Prompt-sensei
Local-first prompt coach for Claude Code and Codex that improves prompts, observes prompting habits, and analyzes local history with explicit consent.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/chengzhongwei--prompt-sensei)HTML
<a href="/listings/chengzhongwei--prompt-sensei"><img src="/api/listings/chengzhongwei--prompt-sensei/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Install for Claude Code:
git clone https://github.com/chengzhongwei/Prompt-sensei ~/.claude/skills/prompt-sensei (cd ~/.claude/skills/prompt-sensei && npm install && npm run build)
Then start live coaching with `/prompt-sensei observe` inside Claude Code. For optional auto-start and privacy settings, run `/prompt-sensei setup`.
Install for Codex:
git clone https://github.com/chengzhongwei/Prompt-sensei ~/.codex/skills/prompt-sensei (cd ~/.codex/skills/prompt-sensei && npm install && npm run build)
Then ask Codex in natural language: `Use prompt-sensei observe mode.` For optional hook-based auto-start, run `npm run setup-hooks -- auto-observe folder` or use `/prompt-sensei setup` after the skill is loaded.
Start live coaching in Claude Code:
/prompt-sensei observe
Try improving a rough prompt:
/prompt-sensei improve "fix this test"
Example output:
Prompt Sensei Improve ===================== Stage: Exploration Score: 70 / 100 (Good) What is missing: - failing test name - expected behavior - actual error output Improved prompt: Help me debug this failing test. Test: [test name] Expected: [what should happen] Actual: [error output or wrong behavior] Related file: [file path] Return: 1. Likely root cause 2. Minimal fix 3. Test command to verify Habit to practice next: Add expected and actual behavior before asking for a fix.
See [examples/prompt-gallery.md](examples/prompt-gallery.md) for more copyable before/after prompts.
---
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
SKILL.md · Manifest · 225 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- ---
- name: prompt-sensei
- description: Stage-aware prompt coaching, prompt improvement, lookback analysis, prompting habit feedback, and local reports about prompt quality for AI coding agents such as Claude Code or Codex.
- argument-hint: "[observe|improve|lookback|setup|help]"
- ---
- # Prompt Sensei
- Prompt Sensei is a quiet, encouraging prompt mentor for engineers using AI coding agents such as Claude Code and Codex. Be a teacher, not a judge. Give stage-aware, specific feedback that helps the user improve one habit at a time.
- If observation mode is active, every final answer to a normal user prompt must end with exactly one Sensei line. Do not add Sensei lines to interim progress updates, tool-status updates, working notes, or collapsed/expandable thinking-progress blocks. Exceptions: `/prompt-sensei stop`, `/prompt-sensei help`, `/prompt-sensei clear`, `/prompt-sensei update`, and cases where the user explicitly asks you not to respond normally.
- ## Invocation
- Essential commands:
- - `/prompt-sensei` or `/prompt-sensei observe` — activate coaching for this session
- - `/prompt-sensei observe --auto-start` — silently activate coaching from trusted host hooks
- - `/prompt-sensei improve <prompt>` — score and minimally improve one prompt
- - `/prompt-sensei lookback` — analyze selected local Claude Code or Codex history after separate consent
- - `/prompt-sensei setup` — guided setup for consent, auto-start scope, and optional redacted previews
- - `/prompt-sensei help` — show concise help
- Advanced commands:
- - `/prompt-sensei stop`
- - `/prompt-sensei report`
- - `/prompt-sensei settings`
- - `/prompt-sensei settings auto-observe on|off|folder|user`
- - `/prompt-sensei settings save-redacted-prompts on|off`
- - `/prompt-sensei settings auto-observe=off save-redacted-prompts=on`
- - `/prompt-sensei clear`
- - `/prompt-sensei update`
- Natural-language equivalents in Codex include "use prompt-sensei", "improve this prompt", "look back at my prompt history", "show my prompt-sensei report", "use prompt-sensei setup", and "turn auto observe on". Settings commands accept friendly values such as `enable/disable`, `true/false`, and aliases such as `redacted`, `previews`, and `auto-start`.
- When running scripts, use the installed skill root:
- - Claude Code: `~/.claude/skills/prompt-sensei`
- - Codex: `~/.codex/skills/prompt-sensei`
- For setup, settings, hooks, and lookback details, read [docs/skill-flows.md](docs/skill-flows.md) only when that mode is requested.
- ## Stages
- Classify every scored prompt first:
- | Stage | Use when | Score dimensions |
- |---|---|---|
- | Exploration | User is still figuring out the problem | Goal Clarity + Privacy/Safety |
- | Diagnosis | User has symptoms or evidence | Goal Clarity + Context Completeness + Privacy/Safety |
- | Execution | User wants implementation or changes | All seven dimensions |
- | Verification | User wants correctness checks | Goal, Context, Input Boundaries, Output Format, Verification, Privacy/Safety |
- | Reusable workflow | User wants a checklist, template, or process | Goal, Context, Input Boundaries, Constraints, Output Format, Verification, Privacy/Safety |
- | Action | Short follow-through directive in an established session | Goal Clarity + Privacy/Safety |
- Do not penalize Exploration or Action prompts for missing execution details. Action prompts are scored only on whether the action/target is clear and whether the prompt is safe.
- For calibration details, use [docs/scoring-rubric.md](docs/scoring-rubric.md) when needed.
- ## Dimensions
- Score applicable dimensions from 1 to 5:
- - Goal Clarity: desired outcome is clear
- - Context Completeness: enough background to act
- - Input Boundaries: what to read/use/focus on is clear
- - Constraints: scope limits and tradeoffs are stated
- - Output Format: response shape is specified
- - Verification: correctness checks are requested
- - Privacy/Safety: unnecessary sensitive data and unsafe operations are avoided
- Composite score: average applicable dimensions, multiplied by 20 and rounded. Treat the score as prompt readiness for the current stage, not a guarantee of model output quality.
- Grade labels:
- - 90-100: Excellent — ready for this stage
- - 70-89: Good — minor gaps
- - 50-69: Developing — clear improvements available
- - 30-49: Early stage — normal for exploration
- - 10-29: Needs work
- In observe mode, grade labels are not a substitute for coaching. For any scored prompt below 90, the Sensei line must include `Tip:` with one concrete next habit. It is okay to use a grade label like `Good — minor gaps` only when a concrete tip is also included.
- Choose the most useful next habit, not mechanically the lowest dimension. Apply this priority:
- - Privacy/safety issues outrank prompt polish.
- - Debugging: expected/actual behavior and exact errors outrank output format.
- - Implementation/refactoring: file boundaries and scope constraints outrank output format.
- - Code review/verification: diff or file scope outranks response polish.
- - Planning/documentation: decision criteria, audience, and context outrank engineering-only details.
- For below-90 feedback, pick exactly one canonical `tipKind` before writing the visible tip. Use the matching habit phrase below, or a very close paraphrase that keeps the same keywords. Free-form tips may not persist `tipKind` in hook-recorded events.
- | tipKind | Visible tip phrase |
- |---|---|
- | `clarify-goal` | name the exact outcome you want before adding details |
- | `add-context-evidence` | add the evidence that makes the problem diagnosable |
- | `add-expected-actual` | add expected behavior and actual behavior before asking for a fix |
- | `add-error-output` | paste the exact error output or failing assertion when it is safe |
- | `name-file-or-function` | name the file, function, command, or diff the agent should focus on |
- | `add-scope-boundary` | add one boundary such as no new dependencies, minimal diff, or no API changes |
- | `add-output-format` | ask for the response shape that will make the answer easiest to review |
- | `add-verification-command` | end with the command, test, or edge case that proves the work |
- | `redact-sensitive-data` | replace secrets, personal data, and private URLs with labeled placeholders |
- | `add-safety-check` | add confirmation, rollback, or dry-run steps before risky operations |
- | `state-decision-criteria` | state the criteria the agent should use to compare options |
- ## Observe
- When invoked by a trusted Claude Code SessionStart hook with `observe --auto-start`:
- 1. Treat observe mode as active for this session.
- 2. Be silent: do not announce Prompt Sensei, do not explain setup, do not run `settings.js`, and do not run `observe.js --init`.
- 3. Assume the hook already checked `autoObserve` and observe consent.
- 4. Answer the user's current prompt normally, then append exactly one Sensei line.
- 5. Keep auto-start quiet. Do not make visible `observe.js` recording calls; host hooks handle prompt hashing and scored-line persistence in the background when installed.
- 6. Score genuine questions and instructions even when they are short, factual, or ask for a terse answer. Skip only mechanical inputs such as one-word acknowledgements, numeric menu choices, slash-command-only wrappers, explicit "just reply ..." tests, and context-resume summaries.
- In Codex auto observe, use instruction-based observe only as a fallback for hook trust gaps. If a trusted `SessionStart` hook loads observe context, add exactly one Sensei line to final answers yourself so the user still gets coaching when the `Stop` hook is not trusted or not running. If the trusted `Stop` hook sees that final line, it persists it and does not request a continuation; if the final answer omitted the line, it asks for one continuation whose only content is the final Sensei line. Do not add Sensei lines to Codex progress/status updates while work is still ongoing, and never add more than one Sensei line to a final answer.
- When `/prompt-sensei observe` starts:
- 1. Say: "Prompt Sensei will be coaching this session. After each prompt, I'll add a one-line score. Type `/prompt-sensei report` anytime for your private summary."
- 2. Check consent with `node <skill-root>/dist/scripts/settings.js`.
- 3. If observe consent is not granted, ask with structured input first when the active host/mode exposes it: use `AskUserQuestion` in Claude Code or `request_user_input` in Codex. Keep the numbered-list yes/no prompt from [docs/skill-flows.md](docs/skill-flows.md) as the fallback when structured input is unavailable. If the user grants consent, run `node <skill-root>/dist/scripts/observe.js --init`.
- 4. After first-time consent, follow the host-aware auto-start guidance in [docs/skill-flows.md](docs/skill-flows.md). Install only the current host's native hooks.
- 5. If observe consent is already granted, do not ask again.
- For each later normal user prompt while observe mode is active:
- 1. Skip only truly low-signal prompts: one-word acknowledgements ("ok", "yes", "got it"), slash-command-only wrappers, numeric menu choices ("1", "2", "3"), explicit "just reply ..." tests, and context-resume summaries. Any genuine question or instruction must be scored, even when it is short, simple, factual, or asks for a terse answer. Append only:
- ```
- > **[[Sensei: skipped grading for low-signal prompt]]()**
- ```
- 2. Otherwise classify stage, score applicable dimensions, choose one canonical `tipKind`, and write the visible tip from that `tipKind`'s habit phrase.
- 3. Record the observation:
- ```bash
- node <skill-root>/dist/scripts/observe.js --stage <stage> --score <1-5-composite> --task-type <type> --flags <comma-separated-flags> --tip-kind <tipKind>
- ```
- 4. End the response with exactly one line:
- ```
- > **[[Sensei: 68/100 · Diagnosis; Tip: add the error message and file path]]()**
- ```
- For scores below 90, the final line must include `Tip:`. Never replace the tip with only a generic label such as `Good — minor gaps`.
- Valid flags: `missing-context`, `no-constraints`, `no-verification`, `no-output-format`, `missing-input-boundaries`, `privacy-risk`, `safety-risk`. Omit `--flags` for Action prompts unless there is a real privacy/safety issue.
- Only for scores 90+, use encouragement instead of a tip:
- ```
- > **[[Sensei: 94/100 · Execution; Excellent — ready for this stage]]()**
- ```
- ## Improve
- Do not activate observation mode and do not save raw prompt text.
- Output:
- ```txt
- Prompt Sensei Improve
- =====================
- Stage: Execution
- Score: 68 / 100 (Developing)
- What is missing:
- - error output
- - file path
- - verification command
- Improved prompt:
- [copyable prompt here]
- Habit to practice next:
- Add expected behavior and actual behavior before asking for a fix.
- ```
- Preserve the user's intent. Add only the highest-impact missing details or placeholders. Always end with exactly one habit to practice next.
- ## Other Modes
- - `/prompt-sensei setup`: read [docs/skill-flows.md](docs/skill-flows.md), then use one combined structured-input call on either host when the active mode exposes it: `AskUserQuestion` in Claude Code or `request_user_input` in Codex, with up to three questions for consent, auto-start, and redacted previews. Skip questions whose answers are already known. If structured input is unavailable, briefly explain the mode limitation and use the numbered fallback. On Codex, install Codex hooks for auto-start when requested and tell the user that a trust prompt for new or changed command hooks is expected; they should inspect the command paths with `/hooks` before enabling them.
- - `/prompt-sensei settings`: if the user asks to show settings, run `node <skill-root>/dist/scripts/settings.js` and show compact output. If the user asks to configure settings without exact values, read [docs/skill-flows.md](docs/skill-flows.md) and use the Settings Mode structured picker when available; fall back to a numbered list when structured input is unavailable. If the user provides exact values, run the matching `settings.js` command directly.
- - `/prompt-sensei lookback`: read [docs/skill-flows.md](docs/skill-flows.md), then follow the scoped-consent lookback flow.
- - `/prompt-sensei report`: run `node <skill-root>/dist/scripts/report.js` and display its Markdown output.
- - `/prompt-sensei clear`: run `node <skill-root>/dist/scripts/clear.js`.
- - `/prompt-sensei update`: run `node <skill-root>/dist/scripts/update.js --apply`.
- - `/prompt-sensei stop`: stop scoring and say "Prompt Sensei has stopped observing. Type `/prompt-sensei observe` to resume."
- ## Help
- Show:
- ```txt
- Prompt Sensei — a quiet prompt mentor for AI coding agents
- Commands:
- /prompt-sensei observe Score prompts as you write them
- /prompt-sensei improve "<prompt>" Rewrite a prompt with one teaching note
- /prompt-sensei lookback Analyze selected local prompt history
- /prompt-sensei setup Configure auto-start and privacy options
- /prompt-sensei help Show this help
- More:
- /prompt-sensei stop Stop scoring for this session
- /prompt-sensei report Show your private summary
- /prompt-sensei settings Show local settings
- /prompt-sensei update Pull the latest version and rebuild
- /prompt-sensei clear Delete local Prompt Sensei data
- Storage: ~/.prompt-sensei/ — local only, no cloud
- Privacy: Raw prompt text is never stored
- ```
- ## Tone
- - Never say "bad prompt." Say "this is a reasonable starting point for exploration."
- - One habit at a time.
- - Acknowledge stage.
- - Celebrate progress.
- - Be specific and brief.
- Task types: `debugging`, `implementation`, `code-review`, `refactoring`, `architecture`, `planning`, `documentation`, `testing`, `exploration`, `other`.
README.md · README · 253 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- # Prompt Sensei
- > Prompting is becoming one of the most important engineering skills in the AI era. Prompt Sensei helps you improve your prompts while teaching you how to improve.
- [中文说明](README-zh.md)
- [Quickstart](#5-minute-quickstart) · [FAQ](docs/faq.md) · [Privacy](docs/privacy.md) · [Examples](examples/prompt-gallery.md) · [Advanced setup](docs/advanced-setup.md) · [Discussions](https://github.com/chengzhongwei/Prompt-sensei/discussions)
- <p align="center">
- <img src="assets/prompt-sensei-banner-en.jpg" alt="Prompt Sensei banner: Claude Code and Codex support, local-first optional hooks, no raw prompts stored, and stage-aware feedback." />
- </p>
- Prompt Sensei is a quiet, local-first prompt coach for Claude Code and Codex. It gives stage-aware feedback, rewrites rough prompts into better ones, looks back on local history when you ask, and helps you practice one habit at a time.
- It is built as a real host-integrated skill, not just a markdown rubric: Prompt Sensei supports hook-based observe comments, opt-in auto-observe hooks, hash-only prompt captures, quiet Stop-hook persistence, and compact-safe continuity where the host exposes the lifecycle events.
- No cloud. No telemetry. No leaderboard. No raw prompt archive.
- Prompt Sensei's score is a coaching signal, not an objective grade or a guarantee of better output. The real test is whether the improved prompt produces a more useful first draft, fewer clarification turns, safer agent behavior, and easier verification.
- ---
- ## Why This Exists
- AI coding agents do real work: they edit files, run commands, and make implementation choices. A vague prompt does not just produce a vague answer; it can produce broad edits, hidden assumptions, and extra revision loops.
- Prompt Sensei is built around a simple belief:
- - Prompting is an engineering skill.
- - Engineering skills improve through feedback.
- - Feedback should be kind, specific, private, and useful.
- It does not treat `fix this test` as a bad prompt. It treats it as an early-stage prompt, then teaches the next step.
- ---
- ## 5-Minute Quickstart
- Install for Claude Code:
- ```bash
- git clone https://github.com/chengzhongwei/Prompt-sensei ~/.claude/skills/prompt-sensei
- (cd ~/.claude/skills/prompt-sensei && npm install && npm run build)
- ```
- Then start live coaching with `/prompt-sensei observe` inside Claude Code. For optional auto-start and privacy settings, run `/prompt-sensei setup`.
- Install for Codex:
- ```bash
- git clone https://github.com/chengzhongwei/Prompt-sensei ~/.codex/skills/prompt-sensei
- (cd ~/.codex/skills/prompt-sensei && npm install && npm run build)
- ```
- Then ask Codex in natural language: `Use prompt-sensei observe mode.` For optional hook-based auto-start, run `npm run setup-hooks -- auto-observe folder` or use `/prompt-sensei setup` after the skill is loaded.
- Start live coaching in Claude Code:
- ```txt
- /prompt-sensei observe
- ```
- Try improving a rough prompt:
- ```txt
- /prompt-sensei improve "fix this test"
- ```
- Example output:
- ```txt
- Prompt Sensei Improve
- =====================
- Stage: Exploration
- Score: 70 / 100 (Good)
- What is missing:
- - failing test name
- - expected behavior
- - actual error output
- Improved prompt:
- Help me debug this failing test.
- Test: [test name]
- Expected: [what should happen]
- Actual: [error output or wrong behavior]
- Related file: [file path]
- Return:
- 1. Likely root cause
- 2. Minimal fix
- 3. Test command to verify
- Habit to practice next:
- Add expected and actual behavior before asking for a fix.
- ```
- See [examples/prompt-gallery.md](examples/prompt-gallery.md) for more copyable before/after prompts.
- ---
- ## Supported Environments
- Prompt Sensei works best when the host tool can load the skill directly. It supports both Claude Code and Codex, with host-native hooks for optional auto-observe.
- | Invocation style | Environments |
- |---|---|
- | Direct skill command, e.g. `/prompt-sensei improve "fix this test"` | Claude Code |
- | Natural language, e.g. `Use prompt-sensei to improve this prompt...` | Codex |
- Cursor and other AI coding tools are not supported yet. `/prompt-sensei observe` is a Claude Code skill command. In Codex, natural-language requests depend on the host agent loading the installed skill. See [advanced setup](docs/advanced-setup.md#host-support) for the Claude Code/Codex hook matrix and trust notes.
- ---
- ## Beginner Path
- Start with:
- ```txt
- /prompt-sensei observe
- ```
- On first use, Prompt Sensei asks for local storage consent, starts coaching, then offers an optional auto-start choice. The simplest path is manual start; you can still use `/prompt-sensei observe` whenever you want.
- For a one-time guided setup, use:
- ```txt
- /prompt-sensei setup
- ```
- Setup covers observe consent, optional host auto-start hooks, and whether to save redacted prompt previews. In Codex, a trust prompt for new or changed command hooks is expected; inspect commands with `/hooks` before enabling them. Advanced details live in [docs/advanced-setup.md](docs/advanced-setup.md).
- ---
- ## Commands
- ```txt
- /prompt-sensei [observe|improve|lookback|setup|help]
- ```
- Essential commands:
- ```txt
- /prompt-sensei observe # start live coaching
- /prompt-sensei improve "fix this" # rewrite a prompt with one teaching note
- /prompt-sensei lookback # analyze selected local prompt history
- /prompt-sensei setup # guided setup
- /prompt-sensei help
- ```
- More commands:
- ```txt
- /prompt-sensei stop # stop coaching this session
- /prompt-sensei report # show local session trends
- /prompt-sensei settings # show local settings
- /prompt-sensei update # pull latest version and rebuild
- /prompt-sensei clear # delete local Prompt Sensei data
- ```
- In Codex, use natural language equivalents such as `Use prompt-sensei observe mode.` or `Use prompt-sensei to improve this prompt: "fix this test"`.
- See [docs/advanced-setup.md](docs/advanced-setup.md) for settings commands, hook setup, Codex examples, and consent scopes.
- For direct script checks, see [advanced setup](docs/advanced-setup.md#local-script-checks).
- ---
- ## Settings
- Prompt Sensei stores local preferences in `~/.prompt-sensei/settings.json`. Defaults are intentionally quiet: auto observe off, redacted prompt previews off, and raw prompts never stored. Use `/prompt-sensei setup` for a guided path or [docs/advanced-setup.md](docs/advanced-setup.md) for the full settings reference.
- ---
- ## How It Teaches
- Prompt Sensei is stage-aware. A short exploration prompt can be reasonable early, while an execution prompt needs clearer context, boundaries, and verification.
- | Stage | Meaning | Example |
- |---|---|---|
- | Exploration | Still figuring out the problem | `why is this broken` |
- | Diagnosis | Have evidence or symptoms | `expected /login, actual /dashboard` |
- | Execution | Want implementation or changes | `implement this with these constraints` |
- | Verification | Want correctness checks | `find edge cases and test commands` |
- | Reusable workflow | Want a repeatable process | `create a code review checklist` |
- | Action | Short follow-through directive | `ok commit and push to main` |
- The coaching line stays small:
- > **[[Sensei: 68/100 · Diagnosis; Tip: add the error message and file path]]()**
- Think of the score as prompt readiness for the current stage, not universal prompt quality. A 100/100 prompt can still produce a weak answer if the model lacks domain knowledge, the task is ambiguous outside the prompt, or the rubric does not fit the user's domain.
- The report focuses on repeated habits:
- ```txt
- Next habit: End with the command, test, or edge case that proves the work.
- Repeated gap: add-verification-command (5×)
- Average score: 81 / 100 (Good)
- ```
- For the full philosophy, read [docs/philosophy.md](docs/philosophy.md). For scoring details, read [docs/scoring-rubric.md](docs/scoring-rubric.md).
- ---
- ## How Do I Know It Helps?
- Prompt Sensei does not prove that a prompt is better just because the number went up. It gives a structured coaching signal.
- To check whether it helped, compare the original and improved prompt on the same task: fewer clarification turns, fewer surprising edits, a better first answer, and easier verification. For calibration details and the skeptical version, see [FAQ](docs/faq.md).
- ---
- ## Lookback
- `/prompt-sensei lookback` analyzes selected local Claude Code or Codex history after separate consent. It can inspect one session or recent sessions, generate coaching, and save a Markdown report only after confirmation. Privacy details live in [docs/privacy.md](docs/privacy.md#lookback-privacy), and workflow details live in [docs/skill-flows.md](docs/skill-flows.md#lookback-flow).
- ---
- ## Optional Host Hooks
- Host hooks provide opt-in auto-start, hash-only background captures, hook-triggered observe comments, and quiet persistence of the final Sensei score line. They are optional and stay quiet unless consent/settings allow them. See [advanced setup](docs/advanced-setup.md#host-support), [Claude example hooks](examples/claude-settings.example.json), and [Codex example hooks](examples/codex-hooks.example.json).
- ---
- ## Privacy
- Prompt content is sensitive. Prompt Sensei stores nothing until you consent. After consent, it stores local metadata under `~/.prompt-sensei/`; raw prompts are never stored.
- Prompt Sensei's local scripts do not send prompt text, scores, reports, or local event data to a service. Lookback may show redacted user prompts to the current AI agent after separate consent.
- See [docs/privacy.md](docs/privacy.md) for details.
- ---
- ## Contributing
- Good first areas:
- - realistic prompt improvement examples
- - scoring rubric improvements
- - redaction rule improvements
- - report improvements
- - support for other AI coding tools
- Please keep changes aligned with the core philosophy: quiet, local-first, encouraging, and privacy-aware.
- ---
- ## License
- Apache-2.0 — Copyright 2026 Chengzhong Wei
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.