Rohit Ghumare
Community@rohitg00 · UK
Engineer | DevRel | GDE | CNCF Ambassador | Docker Captain | AWS CommunityBuilder | Ex - Solo.io, Cerbos, Oracle, Reliance Jio
Agent Skills by Rohit Ghumare
Showing 124 vetted skills indexed across 5 GitHub repositories.
deep-dive
Conducts autonomous web research using DAG-based query planning and parallel subagents.
agentkit-seo
Routes broad SEO requests to the correct platform-specific optimization module.
claude-memory-kit
Persists Claude Code session context across projects using a hot cache and knowledge wiki.
manage-skills
Manages AI agent skill files across eleven coding tools via terminal commands.
parallel-worktrees
Creates and manages git worktrees for parallel coding sessions across branches.
domain-modeling
Build a shared-language glossary and bounded contexts before writing code.
insights
Computes session analytics, correction trends, and learning metrics from project memory and git history.
wrap-up
Audits session changes, runs quality checks, and captures learnings before ending a coding session.
sprint-status
Tracks parallel Claude Code sessions with session headers and structured status lines.
context-optimizer
Diagnose and reduce token usage through compaction, MCP audits, and subagent delegation.
mcp-audit
Audit connected MCP servers for token overhead, redundancy, and security risks.
auto-setup
Detects project type and configures quality gates, hooks, and settings for new codebases.
survey-generator
Generates structured literature surveys with real bibliographies into FTS5-indexed wiki pages.
wiki-research-loop
Auto-grows a wiki by running a budget-capped BFS research loop over pluggable source fetchers.
wiki-builder
Creates and indexes persistent markdown research wikis in a SQLite FTS5 knowledge base.
bug-capture
Convert conversational bug reports into durable GitHub issues written in domain language.
permission-tuner
Analyzes permission denial patterns and generates allow and deny rules for Claude Code settings.
skill-optimizer
Optimizes SKILL.md files by training on accumulated learn-rule corrections via a validation-gated loop.
safe-mode
Blocks destructive Bash commands and restricts file edits using PreToolUse hooks.
file-watcher
Configure Claude Code file watching hooks to react to config and dependency changes.
compact-guard
Preserves task state and critical files across Claude Code context compaction cycles.
writing-guidelines
Applies clear-writing standards to documentation, UI copy, and error messages.
llm-gate
Validates commits, commands, and file writes using LLM-powered prompt hooks.
context-engineering
Manages AI session context through writing, selecting, compressing, and isolating token budgets.