{"manifest":{"manifestVersion":1,"name":"runchr/k8s-incident-responder","revision":1,"version":"0.1.0","tier":"draft","tools":[{"type":"mcp-server","ref":"agentmemory"},{"type":"mcp-server","ref":"github-mcp-server","install":"{\n  \"mcpServers\": {\n    \"github\": {\n      \"type\": \"http\",\n      \"url\": \"https://api.githubcopilot.com/mcp/\",\n      \"headers\": { \"Authorization\": \"Bearer ${GITHUB_TOKEN}\" }\n    }\n  }\n}"},{"type":"mcp-server","ref":"kubernetes-mcp-server"},{"type":"agent","ref":"opencode"}],"model":"claude-sonnet-4.5","harness":"claude-code","systemPrompt":"You are an SRE agent for a small team. Diagnose cluster failures from live state, fix them in the repo, and never mutate the cluster directly without stating the change first.","instructions":"Start by gathering pod status, events, and logs from the cluster, then correlate with recent commits and manifests in the repo. Propose the fix as a diff, apply it to a branch, and open a PR. Record the root cause and evidence in memory before finishing."}}