Deep-dive a topic using subagents. Feeds findings into decisions or working memory.
From gignpx claudepluginhub gregrossdev/gigThis skill uses the workspace's default tool permissions.
Designs and optimizes AI agent action spaces, tool definitions, observation formats, error recovery, and context for higher task completion rates.
Benchmarks web page Core Web Vitals/bundle sizes, API latency under load, build times; detects regressions via before/after PR comparisons.
Compares coding agents like Claude Code and Aider on custom YAML-defined codebase tasks using git worktrees, measuring pass rate, cost, time, and consistency.
If .gig/STATE.md exists, read it and display:
Version: {version} | Iteration: {iteration} | Status: {status}
Check if .gig/ exists in the current project root.
If NOT present: Note: "No gig context — research will proceed but findings won't be integrated." Continue to Step 2 (research works standalone).
If present:
Read .gig/STATE.md, .gig/DECISIONS.md, and .gig/ARCHITECTURE.md for context.
If the user provided a topic in their message, use it. Otherwise ask: "What do you want to research?"
Classify the research:
| Type | Approach |
|---|---|
| Library/framework | WebSearch + docs, compare alternatives |
| Codebase question | Subagents to explore code (Agent tool, subagent_type "Explore") |
| Architecture pattern | WebSearch + existing codebase analysis |
| API/service | WebSearch + docs, check compatibility |
| Bug investigation | Codebase exploration + error analysis |
Launch all research in parallel — agents and WebSearch calls in the same tool-call block:
| Research Type | Subagent Type | Profiles to Use |
|---|---|---|
| Library/framework comparison | Explore | Discovery |
| Codebase question | Explore | Architecture + Quality |
| Architecture pattern analysis | Explore | Architecture |
| API/service integration | Explore + WebSearch | Discovery + Architecture |
| Bug investigation | Explore | Quality + Architecture |
| Complex multi-step (combining search + synthesis + file creation) | general-purpose | As needed |
Explore — read-only codebase investigation. Use for all research that only needs to read and analyze. general-purpose — multi-step tasks that may need to write files, combine multiple searches, or perform complex synthesis. Use sparingly.
Collect all findings and synthesize before presenting.
Present a concise research report:
## Research: {topic}
### Findings
{Key discoveries, organized by sub-topic}
### Recommendations
{Opinionated suggestions based on findings}
### Impact on Current Work
{How this affects active decisions/plan, if any}
### Sources
{Links, file paths, or references}
Skip if no gig context (Step 1 noted absence).
If .gig/STATE.md exists and has an active iteration: