PromptIQ

中文 | English

Strict session reviews for power users collaborating with AI.

What It Does

Run /score at the end of a working session to get:

• a strict review across 8 collaboration dimensions
• a deterministic total with caps for weak evidence and low-complexity sessions
• a blunt Why It Is Not Higher section
• one concrete next-session drill

Run /draft before sending a prompt — give it your intent or a rough draft, get back a stronger, paste-ready version with an explanation of what changed.

Install

One-liner (any agent):

Read https://raw.githubusercontent.com/Explainix/promptiq/main/install.md and follow the instructions to install PromptIQ.

Claude Code

/plugin marketplace add Explainix/promptiq
/plugin install promptiq@explainix
/reload-plugins

Or from the terminal:

claude plugin install promptiq@explainix

Then type /score at the end of any session.

Codex

git clone https://github.com/Explainix/promptiq ~/.codex/skills/promptiq

Then type /score at the end of any session.

Usage

1. Work through a real session.
2. Run /score.
3. Read Why It Is Not Higher before looking at the total.
4. Apply the Next Session Drill next time.
5. Run /draft before your next prompt to write it stronger from the start.

Who It Is For

• Frequent Claude Code / Codex users who want sharper steering habits
• People who want calibrated feedback, not softer praise
• Anyone who suspects their prompts are vague but can't see where

Not for trivial one-liner sessions or generic personality tests.

How Scoring Works

The local engine handles N/A filtering, total calculation, score caps, confidence, and trend tracking. The model judges the session — the engine makes the result strict and stable.

Calibration rules:

• short, low-complexity, or low-confidence sessions are capped
• scores above 7.5 require evidence
• scores above 8.5 are intentionally rare

Example Output

• examples/sample-report.md — representative /score report
• examples/draft-sample.md — representative /draft result

Privacy

History is stored locally at ~/.promptiq/history.json. PromptIQ does not upload it.

Local Development

python3 -m unittest discover -s tests -v
python skills/score/scripts/promptiq.py doctor

License

MIT