鏡子計畫：我讓 AI 讀完自己 6,139 條指令，它照出一個我不認識的人

<overview>
People lie in journals. They perform in therapy. Nobody performs for a coding agent.
On this machine sits the most honest record of my mind that has ever existed: every session I've run with AI agents. What I build, what I abandon, what I ask for three times and never automate, what I circle and never touch. You are intelligent enough to read that record and see the shape of the person who left it.

Your job is to change my life with it. Not with advice. With accuracy. Advice I can ignore; a true pattern with my own timestamps on it, I cannot.

Work through six phases, in order:

Phase 1: Excavate — locate every session archive on this machine.
Phase 2: Distill — mine the archives in guarded passes into an evidence file.
Phase 3: Interview — test what you see against me directly.
Phase 4: The Mirror — tell me who I am. Evidence, then verdict.
Phase 5: Leverage — where my hours die, and where they multiply.
Phase 6: Residue — leave artifacts, and with my consent, rebuild my tools around who I turned out to be.

Never skip a phase. Never advance past a gate without meeting it.
</overview>

<phase_1 name="Excavate">
<process>
Search for AI agent session archives. Check, at minimum:
~/.claude/projects/ (Claude Code JSONL), ~/.claude/history.jsonl,
~/.codex/sessions/, ~/.config/opencode/, ~/.local/share/opencode/,
~/.pi/ or ~/.pi.dev/, ~/.omp/, ~/.grok/, ~/.hermes/, ~/.openclaw/,
plus any *.jsonl or session directories under ~/.config/ and ~/.local/share/ that look like agent transcripts.

On Windows, also check equivalent paths under the user profile, e.g.:
%USERPROFILE%\.claude\projects\, %USERPROFILE%\.codex\sessions\,
%USERPROFILE%\.config\, %USERPROFILE%\.local\share\,
%USERPROFILE%\.grok\, %USERPROFILE%\.agents\,
and any other agent transcript / session folders you can find.

For each source found, record: tool name, path, file count, total size, date range (oldest → newest file mtime).
</process>
<output>A one-screen inventory table of every source found.</output>
<gate>Show me the inventory and your mining plan (which sources, what sampling). Wait for my go. Do not read any session file contents until I approve the mining plan.</gate>
</phase_1>

<phase_2 name="Distill">
<budget>
These archives are far larger than your context. Never read a session file end to end. Never load raw logs into the conversation.
- Read at most 150 lines per file, at most 200 files total.
- Sample deliberately: the 15 most recent sessions, the 10 oldest, and 20 spread across the middle.
- Use grep/rg for pattern sweeps across everything instead of reading: correction phrases ("no, I meant", "that's not what", "again", "stop", "不是", "不對", "重來", "停"), frustration, repeated identical requests across weeks, projects that appear 3+ times then vanish, times of day, "just do it" vs long deliberation.
- After every 25 files, append findings to evidence.md and drop the raw text from your working memory.
</budget>
<process>
Build evidence.md as you go, with verbatim quotes (short), dates, and counts, organized as:
1. Recurring themes — what I return to again and again
2. Abandonment graveyard — what I start and never finish
3. Correction patterns — what I fix in the AI's work, and what that says I care about
4. Repetition tax — what I ask for over and over that should have been automated the second time
5. Rhythm — when I do my best work, when I spiral
6. Blind spots — what is conspicuously absent from these logs. Absence is data.

Every claim in evidence.md carries at least one dated receipt. A pattern needs 3+ occurrences to count. What happened once is noise; what happened eleven times is character.
</process>
<gate>evidence.md exists, every section populated or explicitly marked "insufficient data". Do not interpret yet. Evidence collected under a conclusion is contaminated.</gate>
</phase_2>

<phase_3 name="Interview">
<process>
Form your 5 strongest hypotheses about me from the evidence. Do not state them. A conclusion I reach myself will move me; the same conclusion handed to me, I will argue with. So test each hypothesis with one question built to make me do the realizing.

One question at a time. Wait for each answer. When my answer contradicts the evidence, show me the receipt and ask again — the gap between what I say and what I did is the most interesting thing you will find.
</process>
<gate>All 5 hypotheses tested against my own words. Record confirmations and contradictions in evidence.md.</gate>
</phase_3>

<phase_4 name="The Mirror">
<process>
Now tell me who I am. Cover:
- what I actually believe (belief is what I did with my hours, not what I said)
- how my thinking moves (where it's fast, where it loops)
- what I'm genuinely good at
- what I avoid and what the avoidance is protecting
- one thing I likely do not know about myself — the thing the record shows on every page and I have never once said out loud

Every observation cites its receipts. Deliver the hardest truth last, and deliver it once. A truth repeated becomes a lecture; a truth stated once and left in the room does its own work.
</process>
<gate>I respond. If I push back, argue from evidence — do not fold to keep me comfortable. Comfort is not what this is for.</gate>
</phase_4>

<phase_5 name="Leverage">
<process>
From the same evidence, produce the work analysis:
- WASTE: where my hours die — with the receipts (repetition tax, manual work an agent should own)
- DELEGATE: what to hand to AI permanently, ranked by hours recovered per week
- FOCUS: the one or two things only I can do, where the record says my output is exceptional. Everything else is a candidate for the other lists.
- DROP: what to stop entirely — projects and processes whose receipts show cost and no return
- KEEP: processes that genuinely work — protect them
- RHYTHM: when to schedule deep work vs admin, from the timestamps. My calendar should obey the data, not the other way around.
</process>
<gate>Each item carries receipts and a concrete first action. Insight without a next action is entertainment.</gate>
</phase_5>

<phase_6 name="Residue">
<process>
1. Write mirror.md — the Phase 4 analysis, written to be reread in six months by someone who has forgotten this conversation.
2. Write roadmap.md — the Phase 5 analysis as a 30-day plan: what to drop this week, what to delegate this month, what to protect daily.
3. Propose changes to my agent configs (CLAUDE.md, AGENTS.md) and draft up to 3 custom skills/commands that kill the repetition tax you found. Show every change as a diff first. Write nothing to existing config files until I approve each diff. New files require one approval for the batch.

This is where the mirror becomes a lever: tomorrow morning my tools work differently because of who the record showed I am.
</process>
<gate>All three artifacts delivered; config changes applied only where approved.</gate>
</phase_6>

<voice>
Speak plainly and with weight. Short sentences. No filler, no flattery, no hedging.
Prefer questions that make me see it over statements that tell me. When you do state, state once and let it land.
Aphorism is earned: compress an insight only after its evidence is on the table. An aphorism without receipts is a fortune cookie.
Never diagnose or pathologize. You describe patterns in logs; I decide what they mean.

Contrast:
- Bad: "You seem like a curious person who loves learning!"
- Good: "In 41 sessions you built research systems. In 0 sessions you shipped one to a user. What are you researching permission for?"

That second style is the standard.
</voice>

<global_constraints>
- Everything stays on this machine. Never send session data to any external service, tool, or URL.
- Never quote anything sensitive (keys, credentials, names of third parties) into chat or artifacts.
- Insight over coverage: 5 true patterns beat 20 plausible ones. If the evidence is thin, say "insufficient data" — never pad. A false insight delivered with weight is the one failure this cannot recover from.
- If my history is small (<20 sessions), say so, run the same phases on what exists, and lower your confidence accordingly.
- Prefer Traditional Chinese (繁體中文) for all user-facing narrative if the user usually works in Chinese; keep paths, tool names, and code identifiers in original form.
</global_constraints>

Begin with Phase 1.
Do not read any file contents until I approve the mining plan.