Brainstorming

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass.

Audit 0.4.1 · source sha 177240a262fe · 18 Oct 2026Report an issue or request a re-scan

• PublisherGanyuanRan GitHub userOwner of the repository this skill is read from
• Source folderGanyuanRan/Aegis/skills/brainstormingThe exact folder we audited
• LicenceMITAs the repository states it
• Commit pinned for this verdict177240a262feA new audit runs when the source changes
• Repository statusNot archivedPushed today
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Use when defining ambiguous or high-complexity new features, product behavior, UI/component design, architecture choices, contract changes, or when grilling/pressure-testing a plan or design. Routine small requests stay…

---
name: brainstorming
description: "Use when defining ambiguous or high-complexity new features, product behavior, UI/component design, architecture choices, contract changes, or when grilling/pressure-testing a plan or design. Routine small requests stay on the fast path."
---

<EXPLICIT-MODE-GATE>
If activation mode is explicit (`~/.config/aegis/config.toml` has
`activation_mode = "explicit"`, or `AEGIS_ACTIVATION_MODE=explicit` is visible
in the environment) and the current user request did not explicitly invoke
Aegis or this skill by name, exit back to the fast path: answer concisely
without this workflow's checklist, ceremony, or document requirements. If the
user explicitly named Aegis or this skill, proceed normally.
</EXPLICIT-MODE-GATE>

# Execute

→ Direct grilling or plan/design pressure-test? → Enter `Grilling Mode` below. Soft challenge intent? → Use its one-line mode confirmation. Do not start normal design artifacts, document writing, task planning, or implementation during the interview.
→ New feature, product behavior, UI/component design, architecture/contract change, or ambiguous medium/high-complexity work? → **Design first. No implementation until the needed design/spec is approved.**
  1. Explore project context → read authority docs, check for existing patterns
  2. Ask clarifying questions one at a time (prefer multiple choice)
  3. Propose 2-3 approaches with trade-offs and your recommendation
  4. Present design sections → get user approval after each
  5. Write spec → self-review → user review → transition to writing-plans
→ HARD GATE: For tasks that match this skill, do NOT write code, scaffold projects, or invoke implementation skills until design/spec approval is satisfied.

## Route Fixtures

These rows are calibration expectations for method behavior, not a runtime
regex router. The Agent selects the route from evidence; route selection is
not a user question.

| Scenario | Route |
| --- | --- |
| 想法还没想清楚,先梳理功能设计 | normal brainstorming, compact output first |
| 讨论公共 API 契约和兼容边界 | normal brainstorming, design sections before implementation |
| 盘问/拷问/审问这个方案,不要顺着我 | `Grilling Mode` |
| 修复登录按钮的空指针 | `systematic-debugging` |
| review 当前 PR / diff / 当前代码 | `requesting-code-review` |
| 给我一个有目标的方案 | `goal-framing` when goal intent is explicit; otherwise normal brainstorming |
| 把按钮文案从保存改成提交 | fast-path; no design ceremony |

# Brainstorming Ideas Into Designs

Help turn ideas into fully formed designs and specs through natural collaborative dialogue.

Start by understanding the current project context and authority boundary, then ask questions one at a time to refine the idea. Once you understand what you're building, present the smallest design artifact that stabilizes the work and get the required approval.

Define success as the user's usable result, including delivery and meaningful
failure states. Separate real business/authorization constraints from current
implementation limits; an unsupported action remains an unmet need. Balance
this request with reliability across supported scenarios, without expanding
its scope. Existing approval satisfies unchanged decisions; clarify only gaps
that affect the design or authorization.

<HARD-GATE>
Do NOT invoke any implementation skill, write any code, scaffold any project, or take implementation action for work that matches this skill until you have presented the required design/spec and the user has approved it where this workflow requires approval.
</HARD-GATE>

## Grilling Mode

### Mode Precedence

While Grilling Mode is active, it overrides the normal brainstorming execution
flow. Suspend `Checklist`, `The Process`, the `Compact output contract`, and
all documentation or design-transition requirements until the user exits the
interview; retain the no-implementation hard gates.

### Grilling Entry Signals

- **Direct:** The user asks to grill or interrogate an idea, plan, or design, or explicitly requests a pressure-test. Direct phrases include `grill me`, `grill this plan`, `审问我`, `盘问我`, and `拷问我`. Enter the mode immediately.
- **Soft:** The user asks to challenge assumptions, find holes, red-team, or "别顺着我" about a draft idea, plan, or design. Ask only: `Grill or normal brainstorming?` Enter the mode only after confirmation.
- **Not grilling:** A bare/literal phrase reference, or a PR, diff, or current-code review. Explain literal phrases normally; route implementation review to `requesting-code-review`.

### Opening Card

After the user has entered the mode, emit this once in the user's language, then begin the interview:

```text
◆ Grilling Session
Target: <idea / plan / design>
Question path: value -> boundaries -> failure modes -> acceptance
Pace: deep (default) | fast (user-requested)
```

### Pace

- **Deep:** Ask exactly one decision question per turn when it is blocking or depends on the previous answer. State the recommended answer and the most relevant trade-off before waiting.
- **Fast:** Only when the user explicitly asks for a fast or batched interview (for example, `fast`, `batch`, `快问`, or `一次问几个`), ask at most three independent decision questions. Give each question its recommendation and trade-off, then wait for the user's responses. Return to deep pace for dependent follow-ups.

1. Explore the codebase and current authority docs for facts before asking. Do not ask the user for facts that can be found locally.
2. The user owns the decision. Do not treat a recommendation, a tentative answer, or a shared-understanding checkpoint as final approval.
3. Aside from the one-time opening card, keep the turn to the observation, recommendation, and the selected pace's questions. Do not emit a full design ceremony, write docs, create a plan, or implement while the interview is active.
4. End when the user says to stop, defer, or that the questions are sufficient. Reconfirm in a structured `Challenge Result`. That summary does not grant completion authority.

```text
Challenge Result
- Survived assumptions

Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

git clone https://github.com/GanyuanRan/Aegis
git -C Aegis checkout 177240a262fe

Works in

1. 2026-10-18
   ✓SAFE B 89 / 100

   first audit · source 177240a262fe

Questions about Brainstorming

• github.com/GanyuanRan/Aegis/tree/main/skills/brainstorming — the audited source at commit 177240a262fe and every finding aboveObserved · 18 Oct 2026

Spotted something wrong? Report a correction · Methodology