项目文件夹

文件
wehub-resource-sync a1fa97429b
Deploy Documentation to Pages / build (push) Has been cancelled
Deploy Documentation to Pages / deploy (push) Has been cancelled
Release / Tag + GitHub Release (push) Has been cancelled
Sync Codex Skills Symlinks / sync (push) Has been cancelled
chore: import upstream snapshot with attribution
2026-07-13 12:41:47 +08:00

4.8 KiB

title, description
title description
/cs-reflect — Slash Command for AI Coding Agents /cs:reflect — Mid-conversation reflection: halts current thread, re-reads full conversation from original goal forward, runs 5-dimension analysis. Slash command for Claude Code, Codex CLI, Gemini CLI.

/cs-reflect

:material-console: Slash Command :material-github: Source

Command: /cs:reflect

The cs-reflect persona pauses execution and honestly reassesses where the conversation has been heading.

When to Run

  • Conversation has gone 10+ turns deep on implementation details without strategic check-in
  • Repeated dead-ends or pivots within a short span
  • You suspect the framing has drifted from the original goal
  • You want a bias check before committing to next steps
  • Pre-decision sanity check on a substantive direction change

When NOT to Run

  • Quick lookups or factual questions
  • Conversations <5 turns deep (not enough to reflect on)
  • Mid-task when you just need execution, not reassessment

What You Get

A flowing-prose reassessment covering:

  1. Macro Perspective — original goal vs current direction; drift detection
  2. Gap Analysis — unverified assumptions, missing stakeholders, skipped constraints, dismissed alternatives
  3. Reflective Inquiry — right problem vs adjacent easier one? Simpler path overcomplicated? Harder valuable path avoided?
  4. Bias Check — confirmation / sunk cost / anchoring / complexity / recency
  5. Contextual Alignment — does direction serve goals + best use of time

Closing with one of three recommendations:

  • Continue — and why (specific evidence)
  • Pivot to {direction} — and what to drop
  • Pause for {question} — and which question to answer first

Trigger Phrases (auto-invoke without /cs:)

Explicit:

  • "reflect"
  • "take a step back" / "step back"
  • "zoom out"
  • "are we missing something"
  • "bigger picture"
  • "what are we missing"
  • "let's pause"
  • "sanity check this"
  • "are we on track"
  • "are we overthinking this"
  • "forest for the trees"

Implicit (no phrase needed):

  • 10+ turns of implementation detail without strategic check-in
  • User shows signs of frustration or stuck-ness
  • Repeated dead-ends or pivots within a short span

Discipline

  • Re-read FULL conversation — from original goal forward, not just recent turns
  • Honest output — no manufactured problems when path is solid; specific reasoning when validating
  • Flowing prose — no headers, no bullet lists
  • Specific evidence — anchor every observation to specific conversation moments
  • Closing recommendation mandatory — Continue / Pivot / Pause every time
  • Low-intake — max 1 optional clarifier (only when context is too thin)
  • No name references — generic second-person throughout

Workflow

# When triggered, the skill:
# 1. Halts current thread (no continuation of the in-progress task)
# 2. Re-reads full conversation from original goal
# 3. Runs 5-dimension analysis in head
# 4. Delivers flowing-prose reassessment

# Optional pre-flight: scan for bias patterns + depth signals
python ../skills/reflect/scripts/conversation_depth_analyzer.py --conversation /tmp/transcript.txt
python ../skills/reflect/scripts/bias_pattern_detector.py --conversation /tmp/transcript.txt

# Post-flight: validate output meets discipline
python ../skills/reflect/scripts/directional_recommendation_validator.py --output /tmp/output.txt

Stop Conditions

  • Reflection complete + closing recommendation delivered → done
  • Context too thin → 1 clarifying question, then run
  • User says "stop reflecting" → drop back to task immediately

Anti-Patterns Rejected

  • Hardcoded user names or specific domain references
  • Structured-report output (headers, bullets) when prose is required
  • Manufactured problems when things are actually fine
  • Vague reassurance ("looks good!") instead of specific reasoning
  • Reassessing only recent turns instead of the full conversation
  • Skipping the closing directional recommendation

Version: 1.0.0 Source: Path-B direct conversion of megaprompts/02-reflect-megaprompt.md