Skip to content

tobihagemann/turbo / claude/skills/resolve-open-items

resolve-open-items

Verified

Walk the open items listed above the prompt and settle each through AskUserQuestion. Use when the user asks to "resolve open items", "walk the open items", or "what's waiting on me".

Resolve Open Items

Settle each open item with one question per item, acting on every answer before the next round.

Step 1: List the Open Items

Call the mcp__turbo__open-items tool with action set to list. When it answers that there are no open items, present "No open items." Then use the TaskList tool and proceed to any remaining task.

Step 2: Settle Each Item

Use AskUserQuestion, one question per item in the order list returns them, filling each call to its four-question limit:

  • Header — #<id>
  • Question — the item's line, plus whatever its detail adds that the answer depends on
  • Options — at most four, by the kind of item. Give each option a description saying what choosing it does and what it gives up.
    • Decision, where the detail names choices. Fill the four options in this order and leave out whatever does not fit:

      1. The choice the detail recommends, with "(Recommended)" appended and the reason in its description. When no choice earns it with conviction, the most likely choice, unlabeled.
      2. Get a second opinion, when the decision is costly to reverse (it establishes a pattern others will follow, defines an interface, commits to a data shape, or imports a pattern the codebase has not used) or no choice earns "(Recommended)" with conviction. Leave it out when asking about an item again after a consultation.
      3. The other choices, most likely first.
      4. Not yet, then Note for later.

      End the question by naming each choice or deferral left out as available under "Other".

    • Action, where only the user can do it: Done, Not yet, and Note for later.

    • Open question, an item with no listed choices: the answers the conversation supports, then Not yet and Note for later, leaving out whatever does not fit and naming it in the question as available under "Other". When the conversation supports no answer, invite it under "Other".

After each call, act on its answers before the next call. An answer given under "Other" that names a choice, "not yet", or "note for later" takes that option's branch.

  • A choice — carry it out, then call the tool with action set to resolve and the item's id.
  • Done — the user already did it. Acknowledge it and resolve the item without repeating the action.
  • Note for later — run the /backlog skill with the item. Resolve the item only when that skill wrote the entry. When it could only print the entry, leave the item open and say so.
  • Not yet — leave the item open.
  • Get a second opinion — run the /consult-codex skill. On a technical choice, ask for the soundest choice on technical merit alone, independent of how closely it conforms to the original scope, and carry any scope cost back with the answer. On a question of product intent, ask what each choice commits to, what reversing it costs, and what the prevailing convention is. Then ask about that item again with the answer in hand.
  • An instruction or a question — act on it. When that settles the item or makes it moot, resolve it. Ask about it again only while it stays open.

Before the next call, resolve any unasked item that earlier answers settled or made moot, and leave it out of the call. Repeat until no item remains to ask. Then call the tool with list again, and settle any item it shows that this walk has not asked about, except an action this walk itself added, which stays open.

Step 3: Report

Output, by #<id>, what was settled and what stays open. Then use the TaskList tool and proceed to any remaining task.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall