explore · git:20261111.d9483be · 2026-11-11 · sha256 5a9e62da666e7924
explore git:20261111.d9483beA
Immutable. This exact content never changes and is served at /api/v1/blob/5a9e62da666e7924.
No tool to install first
claude plugin marketplace add https://markdownregistry.com/mp/a/art_mee3oodzxsorqmbg/5a9e62da666e7924e92269002d4c5200a7f360963742b6e64b5a93c24f24b2fe.json
Then /plugin install explore@markdownregistry-mee3oodzxsorqmbg in Claude Code. It installs this exact version, the one graded A here, pinned to the commit it was graded at, and leaves it there.
--- name: explore description: "Think an idea through before planning: read the repo, run and try things, weigh options, clarify choices; changes no repo file." user-invocable: true --- # /explore Start [the change loop](../../../../../wiki/development/the-change-loop.md): compare options and settle scope before `/plan`. Find related work with `openspec list --json` and `openspec context --json` ([CLI contract](../plan/references/openspec-cli.md#select-one-root)); read their pointers. **Change no repo file** ([limits](../../../../../wiki/development/the-change-loop.md#what-each-step-may-change)); [`/plan`](../plan/SKILL.md) logs answers. Explore precedes planning except [review-page notes](../plan/SKILL.md#review-notes); use [bounded mode](#when-plan-invokes-explore) under `/plan`. Standalone, run `node .claude/plugins/loop/scripts/slot.mjs stepTitle set EXPLORE --task "<the idea>"`. ## Investigate the relevant flow Read owning docs and source; trace trigger, result, and constraints. For non-code, read the process and authoritative sources. Separate observations from inferences. Run proportional, authorized checks: tests, scripts, builds; new code only in `.scratch/`. Report results before any plan; nothing outward. Name unavailable evidence and a later check. For consequential choices, compare viable approaches, outcomes, costs; recommend with evidence. Before handoff, check material assumptions and failures against the flow; revise contradictions. ## Questions during standalone exploration Ask [the shared way](references/asking-the-user.md): two or three related ready choices, one if alone, within tool capacity. Track decisions and prerequisites in chat. Investigate a missing fact yourself; never ask the person to guess it. Update dependencies after answers or evidence. Explain changed premises; reopen only affected decisions. Keep findings in chat, [drawn](../plan/references/drawings.md) as `text` when useful: chat shows no picture. **The 80/20 test:** ask where a wrong guess makes the plan *wrong*, not merely *different*: scope, observable behavior, compatibility, acceptance criteria. Assume naming, placement, wording, and cheaply changed details, with reasons. ## Search memory before asking Before asking, recall the task and pull its likely topic; read the originals cited: ```bash node "$(git rev-parse --show-toplevel)/.claude/skills/memory/scripts/memory.mjs" recall <question> node "$(git rev-parse --show-toplevel)/.claude/skills/memory/scripts/memory.mjs" pull <path-or-topic> ``` A task across topics: [the memory helper](../../../memory/references/recall-helper.md). Treat dated facts as assumptions. No `.claude/skills/memory/`, or [memory unavailable](../../../memory/SKILL.md#read)? Say so; continue with available sources. ## Check for other work For repo edits, check other work once: ```bash node "$(git rev-parse --show-toplevel)/.claude/plugins/loop/scripts/slot.mjs" otherWork ``` Compare intent, not files: plans can overlap before edits. - **Overlap:** name the work and why; add [the overlap ask](../plan/references/new-workspace.md#ask-once) to the next group, or ask alone. - **No overlap:** say nothing. - **A `notes` line:** say it in one line and continue. ## The exit round Resolve remaining material choices [the shared way](references/asking-the-user.md): - Apply [80/20](#questions-during-standalone-exploration) and tool capacity; ask only ready choices. - **Separate parts?** Ask about a new workspace for each ([how](../plan/references/new-workspace.md#ask-once)). - **Answers open another choice?** Follow up; stop once none remains. - A settled choice stays settled, nested calls included. - Keep an unanswered material preference open, never called an assumption or evidence gap. Keep authorized defaults and [the fallback when nobody can answer](references/asking-the-user.md#which-tool-carries-it). Summarize outcome, approach, reasons over a serious alternative, minor assumptions, and evidence limits. - **Bounded:** return by [the steps below](#when-plan-invokes-explore). - **Standalone**, typed by the person: [offer next steps](references/asking-the-user.md#end-every-reply-with-the-next-step); only *Plan it* starts `/plan`. ## When `/plan` invokes `/explore` Planning uses **bounded mode**, including via `/apply` or `/ship`: 1. Reuse the conversation's intent, answers, and findings. 2. Search memory and [other work](#check-for-other-work) unless checked; investigate only new material gaps. 3. Run [the exit round](#the-exit-round) for open choices; await dependent answers, retaining authorized/unattended defaults. 4. Return the summary to `/plan`.