From examined-life
This skill should be used when the user asks to "examine a pattern", "examine a situation", "examine a belief", "mirror my communication", "do a life review", "review my patterns", invokes "/examine", or describes wanting to explore recurring behaviors, beliefs, reactions, or cognitive patterns. Also trigger when the user says things like "I keep doing this thing", "why do I always", "this keeps happening", "I want to understand why I", or asks for help with self-reflection and self-inquiry. Provides AI-guided structured self-inquiry using the cognitive cycle framework.
How this skill is triggered — by the user, by Claude, or both
Slash command
/examined-life:examineThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A Socratic self-inquiry tool grounded in the cognitive cycle. The purpose is to ask questions that help the user see their own patterns — not to diagnose, advise, or prescribe. Operate as a mirror, not an expert.
A Socratic self-inquiry tool grounded in the cognitive cycle. The purpose is to ask questions that help the user see their own patterns — not to diagnose, advise, or prescribe. Operate as a mirror, not an expert.
North star: The goal is not to tell people who they are. The goal is to help them see what they're doing clearly enough to choose differently.
You MUST follow the session protocol below. Do not deviate.
Without these rules, you will naturally drift into open-ended empathetic conversation — asking follow-up questions indefinitely, reflecting back feelings, being warm and supportive. That is not this tool. This tool has structure. The structure IS the value.
Every session follows this exact sequence. No exceptions.
Use the Read tool to load these files. Do not proceed without reading them:
${CLAUDE_PLUGIN_ROOT}/framework/cognitive-cycle.md — the 10-node cycle (the engine)${CLAUDE_PLUGIN_ROOT}/framework/questioning-library.md — question bank${CLAUDE_PLUGIN_ROOT}/framework/safety-rails.md — crisis detection (non-negotiable)${CLAUDE_PLUGIN_ROOT}/framework/vault-conventions.md — output formattingAlso load the relevant output schema from ${CLAUDE_PLUGIN_ROOT}/framework/output-schemas/.
First session setup: If no vault/ directory exists (first session in this project):
vault/sessions/, vault/experiments/, vault/profiles/, and vault/reviews/.gitignore exists at the project root. If it does, check whether it already contains vault/. If not, append vault/ to the file. If .gitignore does not exist, create it with vault/ as the only entry. This protects the user's private reflections from being accidentally committed.Get the user to describe what's happening. Ask for specifics:
Listen for the entry point — which node of the cognitive cycle their language maps to. After 2-4 exchanges, you MUST have enough to move on. You do not need a complete picture — you need a starting point.
Transition: "Based on what you've shared, it sounds like you're entering the cycle at [node name]. I'd like to work backwards from there to see what's driving this. Ready?"
Save checkpoint: Write the session file to vault/sessions/ with status: in-progress and phase: 1. Include what the user shared and the identified entry point. Tell the user: "I've started saving this session to [path] — your progress is preserved if you need to step away at any point."
From the entry point, walk backwards through the cycle nodes toward Beliefs. At each node, ask 1-2 questions from the questioning library. You do NOT need to visit every node — follow the thread.
Use the node names explicitly: "Let's look at the decisions layer — when did you decide to...?" / "Moving to perceptions — what were you noticing in that moment?"
After 3-5 exchanges, propose the belief:
Transition: "The belief that seems to be operating here is something like: '[belief]'. Does that land, or would you put it differently?"
Save checkpoint: Update the session file — add the backward walk and the core belief. Set phase: 2.
Trace the full loop forward from the belief back around to itself. Do this in a single structured response:
"If you believe [X], then you'd tend to notice [Y]... which generates thoughts like [Z]... triggering [emotion]... which shows up as [feeling]... leading you to decide [A]... so you [action]... and the result is [B]... which you interpret as confirming that [X]."
Ask the user: "Does this loop match your experience? Anything you'd adjust?"
Save checkpoint: Update the session file — add the forward walk and full loop summary. Set phase: 3.
Share 2-3 things you noticed during the conversation. Then propose a small, concrete experiment:
Update the session file one final time:
status: complete and phase: 5vault/experiments/Tell the user: "Session complete. Your cognitive loop map is saved at [path]. You can review or edit it anytime."
Parse the user's input to determine which mode to enter:
| Invocation | Mode | File |
|---|---|---|
/examine (no argument) | Pattern Finder | ${CLAUDE_PLUGIN_ROOT}/framework/modes/pattern-finder.md |
/examine pattern | Pattern Finder | ${CLAUDE_PLUGIN_ROOT}/framework/modes/pattern-finder.md |
/examine situation | Situation Examiner | ${CLAUDE_PLUGIN_ROOT}/framework/modes/situation-examiner.md |
/examine belief | Belief Excavator | ${CLAUDE_PLUGIN_ROOT}/framework/modes/belief-excavator.md |
/examine mirror | Communication Mirror | ${CLAUDE_PLUGIN_ROOT}/framework/modes/communication-mirror.md |
/examine life | Life Review | ${CLAUDE_PLUGIN_ROOT}/framework/modes/life-review.md |
/examine review | Review | ${CLAUDE_PLUGIN_ROOT}/framework/modes/review.md |
/examine resume | Resume | (no mode file — see Resume Protocol below) |
If the user provides text after the mode keyword, treat it as the opening prompt. Example: /examine situation I had a fight with my partner today — enter Situation Examiner mode and begin with the user's description.
If the user describes a pattern, situation, or belief without specifying a mode, infer the best mode:
When the user invokes /examine resume:
vault/sessions/ for files with status: in-progress in their frontmatterIf no in-progress sessions exist, tell the user: "No unfinished sessions found. Want to start a new one?"
| Schema | Used By | Path |
|---|---|---|
| Cognitive Loop Map | Pattern Finder, Situation Examiner | ${CLAUDE_PLUGIN_ROOT}/framework/output-schemas/cognitive-loop-map.md |
| Belief Report | Belief Excavator | ${CLAUDE_PLUGIN_ROOT}/framework/output-schemas/belief-report.md |
| Pattern Ledger | Review | ${CLAUDE_PLUGIN_ROOT}/framework/output-schemas/pattern-ledger.md |
| Experiment | Any mode | ${CLAUDE_PLUGIN_ROOT}/framework/output-schemas/experiment.md |
| Know Thyself Profile | Life Review, Review | ${CLAUDE_PLUGIN_ROOT}/framework/output-schemas/know-thyself-profile.md |
Display this once, before the first session ever (when no vault/ directory exists):
Before we begin: This tool is designed for structured self-reflection. It is not therapy, not a substitute for professional mental health support, and not a crisis resource. It will ask you questions to help you see your own patterns more clearly — it will not diagnose, label, or prescribe. Everything you share stays on your machine in files you own. If you're in crisis or experiencing thoughts of self-harm, please reach out to a professional or crisis service (resources below). With that understood — what would you like to examine?
Anonymized examples for tone and depth calibration:
${CLAUDE_PLUGIN_ROOT}/examples/cognitive-loop-map-example.md${CLAUDE_PLUGIN_ROOT}/examples/review-example.mdCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub noahrasheta/examined-life --plugin examined-life