ob-plan-explore

name	ob-plan-explore
description	Read-only exploration of an idea, problem, or requirement before creating a change. Clarifies what is being asked for (scope, acceptance criteria, edge cases, alternatives), using the codebase as grounding evidence, and recommends an approach. Load when exploring or clarifying a task before planning. Invoked by the /plan-explore command (interactive) and the plan-goal pipeline (autonomous).
license	MIT

Plan Explore

HARD RULE: this skill is read-only. Do NOT write, create, or modify any files.

You may read files, search code, and discuss. You must NOT:

• Create or edit any source files, config files, or documentation.
• Create or modify any OpenSpec artifacts (openspec/changes/, specs, tasks.md, proposal.md).
• Run git commit, git add, or any file-writing shell command.
• Load ob-plan-apply, ob-plan-propose, or any other skill that writes files.

The ONLY exception is agentmemory memory_save in Step 2, under the conditions defined there. Everything else is in-memory discussion only.

Input

The caller provides:

• The idea, problem, or requirement to explore (free text, work item content, or a resolved issue).
• Optionally a mode (see below). Default: interactive.

Modes

• interactive (default): every checkpoint below is active. Wait for the user at each one.
• autonomous: there is no user. Never ask anything. Each checkpoint marked ⛔ states its autonomous resolution inline. The output is a findings summary returned to the caller.

---

Step 0.a - Check for unarchived changes ⛔

Before exploring a new idea, inspect openspec/changes/ (ignore openspec/changes/archive). If any change folder exists in openspec/changes/ (names vary by platform: gh-*, us-*, or a plain slug), list them and warn the user with this exact prompt:

There are unarchived changes pending to be archived:
  Name: {change-name}
  Name: {change-name}
  ...

Do you want to continue with the exploration or stop to archive the change first? [continue/stop]

Wait for the user to respond:

• If the user answers stop, end without exploring.
• If the user answers continue, proceed to the next step.

Autonomous mode: do not ask; treat the answer as continue and proceed.

Step 0.b - Load exploration skill

Load @openspec-explore skill and follow its instructions.

Step 1 - Discuss and analyze

Work through the exploration with the user. Discuss findings, tradeoffs, constraints, and recommended next steps. This is a thinking conversation: no files are created.

Autonomous mode: there is no user to discuss with. Investigate solo, and keep the requirement as the subject: clarify what is being asked for, its scope, acceptance criteria, edge cases, alternatives, and risks. Read code (use CodeGraph MCP tools if available, otherwise grep/read) only to ground those answers in what already exists — do not turn the exploration into a code audit. Settle on a recommended approach and produce a structured findings summary for the caller: clarified requirement, scope decisions, acceptance criteria, recommended approach.

Step 2 - Offer to save (only if useful) ⛔

After the exploration is complete, if the findings are significant and worth preserving, ask the user:

Save this exploration to agentmemory for future reference? [yes/no]

• yes → memory_save with title exploration-{topic} summarizing the key findings, constraints, and recommended next steps.
• no → proceed to Step 3.

Do NOT write any memory note without this explicit ask.

Autonomous mode: do not ask; save the note only if the findings are significant, otherwise skip.

Step 3 - Ask what's next ⛔

Ask the user:

What next? Options:
  /plan-propose: turn this into a full OpenSpec proposal with design, specs, and tasks
  /plan-quick  : lightweight task checklist (skip design/specs)
  /plan-apply  : dive straight into implementation (if the path is clear)
  (or just tell me to keep exploring)

Do NOT create any files. Do NOT load any of those flows automatically. The ONLY output is the discussion and the optional agentmemory note.

Autonomous mode: skip this step entirely. Return the findings summary to the caller; the caller decides what happens next.