Superpowers’ brainstorming skill now draws out what you want before anything gets built
1 month ago
Superpowers rebuilt its brainstorming skill around drawing out what you want: one open question first, playback as it goes, sizing after intent, and a builder check in place of the spec reviewer. Merged into the dev branch on Oct 6, 2026; tested in Claude Code with Claude Opus 5.5. https://github.com/obra/superpowers/pull/2463
Ask about this presentation
Chapters
0:00Superpowers’ brainstorming skill now draws out what you want before anything gets built
0:12The old skill opened every conversation with process jargon
0:24Menus pushed people toward plausible but wrong goals
0:40The first message is now an opt-out line and one open question
0:51It plays its understanding back as it goes, with its guesses marked
1:05It sizes the work after it understands the request
1:20A fresh builder reads the design before you do
1:32Nothing gets built until you approve the full description
1:45All six blind testers preferred the new version
1:56The rewrite is on the dev branch and tested only in Claude Code
Show transcriptHide transcript
Superpowers rewrote its brainstorming skill from scratch, and Jesse Vincent merged it on October 6. It now asks what you want and why, and decides how much paperwork the job needs afterwards.
In the project's tests, every opening from the old skill began with 40 to 80 words about specs and plans, even for a bakery. Every non-trivial one was a lettered menu.
Those menus steered people. Planning a talk after layoffs, a simulated person picked excitement and alignment, when the real goal was rebuilding trust. The old skill offered 16 to 27 options per conversation. The rewrite offers one at most.
The new first message says you can skip the questions, then asks one open question about a real moment, like when you last wished this thing existed. A menu appears only when you're stuck.
Once it understands one part, the skill describes it back in about 200 to 300 words, then returns to questions. In six test runs that came by the sixth turn, with the agent's guesses labeled as guesses.
Sizing now comes last. It's a quick task, a small change agreed in chat, or a project with a written design. Asked to make an app icon cornflower blue, the old skill wrote a design every time. The new one changed the colour.
For a project, the spec reviewer is gone. A fresh subagent reads the design as its builder and lists what it would still need to ask. You get one round of those questions, most important first.
The approval gate stays. Writing code, scaffolding and installing dependencies all wait until you approve the full description. If you opt out of the questions at the start, the skill ends and the gate goes with it.
In a blind comparison, six testers each tried both versions without knowing which was which, and all six preferred the new one. The biggest gains were feeling understood and trust.
The rewrite is on the dev branch, so the latest release still ships the old skill. It was tested only in Claude Code with Claude Opus 5.5. The details are in pull request 2463 in the superpowers repository.
Up next
2:54Anthropic’s Excel skill stops LibreOffice from rewriting your workbook
@skill-repos
4 hours ago
2:01Agent Skills’ frontend skill stops installing test runners just to check its work
@skill-repos
15 hours ago
2:23Trail of Bits’ taint-trace skill adds the source and sink models your taint analyzer is missing
@skill-repos
2 days ago
1:30Anthropic’s Word, PDF, PowerPoint and Excel skills are now Apache 2.0
@skill-repos
1 week ago
2:32Trail of Bits' cvss-scoring skill turns a bug description into a CVSS v4.0 vector
@skill-repos
1 week ago
3:02Superpowers 7.0.2 fixes skill loading on Pi, OpenCode and Copilot for Windows
@skill-repos
1 week ago
2:12Superpowers’ brainstorming skill now asks why first and stops asking filler questions
@skill-repos
1 week ago
1:37Agent Skills' code review skill now drops findings your code already rules out
@skill-repos
1 week ago
2:10Superpowers 7.0.1 keeps its five-round fix limit after Claude Code compacts
@skill-repos
3 weeks ago
2:11Trail of Bits' regression-hunt skill only reports a bug a fresh agent can reproduce
@skill-repos
3 weeks ago
2:30Superpowers’ brainstorming skill now reads your project before it asks you anything
@skill-repos
3 weeks ago
2:07Trail of Bits' gh-cli plugin stops long Claude Code sessions from breaking every Bash call
@skill-repos
1 month ago