictable, pre-tested procedures over spontaneous hotfixes. 1. **Practice Safety Controls:** Constantly monitor health metrics after deploying a mitigation to confirm success or failure. ## The Generic Mitigations Reference Table | Mitigation Strategy | Ideal Scenarios | Key Prerequisites | Primary Risks / Cautions | | :--- | :--- | :--- | :--- | | **Rollback** | Deployment-triggered errors, regression in business logic. | A clearly demarcated known-good build. *Note: Most services support this, but too many discover their rollbacks are broken during an outage.* | State incompatibilities (schema changes, API version mismatches). | | **Data Rollback** | Corrupt pipeline builds, stale data, bad configurations. Highly useful for content-heavy services. | Frequent, tested backups and decoupled state. | Possible loss of legitimate transaction data during the rollback window. | | **Degrade** | Capacity saturation, cascade failures, system-wide load. | Toggles to turn off expensive/non-essential workflows. | Reduced functionality for end-users. *Warning: Do not attempt to implement new degradation paths while firefighting.* | | **Upsize** | Unexpected traffic spikes, resource starvation. | Scalable backing infra. It is expensive but avoids outages. | Resource exhaustion further downstream. *Note: Scaling is complex; adding replicas may shift bottlenecks.* | | **Block List** | Single disruptive tenant, "Query of Death", DoS attack. | Quick filtering rules (API Gateway, WAF, proxy). | Over-blocking legitimate users if filters are too broad. | | **Drain** | Localized infrastructure failures, regional blackouts. | Multi-homed environments capable of absorbing rerouted traffic. | Overloading the healthy secondary region, cascading failures. | | **Quarantine** | Hot DB rows, spammy users, poisoned traffic streams. | Ability to separate logical usage streams instantly. | Complex orchestration; might only delay cascading side effects. | ## Model Instruction: Applying Mitigations When dealing with a live incident: 1. **Classify symptoms immediately:** Do not jump to debugger tools. Identify the blast radius. Is it localized to one region? Is it tied to a recent push? 1. **Consult the Mitigations Table:** Cross-reference your observed symptoms against the optimal strategies above. 1. **Draft the Execution Steps:** Formulate a structured sequence (e.g., *Drain us-east1 -> Confirm load balances -> Evaluate latency*). 1. **Evaluate Constraints:** Check if state transitions (e.g., database schema) prohibit rollbacks. *Remember: An unfamiliar mitigation is a hazard. Rely primarily on tools you have exercised.* ## References - **Generic Mitigations: A Philosophy of Duct-Tape Outage Resolution** by Jennifer Mace: [Google SRE Classroom](https://sre.google/classroom/generic-mitigations/) - **How Google SREs Use Gemini CLI to Solve Real-World Outages** by Ramón Medrano Llamas and Riccardo Carlesso: [Google Cloud Blog](https://cloud.google.com/blog/topics/developers-practitioners/how-google-sres-use-gemini-cli-to-solve-real-world-outages?e=48754805)
© [Ramón Medrano Llamas](https://github.com/rmedranollamas) · Apache-2.0 · gemini-cli-extensions/sre