Reflect Skill Claude — Security skill for Claude Code
3-phase /reflect skill for Claude Code: extract session learnings, audit your config against what actually happened, compress CLAUDE.md against codebase truth.
How to install Reflect Skill Claude
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open hansvangent/reflect-skill-claude and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Reflect Skill Claude does
3-phase /reflect skill for Claude Code: extract session learnings, audit your config against what actually happened, compress CLAUDE.md against codebase truth.
Alternatives in Security
- Engage.Exploit — Execute Phase 4 - Exploitation and Access Establishment 348 ★
- Competitive Audit — Capture a competitor website with Playwright and extract its color system, typography, layout patterns, and 's 309 ★
- Skill Threat Modeling — Code-First Deep Risk Analysis Skill for Claude Code - 8-Phase Workflow with Security design review, STRIDE Thr 207 ★
README
/reflect
A Claude Code skill that runs a 3-phase retrospective at the end of any session and keeps your CLAUDE.md sharp over time.
Most sessions end with a close button and you lose the corrections, the preferences, the workflows that finally clicked. `/reflect` extracts what's worth keeping, audits your existing config against what actually happened, and (when the file gets bloated) compresses CLAUDE.md against codebase truth without losing any live rule.
Install
The skill is a single file. Drop it into your global skills directory.
mkdir -p ~/.claude/skills/reflect
curl -fsSL https://raw.githubusercontent.com//reflect/main/SKILL.md \
-o ~/.claude/skills/reflect/SKILL.md
Or clone the repo and symlink it:
git clone https://github.com//reflect.git
ln -s "$(pwd)/reflect/SKILL.md" ~/.claude/skills/reflect/SKILL.md
Restart Claude Code (or open a fresh session). Type `/reflect` to test.
The skill has `disable-model-invocation: true`, so it only fires when you explicitly invoke it. Nothing runs in the background.
Usage
Args are space-separated tokens after `/reflect`. Order doesn't matter.
| Command | What it does |
|---|---|
/reflect |
Phase 1 → Phase 2 → Phase 3 (if size triggers fire) |
/reflect project |
Same, but extends scope to the project's CLAUDE.md and .claude.local.md in cwd |
/reflect compress |
Skip Phases 1 + 2, run only Phase 3 (compression pass) |
/reflect compress project |
Compression pass over both global and project files |
How each phase works
Phase 1: Extract
Walks the conversation top-to-bottom and sorts everything into four buckets:
- Durable facts about you (role, tools, environment, preferences) → appended to
~/.claude/CLAUDE.md - Standing corrections (you said "stop doing X" or "always do Y") → appended with the WHY and WHEN
- Repeatable workflows that pass a conservative "would you actually rerun this?" bar → saved as a new sk
Related Skills
Sota Audit
Audit this codebase against the library — agree the scope first, route from the surfaces rather than from what
Gtd Audit
A Claude skill that audits task lists against GTD (Getting Things Done) principles — Capture, Clarify, Organiz
Release Audit
Use right before cutting a release — it spawns ONE fresh sibling session via CCC that audits only the JUDGMENT
Sessionxray
Security audit of Claude Code session transcripts: what did the agent actually touch.
Game Localization
Engine-agnostic game localization pipeline for AI agents: detect engine -> extract text -> translate with cont
Closeout
Post-project audit. Diffs SCOPE.md deliverables against ground truth using tier-aware verification (code-done
Related Agents
Chronicler
You are a specialized session analyst. Your job is to analyze past sessions, extract learnings, and find relev
Gsd Mempalace Curator
Ship-time MemPalace curation — writes the session diary, proposes/creates cross-project tunnels, mirrors extra
Calendar Audit
Use this agent to review calendar events, log what happened, and handle rescheduling. Can run daily or cover m