菜单
Build and maintain the user's OWN voice profile from continuously captured writing plus explicit samples — status, build/refresh, add samples, capture on/off. Use when the user wants posts in their own voice, wants to see what's been learned about their style, or wants to manage capture.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Write a social post in one of the roster's voices (or the user's own `self` voice) — platform-native, craft-floored, scanner-verified. Use when the user wants a post, caption, announcement, reply, or any short-form text written as a distinct human voice.
One brief, N posts, N genuinely different people — voice-isolated writers, cross-post overlap detection, per-post verification receipts. Use for launching/promoting something across multiple posts or platforms without the posts sharing a fingerprint.
Two-layer AI-tell audit of any text — deterministic scanner (lexicon, constructions, stylometry with line numbers) plus a blind semantic judge. Use when the user wants to know whether text reads AI-generated, what specifically gives it away, and how to fix it.
Rewrite existing text so it reads human — either scrubbed-and-souled in its own register, or fully re-voiced as one of the roster's people. Use when the user has a draft that "sounds like AI" or wants existing copy converted into a specific voice.
Front door and router for the idiolect voice engine. Use when the user mentions idiolect, asks what voices are available or who should write something, wants writing that "doesn't sound like AI", says text "sounds like ChatGPT/AI slop", or asks for human-sounding posts and no more specific idiolect skill has already matched. Also the first-run onboarding.
Create a new voice profile — from a writing corpus (measured, not vibed), or invented to fill a gap in the roster. Extends the roster beyond the 60+ built-ins. Use when the user wants a new/custom voice, a voice "like these samples", or a voice the roster lacks.
| name | self |
| description | Build and maintain the user's OWN voice profile from continuously captured writing plus explicit samples — status, build/refresh, add samples, capture on/off. Use when the user wants posts in their own voice, wants to see what's been learned about their style, or wants to manage capture. |
| argument-hint | status | build | add <file|paste> | on | off | drift |
The roster's voices are fictional. This one is real: a running stylometric picture of how YOU actually write, so drafts in "your voice" are measured against you, not against a compliment. It's also the anti-homogenization move — co-writing with RLHF-tuned models measurably pulls writers toward one style (Padmakumar & He 2024; CHI 2025 showed the pull is strongest on non-Western writers). A maintained self profile is how your idiolect survives the tooling.
ROOT="${CLAUDE_PLUGIN_ROOT:-${CODEX_PLUGIN_ROOT:-$IDIOLECT_ROOT}}"
# unset everywhere (opencode)? ROOT = two directories up from this SKILL.md
IDIO="python3 $ROOT/scripts/idiolect.py"
On Claude Code, a UserPromptSubmit hook feeds each of your prompts through a filter: prose-like messages ≥ 15 words are kept; commands, code, paths, and logs are rejected. Kept text lands in a LOCAL corpus ($IDIO path → self_corpus) that never leaves the machine, never enters the repo, and caps itself (samples outlive prompts when it compacts). self off stops capture instantly; self clear deletes the corpus. On Codex/opencode there's no per-prompt hook — feed it explicitly with add.
Prompt text is a REGISTER of you (terse, imperative, workish), not the whole you. Explicit samples of real writing (your posts, emails, comments) are tagged sample and weigh more in the build. The more samples, the truer the profile.
$IDIO self status: corpus size, sample/prompt split, capture on/off, ready-for-build.$IDIO self add --file <f> (chunks and tags as samples). This is the highest-leverage thing a user can do for profile quality; nudge for 3–5 real posts/emails.clear (destructive).$IDIO self status --json — if not ready_for_build, say what's missing (~800 words or 5 samples) and offer add.$IDIO fingerprint --self --json > fp.json — the measured stylometrics, with the sample/prompt provenance split.$IDIO synth-scaffold --slug self --from-fingerprint fp.json — skeleton with THEIR real numbers.banned list — start it with whatever corporate words they hate). Exemplars: prefer 3 REAL posts from their samples (lightly trimmed) over synthetic ones; synthetic exemplars only with their sign-off.$IDIOLECT_HOME/voices/self.md, run $IDIO validate self (custom voices get the relaxed exemplar threshold), fix until ok./idiolect:write self <platform> — <brief> writes as them, conform-checked against their own parameters.fingerprint --self, diff the numbers against the profile's stylo block, report what changed ("your sentences got 20% shorter since March; em dashes crept in — keep or update?"). Update the profile only with consent; drift is sometimes growth and sometimes contamination, and only the owner knows which.Corpus and profile are local files under the user's home; nothing is uploaded, and this skill never pastes raw corpus entries into chat beyond short quoted fragments needed for the build conversation. If the user asks what's stored: $IDIO self corpus --tail 10 shows them, self clear erases it, self off stops it. Their voice is their asset — treat the profile file as sensitive as an SSH key.