agentsclimarketplace

Teamwork explore

Skill JinPLu/Teamwork/plugins/teamwork-skill/skills/teamwork-explore

Evidence-first skills for Codex, Cursor, and Claude Code: research, debugging, planning, execution, review, and goal loops.

Install
npx -y skills add JinPLu/Teamwork --skill teamwork-explore

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 8 stars8 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

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.

SKILL.md

2.4 KB, as published. Nobody here has run it

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 unformed preference or material direction choice, return a reclassification signal to Collaborate; do not initiate that workflow.

Evidence belongs in the workflow packet or checkpoint that owns the decision. Writer never creates an independent Explore artifact. Native fast path remains outside Explore: tiny/discoverable local reads, ordinary explanations, simple commands, integration, and reads needed for an already authorized change stay with the active owner. If the remaining question is external/current, an unknown-cause failure, an unsettled decision, or mutation, report that boundary.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.