agent-exploration — Claude Code skill: what it does, how to install
agent-exploration — a skill for Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI and OpenCode that splits a broad research question into independent slices and dispatches each to a separate explorer.
no license 259 installs ★ 655 updated 2026-10-20
npx skills add pedronauck/skills --skill agent-exploration -a claude-codeWhat agent-exploration does
When a question spans too many independent areas for a single agent, it identifies those areas, dispatches one explorer per slice, and keeps each worker scoped to its own analysis/NN_analysis_<slug>.md file — no commits, no edits outside that path.
The --agents <num> flag sets an upper bound, not a quota. The skill picks as many slices as the scope justifies, up to 8. Output defaults to .audits/exploration/<timestamp>/; use --path <dir> to redirect. If one slice fails, the rest finish normally. The failed slice is noted in the summary rather than triggering a full restart.
Without --prompt <text> the skill reads the active request from context. --model <model> overrides the environment default. No external runtime is started, and no agent registry is created — only the native explorer mechanisms of the chosen platform are used.
What it helps with
- split a broad research task into independent parallel workstreams
- produce an evidence-backed summary with source links and remaining gaps
- keep each agent's output isolated so workers cannot overwrite each other
- recover from a failed slice without restarting the entire investigation
- redirect output path or override the model for a specific run
Example prompts for the agent
- Run agent-exploration --prompt "How is our monorepo architecture structured" --agents 4
- Analyse the dependency graph with agent-exploration --path .reports/deps --model o3
- Start a multi-area security review: agent-exploration --agents 6 --path .audits/security
How to install agent-exploration
1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill agent-exploration -a claude-code2Manually: put the files in place
~/.claude/skills/agent-exploration/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/agent-exploration/
1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill agent-exploration -a codex2Manually: put the files in place
.agents/skills/agent-exploration/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill agent-exploration -a cursor2Manually: put the files in place
.cursor/skills/agent-exploration/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install pedronauck/skills skills/mine/agent-exploration2Manually: put the files in place
.github/skills/agent-exploration/SKILL.md1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill agent-exploration -a gemini-cli2Platform CLI
gemini skills install https://github.com/pedronauck/skills --path skills/mine/agent-explorationManually: put the files in place
.devin/skills/agent-exploration/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add pedronauck/skills --skill agent-exploration -a opencodeClaude.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 agent-exploration
How do I install agent-exploration in Claude Code?
Run npx skills add pedronauck/skills --skill agent-exploration -a claude-code. The skill is placed at ~/.claude/skills/agent-exploration/SKILL.md.
What does agent-exploration do?
It breaks a research request into independent question slices, runs a separate explorer for each, and writes a final analysis/summary.md with answers, evidence links, disagreements, and gaps.
Does --agents guarantee that many agents will run?
No. --agents <num> is a ceiling. The skill chooses only as many slices as the scope warrants, so the actual count can be lower than the value you pass.