/cheat-score — Single Draft Scoring

Score but no prediction. Users use it to quickly check the draft's composite score and decide whether it's worth entering the formal prediction process.

Overview

[User: Score this draft.md]
  ↓
[Read draft.md + rubric_notes.md]
  ↓
[Score each dimension 0-5 + write one-line reason + calculate composite]
  ↓
[Console output: Scores + composite + recommended next steps]
  ↓
[End — no files written]

Constants

RUBRIC_PATH = rubric_notes.md — Current rubric source
OUTPUT_DETAIL = full — full: includes per-dimension reasons; compact: score table only

💡 Override when calling:
/cheat-score draft.md — OUTPUT_DETAIL: compact

Inputs

Required	Source
`<draft-path>`	Passed as parameter by user; if missing, ask in conversation
`rubric_notes.md`	User project root
`.cheat-state.json`	User project root (used to read current `rubric_version` and mode)

Workflow

Step 1: Pre-check

Read
```
.cheat-state.json
```
→ if not exists, prompt user to run
```
/cheat-init
```
first, then stop
Read
```
<draft-path>
```
→ if not exists or empty → throw error and stop
Read
```
rubric_notes.md
```
to find the currently effective formula section (usually in "Current Scoring Dimensions" or "Composite Score Formula" section)

Step 2: Identify Formula & Dimensions

Parse from

rubric_notes.md

Current rubric_version

Dimension list and weights (e.g.,

ER×1.5 + SR×1.5 + HP×1.5 + QL + NA + AB + SAT

)

Normalization constant (e.g.,
```
/ 8.5 × 2.0
```
)
Meaning of 0-5 scores for each dimension (read from table in "Current Scoring Dimensions" section)

If the format of

rubric_notes.md

does not match expectations (user manually modified the structure) → ask the user which line the current formula is on, do not guess on your own.

Step 3: Delegate to blind sub-agent (no longer inline scoring)

The main conversation has been contaminated by user conversations / sent data / historical retro sections — inline scoring is equivalent to judging with hindsight.

Change to call the
/cheat-score-blind
sub-agent via Task tool, main Claude only handles scheduling + review. See skills/cheat-score-blind/SKILL.md.

Task prompt template (must only contain the following):

Spawn cheat-score-blind sub-agent.

Input:
  script_path: <user-provided draft path>
  rubric_notes_path: rubric_notes.md

Task: Score the above script according to the current formula in rubric_notes. Return strict JSON (see cheat-score-blind SKILL.md Phase 2 schema).
Do not read state file / predictions/ / videos/ any other files.
Do not ask the user — you have no user.

Prohibited content in Task prompt (from "Main Claude Calling Contract" section of cheat-score-blind/SKILL.md):

User conversation references / excerpts
Words like views count / 万 / w / k
Hints like "Previous prediction was X" / "Actual views were Y"
Any
```
predictions/*.md
```
paths

Self-check before calling:

echo "<prompt>" | grep -Ei '播放|阅读|点赞|评论数|实际|retro|复盘|实绩|w$|万$'

→ if hit → revise prompt and resend.

Step 4: Parse sub-agent's returned JSON + review

Sub-agent returns strict JSON. Main Claude:

Parse the dimensions section (includes score + per-dim confidence + reason)

Verify

self_check.any_contamination_signal == false

, otherwise issue a warning

Calculate composite score according to rubric_notes formula (formula logic in main agent, scores from sub-agent)
Do not modify the dimension scores given by the sub-agent — scores are for display only. If the user raises an objection (e.g., "AB should be 4 instead of 3"), main Claude records it to
```
User Override
```
but keeps the sub-agent's original score on file.

If sub-agent returns

refusal != null

```
blocked_contaminated_input
```
→ report that Task prompt contains prohibited fields, ask main Claude to resend
```
script_path_invalid
```
→ check the path
```
rubric_unparseable
```
→ prompt user that rubric_notes.md is damaged
```
non_blind_warning
```
→ still accept dimensions (but all confidence levels are medium), issue a warning

Step 5: Calculate composite + output

Calculate composite score according to current formula. Console output (OUTPUT_DETAIL=full):

📊 [draft.md short title] — Scoring (rubric: v2)

| Dimension | Score | Reason |
|---|---|---|
| ER (Emotional Resonance) | 5 | Extremely concrete: "Checking chat records at 3 AM" |
| HP (Hook Strength) | 5 | IS sentence locks audience in one line |
| QL (Quotable Line Density) | 5 | MVP line "Intermittent hope" is independently shareable |
| NA (Narrative Quality) | 3 | Straightforward, weak arc |
| AB (Audience Breadth) | 5 | Universal for unrequited love/ex-partner themes |
| SR (Social Issue Resonance) | 2 | Pure personal emotion, no social context |
| SAT (Satire Depth) | 4 | Self-referential irony in acknowledgment section |

Formula: (ER×1.5 + SR×1.5 + HP×1.5 + QL + NA + AB + SAT) / 8.5 × 2.0
composite = (5×1.5 + 2×1.5 + 5×1.5 + 5 + 3 + 5 + 4) / 8.5 × 2.0 = **8.24**

📍 Falls into 300k-1M bucket (based on starter-rubrics bucket boundaries)

Next steps:
- If you've finalized the draft and are ready to publish → say "Initiate prediction"
- If you want to revise the draft → score again after revision (multiple scorings leave no trace)
- If you want to see samples with similar historical composite scores → say "Find anchors with composite 8.0-8.5"

When OUTPUT_DETAIL=compact, only output the score table + composite, no reason column.

Step 6: Never do these things

❌ Write any files (including predictions/, rubric_notes.md, candidates.md)
❌ Provide bucket probability distribution (that's cheat-predict's job)
❌ Trigger "published" or "retro" logic
❌ Propose rubric upgrades (even if obvious anomalies are found during scoring, only prompt in console, do not modify rubric)

Key Rules

Scoring goes through sub-agent. Main Claude no longer does inline scoring. See the isolation protocol in cheat-score-blind/SKILL.md
Integer scores only. 4.5, 3.7 are not allowed
Blind scoring first. Sub-agent only looks at script + rubric, naturally blind — that's the entire reason for its existence
Reasons are diagnostic tools. The 1-30 word reason for each dimension is not decorative — used during retro to identify which dimension's judgment was wrong
No file writing. This is the core difference between score and predict. Score is exploration, predict is commitment
Do not calculate candidate composite. The composite field in candidates.md is written in cheat-trends/cheat-recommend — score only serves "specific drafted manuscripts"

Refusals

"Score and predict at the same time" → Refuse. Please use
```
/cheat-predict
```
instead. Reason: predict must go through blind check + write immutable logs, score skips these steps
"Write the score into the observation section of rubric_notes.md after scoring" → Refuse. Observation lifecycle requires that observations must have a "actual performance vs prediction" comparison; scoring alone does not constitute an observation
"Can you just tell me if it will go viral?" → Refuse. Determining specific composite + bucket requires going through the predict process; score only outputs mechanical calculation under the current rubric
"Skip the blind sub-agent and let main Claude score directly" → cheat-score does not accept this escape hatch (unlike cheat-predict; cheat-predict has
```
--skip-blind
```
). Score is lightweight exploration, no reason to abandon isolation. If Task tool is truly unavailable → prompt user to configure it and try again

Integration

It's a pre-exploration step for
```
cheat-predict
```
: users can repeatedly score different draft versions, then predict once they've decided on a version
Score does not update
```
.cheat-state.json
```
— this is a side-effect-free operation
If user scores the same draft ≥3 times consecutively → gently prompt in console: "Repeated scoring may introduce decision fatigue, it's time to make a decision"

cheat-score

NPX Install

Tags

SKILL.md Content (Chinese)