基于 SOC 职业分类
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/benjam3n/reasoningtool --skill gaca命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
Route any input through a branching question tree to narrow down the optimal response strategy before writing. Two stages — PERCEIVE (classify input) then ACT (select response). Covers all prompt types.
Generate exhaustive guesses about user input using ALL search methods with coverage tracking. Guessing is SEARCH through possibility space. Tracks space created vs space covered to ensure comprehensive exploration.
Systematically evaluate and select from a set of guesses, options, or possibilities. Combines ARAW analysis with prioritization to determine which guesses are strong, weak, actionable, or eliminable.
| name | gaca |
| description | Convert gates from yes/no vibe checks into explicit, checkable claim-interfaces. |
| output | {"format":"prose"} |
Input: $ARGUMENTS
Convert gates from "yes/no vibe checks" into explicit, checkable claim-interfaces. A gate that can be passed by saying "yes, looks good" is not a gate — it's a suggestion. Real gates require evidence.
For each gate in the target:
For each gate entry, classify what kind of answer it actually requires:
| Interface Type | Definition | Example | Checkable? |
|---|---|---|---|
| Definition | What does term X mean? | "Is the goal clear?" | Only if "clear" is defined |
| Observation | What do you see? | "Are tests passing?" | Yes — look at test results |
| Provenance | What source supports this? | "Is the claim verified?" | Only if verification source specified |
| Measurement | What metric and value? | "Is performance adequate?" | Only if metric and threshold defined |
| Decision | What choice was made and why? | "Was the approach selected?" | Only if selection criteria exist |
Flag gates that can be passed WITHOUT substantive content:
Vibe-check indicators:
| Indicator | Example | Problem |
|---|---|---|
| "Does the [document] exist?" | "Is there a risk assessment?" | Existence ≠ quality |
| "Was [process] done?" | "Was the design reviewed?" | Process ≠ outcome |
| "Is [quality] adequate?" | "Is the plan complete?" | No standard for "adequate" |
| Binary on continuous | "Yes/no on a spectrum question" | Loses nuance |
| Self-certification | "Do you believe...?" | Fox guarding henhouse |
For each vibe-check gate, rewrite as explicit claims:
Before: "Is the risk assessment complete?" After:
Gate: Risk Assessment Completeness
Claims:
1. "At least 5 risks are identified" — check: count entries
2. "Each risk has likelihood AND severity score" — check: inspect each entry
3. "Top 3 risks have mitigation plans" — check: verify plans exist with specific actions
4. "No category from the standard risk taxonomy is unrepresented" — check: compare against taxonomy
Each claim should be:
For each converted gate, define the minimum check that ensures content (not just format):
GATE: [name]
Pass requires ALL of:
1. [claim] — verified by: [how to check]
2. [claim] — verified by: [how to check]
3. [claim] — verified by: [how to check]
Evidence required: [what must exist, not just what must be claimed]
Verifier: [who checks — must be someone other than the creator]
The most dangerous gates: ones that can be satisfied by formatting alone.
| Gate | Can Be Gamed By | Fix |
|---|---|---|
| [gate] | [how to pass without substance] | [what would prevent gaming] |
GATE AS CLAIM AUDIT:
Gates audited: [N]
| Gate | Current Type | Vibe Check? | Claims After Conversion |
|------|-------------|------------|----------------------|
| [gate] | [interface type] | Y/N | [N claims] |
Vibe-check gates found: [N] of [total]
Passable-without-content: [N] — CRITICAL to fix
Conversion queue:
1. [gate] — [why priority] — [N claims needed]
Total claims produced: [N] (replacing [N] vague questions)