Randhir123

20 mods across 1 repository, 1 stars between them.

propose-fix

02

Randhir123/claude-ops-investigator

Command

Investigate an incident and, if it traces to application code, propose a fix as a draft PR — or apply the same fix-proposal step to a report that already exists.

1 1mo ago A 35 tokens

Randhir123/claude-ops-investigator

Skill Claude CodeCodex

Fallback single-agent Kubernetes incident investigation, for use only when custom modes are unavailable. Prefer the /investigate-incident command (.bob/commands/), which runs the same workflow with scoped, per-mode tool access instead of full Advanced-mode access.

1 1mo ago A 54 tokens

Randhir123/claude-ops-investigator

Agent Claude Code

Top-level coordinator for a Kubernetes incident investigation. Decomposes the incident, delegates to specialized subagents, aggregates their findings, and hands off to incident-reporter for the final report. Use this as the entry point for /investigate-incident.

1 1mo ago A 54 tokens

incident-reporter

05

Randhir123/claude-ops-investigator

Agent Claude Code

Use last, to synthesize evidence gathered by k8s-evidence-collector, prometheus-analyst, log-analyst, and runbook-analyst into a single structured, schema-valid incident report. Never gathers evidence itself.

1 1mo ago A 53 tokens

Randhir123/claude-ops-investigator

Agent Claude Code

Use for current-state Kubernetes evidence — pod listing, pod describe, live pod logs, namespace events, and current resource usage. Not for historical logs (use log-analyst) or metrics trends (use prometheus-analyst).

1 1mo ago A 55 tokens

log-analyst

07

Randhir123/claude-ops-investigator

Agent Claude Code

Use for historical log analysis across pod restarts and deployments via IBM Cloud Logs — error search, probe-failure search, and arbitrary text search. Preferred over live pod logs whenever the incident spans more than the currently-running pod. Not for metrics (use prometheus-analyst) or current pod state (use…

1 1mo ago A 75 tokens

prometheus-analyst

08

Randhir123/claude-ops-investigator

Agent Claude Code

Use for Prometheus metrics — restart counts, CPU/memory usage, HTTP error rate, and latency — to measure incident impact or corroborate a hypothesis with numbers. Not for log inspection (use log-analyst or k8s-evidence-collector).

1 1mo ago A 58 tokens

runbook-analyst

09

Randhir123/claude-ops-investigator

Agent Claude Code

Use to match a reported symptom against known incident runbooks and extract diagnosis steps, likely causes, and safety warnings. Not for live evidence gathering.

1 1mo ago A 35 tokens

PreToolUse

10

Randhir123/claude-ops-investigator

Hook Claude Code

Runs before the agent uses a tool for Bash tool calls, executing block_unsafe_shell.py via python3. From Randhir123/claude-ops-investigator.

1 1mo ago A tokens not measured

PostToolUse

11

Randhir123/claude-ops-investigator

Hook Claude Code

Runs after a tool call finishes for mcp__claude-ops-investigator__.* tool calls, executing audit_mcp_tool_call.py via python3. From Randhir123/claude-ops-investigator.

1 1mo ago A tokens not measured

SubagentStart

12

Randhir123/claude-ops-investigator

Hook Claude Code

Runs when a subagent starts, executing audit_subagent_lifecycle.py via python3. From Randhir123/claude-ops-investigator.

1 1mo ago A tokens not measured

SubagentStop

13

Randhir123/claude-ops-investigator

Hook Claude Code

Runs when a subagent finishes, executing audit_subagent_lifecycle.py via python3. From Randhir123/claude-ops-investigator.

1 1mo ago A tokens not measured

Stop

14

Randhir123/claude-ops-investigator

Hook Claude Code

Runs when the agent finishes a response, executing validate_final_report.py via python3. From Randhir123/claude-ops-investigator.

1 1mo ago A tokens not measured

settings

15

Randhir123/claude-ops-investigator

Settings file Claude Code

Agent settings declaring 5 hook events (PreToolUse, PostToolUse, SubagentStart, SubagentStop) and 7 allowed tools.

1 1mo ago A tokens not measured

Randhir123/claude-ops-investigator

MCP server Claude CodeCodexCursor

MCP server "claude-ops-investigator" as configured in Randhir123/claude-ops-investigator. Runs claude_ops.mcp.server with python. Needs 1 environment variable to run.

1 1mo ago A tokens not measured