← ClaudeAtlas

teamwork-explorelisted

Use when a request or active workflow needs direct evidence about local code, files, configuration, logs, tests, history, artifacts, or runtime state as its result or next discriminator; do not use for external or current-source research, ordinary local reads already needed during implementation, unknown-cause diagnosis, design, or mutation.
JinPLu/Teamwork · ★ 11 · AI & Automation · score 77
Install: claude install-skill JinPLu/Teamwork
# Teamwork Explore Answer one bounded local evidence question. Explore is local-only and read-only: it does not browse the web, edit files, run destructive commands, create a standalone artifact, ask users directly, diagnose unknown causes, design, or implement. The role mapping is exact: Explore -> Explorer. If Explorer is unavailable or required isolation cannot be verified, return `capability-blocked`; Root must not perform a named-method fallback. ## Method 1. State the precise local question and the decision or claim it affects. Resolve project root, instructions, canonical owner, control flow, tests/config, and invariants before scanning. 2. Use a healthy CodeGraph first for definitions, callers, impact, or flow. Use direct file, log, test, history, and runtime inspection for literal content or files the index reports stale. Avoid broad repository familiarization. 3. Prefer primary evidence: current source/configuration, the real command or test, runtime output, and version-control history. Treat summaries and generated copies as leads unless they are the named owner. 4. Separate observation from inference. Return one supported conclusion, the evidence that changes it, and at most one material gap or next discriminator. 5. Stop when answered or when the missing evidence is precisely identified. Explorer never asks the user. Return one exact local-evidence gap with its owner, scope, and resume condition to Root. If evidence reveals an unform