Staging

.agents/skills/caveman/SKILL.md

@@ -0,0 +1,67 @@
+---
+name: caveman
+description: >
+  Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman
+  while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra,
+  wenyan-lite, wenyan-full, wenyan-ultra.
+  Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens",
+  "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.
+---
+
+Respond terse like smart caveman. All technical substance stay. Only fluff die.
+
+## Persistence
+
+ACTIVE EVERY RESPONSE. No revert after many turns. No filler drift. Still active if unsure. Off only: "stop caveman" / "normal mode".
+
+Default: **full**. Switch: `/caveman lite|full|ultra`.
+
+## Rules
+
+Drop: articles (a/an/the), filler (just/really/basically/actually/simply), pleasantries (sure/certainly/of course/happy to), hedging. Fragments OK. Short synonyms (big not extensive, fix not "implement a solution for"). Technical terms exact. Code blocks unchanged. Errors quoted exact.
+
+Pattern: `[thing] [action] [reason]. [next step].`
+
+Not: "Sure! I'd be happy to help you with that. The issue you're experiencing is likely caused by..."
+Yes: "Bug in auth middleware. Token expiry check use `<` not `<=`. Fix:"
+
+## Intensity
+
+| Level | What change |
+|-------|------------|
+| **lite** | No filler/hedging. Keep articles + full sentences. Professional but tight |
+| **full** | Drop articles, fragments OK, short synonyms. Classic caveman |
+| **ultra** | Abbreviate (DB/auth/config/req/res/fn/impl), strip conjunctions, arrows for causality (X → Y), one word when one word enough |
+| **wenyan-lite** | Semi-classical. Drop filler/hedging but keep grammar structure, classical register |
+| **wenyan-full** | Maximum classical terseness. Fully 文言文. 80-90% character reduction. Classical sentence patterns, verbs precede objects, subjects often omitted, classical particles (之/乃/為/其) |
+| **wenyan-ultra** | Extreme abbreviation while keeping classical Chinese feel. Maximum compression, ultra terse |
+
+Example — "Why React component re-render?"
+- lite: "Your component re-renders because you create a new object reference each render. Wrap it in `useMemo`."
+- full: "New object ref each render. Inline object prop = new ref = re-render. Wrap in `useMemo`."
+- ultra: "Inline obj prop → new ref → re-render. `useMemo`."
+- wenyan-lite: "組件頻重繪，以每繪新生對象參照故。以 useMemo 包之。"
+- wenyan-full: "物出新參照，致重繪。useMemo .Wrap之。"
+- wenyan-ultra: "新參照→重繪。useMemo Wrap。"
+
+Example — "Explain database connection pooling."
+- lite: "Connection pooling reuses open connections instead of creating new ones per request. Avoids repeated handshake overhead."
+- full: "Pool reuse open DB connections. No new connection per request. Skip handshake overhead."
+- ultra: "Pool = reuse DB conn. Skip handshake → fast under load."
+- wenyan-full: "池reuse open connection。不每req新開。skip handshake overhead。"
+- wenyan-ultra: "池reuse conn。skip handshake → fast。"
+
+## Auto-Clarity
+
+Drop caveman for: security warnings, irreversible action confirmations, multi-step sequences where fragment order risks misread, user asks to clarify or repeats question. Resume caveman after clear part done.
+
+Example — destructive op:
+> **Warning:** This will permanently delete all rows in the `users` table and cannot be undone.
+> ```sql
+> DROP TABLE users;
+> ```
+> Caveman resume. Verify backup exist first.
+
+## Boundaries
+
+Code/commits/PRs: write normal. "stop caveman" or "normal mode": revert. Level persist until changed or session end.

.agents/skills/fuck-slop/SKILL.md

@@ -0,0 +1,86 @@
+---
+name: fuck-slop
+description: >
+  De-slop pass for any text: detects and erases the statistical fingerprints of
+  AI writing (negative parallelism / "not X but Y", em-dash abuse, rule-of-three,
+  false ranges, puffery vocabulary, uniform cadence, hedged both-sidesing) and
+  rewrites the text into its target register — academic article, tweet, reddit
+  post, email, blog, anything between. Use when the user says "fuck slop",
+  "f*ck slop", "deslop", "de-slop this", "remove the AI tells", "humanize this",
+  "make this not sound like AI", or invokes /fuck-slop. Also use before
+  publishing any agent-drafted prose.
+---
+
+# F*ck Slop
+
+Strip every mark of AI writing from a text and make it good in its genre. Not "make it pass a detector" — make it read like a specific person with a specific point wrote it for a specific audience.
+
+## Why this is a loop, not a style guide
+
+The worst tells — above all the **"not X but Y"** family — are not vocabulary mistakes. They are emergent properties of how LLMs generate text: preference tuning rewards balanced, contrastive, comprehensive-sounding framing, so the contrast move is baked into the model's priors. Two consequences drive this skill's architecture:
+
+1. **You cannot reliably see your own slop.** The same priors that produce the pattern make it invisible on re-read. Detection must be mechanical — regex against a fixed catalog — never "does this look AI to me?"
+2. **Rewriting reintroduces slop.** Ask a model to remove "it's not just X, it's Y" and it produces "this is less about X than Y" — the same move in a wig. So every rewrite gets re-scanned, and the loop runs until the scan is clean.
+
+Workflow: **Scan → Diagnose → Rewrite by meaning → Re-scan → (repeat) → Register check.**
+
+## Phase 0: Fix the target
+
+Before touching the text, establish:
+
+- **Genre and venue** — academic article, tweet, reddit post, LinkedIn, email, blog, docs, marketing. If not stated and not obvious from the text, ask. Genre decides which tells are fatal and what "good" means; see [references/voices.md](references/voices.md).
+- **Audience and stance** — who reads it, and what the author actually claims. Slop is what fills the space where a claim should be; you cannot remove it without knowing the claim.
+- **Constraints** — length limits, required citations, house style.
+
+## Phase 1: Mechanical scan
+
+Run the detection patterns from [references/tells.md](references/tells.md) against the text. If the text is in a file (or you can write it to a temp file), run the grep commands in that reference literally — the catalog is written as runnable `grep -Ein` patterns. Otherwise apply each pattern by hand, line by line.
+
+Produce a finding list: line/sentence, matched pattern, tell category. Also run the two structural checks that regex can't fully catch:
+
+- **Cadence**: flag any run of 3+ consecutive sentences within ±4 words of the same length, and any paragraph where every sentence has the same shape (subject–verb–elaboration).
+- **Formatting**: bold scattered through prose, emoji-decorated headers or bullets, "**Term:** definition" bullet lists, headers on a text too short to need them, a tidy intro–three-points–conclusion skeleton.
+
+Report the findings to the user as a short table before rewriting (category, count, worst example). This is the diagnosis; the user should see what was wrong.
+
+## Phase 2: Rewrite by meaning, not by frame
+
+Go finding by finding. The cardinal rule: **never fix a pattern by paraphrasing the pattern.** Fix it by deciding what the sentence actually asserts, then asserting that.
+
+### The "not X but Y" family — three-way triage
+
+Every negative parallelism gets exactly one of these treatments:
+
+1. **The negation is a strawman** (nobody believes X). Delete the X half entirely and assert Y directly, with whatever evidence the text has.
+   - *"It's not just a tool, it's a fundamental shift in how teams work"* → *"Teams that adopted it stopped holding standups within a month."*
+2. **The contrast is real** (people genuinely hold X). Then earn it: name who holds X, say concretely why Y beats it. A real contrast survives being made specific; slop doesn't.
+3. **The sentence asserts nothing** (the contrast is decoration on an empty claim). Delete the whole sentence. Most cases are this one.
+
+Banned escape hatches — these are the same move and count as new findings: "less about X than Y", "X matters, but Y matters more", "the real X is Y", "the question isn't X, it's Y", "X? Y." (rhetorical-question variant), and the em-dash variant "— not X, but Y".
+
+### Everything else
+
+- **Puffery and inflated vocabulary** (pivotal, seismic, testament, tapestry, landscape, delve…): replace with the plain word, or with the concrete fact the puffery was hiding. "Plays a vital role in" → "does".
+- **Rule-of-three lists**: keep the strongest item, cut the rest — unless all three carry distinct information, in which case keep them and break the rhythm (different lengths, different syntax).
+- **False ranges** ("from X to Y"): if you can't name a meaningful midpoint between X and Y, it's not a range — name the two things or cut one.
+- **Hedged both-sidesing** ("it's worth noting", auto-counterpoints, "while X, it's also true that Y"): commit. One opinion, stated, owned. A counterpoint stays only if the author genuinely concedes it.
+- **Uniform cadence**: vary deliberately. Follow a long sentence with a short one. Fragments are legal. Don't apply a formula (alternating long/short is its own tell) — read the paragraph aloud and break wherever the rhythm is metronomic.
+- **Low specificity**: replace "many companies" / "studies show" / "recent research" with the actual names, numbers, and dates — **only from the source text, the conversation, or verifiable research you actually do**. Never invent specifics. If the author needs to supply one, leave a marked placeholder: `[ADD: which study?]`.
+- **Stock skeleton**: kill throat-clearing openers ("In today's fast-paced world…"), summary conclusions ("In conclusion… Ultimately…"), and engagement-bait endings ("What do you think?"). Start where the point starts; stop when it's made.
+
+### What not to do — overcorrection is also slop
+
+- No fake typos, forced slang, or manufactured "voice". Humanizer-tool output is its own genre of slop.
+- Em dashes are not banned. Humans use them. The tell is density and the double-dash "— not X, but —" move. Budget: at most one em dash per ~150 words, never two in a sentence.
+- Don't trade precision for personality in academic or technical text. There, de-slopping means cutting puffery and committing to claims — not adding attitude.
+- Preserve the author's meaning, claims, and facts exactly. This is a style pass, not a content edit. Flag, don't silently fix, anything that looks factually wrong.
+
+## Phase 3: Verify loop
+
+Re-run the full Phase 1 scan **on your rewritten text**. This step is not optional and not a formality — expect your own rewrite to contain new tells, because the model writing it has the same priors that created them. Fix and re-scan until a pass produces zero pattern hits and the cadence check passes. Cap at 4 passes; if a pattern survives 4 passes, rewrite that sentence from scratch starting from its bare claim ("what fact or opinion is this sentence for?").
+
+## Phase 4: Register check
+
+Check the clean text against its genre profile in [references/voices.md](references/voices.md): right length, right formality, right person, genre-specific tells gone (e.g. on reddit: no bold, no bullet essay; in academic prose: no first-person hot takes added). Then the final test — read it aloud. Anywhere you wouldn't say it to the actual audience, rewrite that sentence.
+
+Deliver: the rewritten text, plus a brief change log (categories fixed, counts, and number of verify passes it took).

.agents/skills/fuck-slop/references/tells.md

@@ -0,0 +1,171 @@
+# AI-Writing Tell Catalog
+
+Detection patterns for the F*ck Slop scan. Patterns are written for `grep -Ein` (extended regex, case-insensitive, line numbers) so they can be run literally against a file:
+
+```bash
+grep -Ein -f /dev/stdin draft.txt <<'PATTERNS'
+<paste patterns from a section below, one per line>
+PATTERNS
+```
+
+When the text only exists in conversation, apply each pattern by hand. A match is a *finding*, not an automatic deletion — every finding goes through the Phase 2 triage in SKILL.md. Density matters: one em dash is nothing; one em dash plus a negative parallelism plus "delve" in the same paragraph is a verdict.
+
+## 1. Negative parallelism — the "not X but Y" family
+
+The highest-priority category. LLMs reach for the negation-then-assertion move roughly once a paragraph; humans use it occasionally and deliberately. It is an emergent generative habit, so expect it to reappear in paraphrased form after every rewrite pass — that is why the scan loops.
+
+```
+not (just|only|merely|simply|solely) [^.;]{2,80}(but|it'?s| — )
+isn'?t (just|only|merely|simply|about) 
+it'?s not (a|an|the|that|about|just) [^.;]{2,80}(it'?s|but)
+(is|was|are|were)n'?t about [^.;]{2,60}\. (it|this|that)'?s about
+less about [^.;]{2,60}(than|and more about)
+more than (just|a mere|simply) 
+not because [^.;]{2,80}but because
+the (question|point|issue|problem|goal|real [a-z]+) is(n'?t| not) (whether|about|just|if)
+(doesn'?t|don'?t|didn'?t|won'?t) (just|merely|simply) [^.;]{2,80}(it|they|he|she|we) 
+no [a-z]+, no [a-z]+(, no [a-z]+)?[,.]? just 
+— not [^—.;]{2,60}, but 
+not only [^.;]{2,80}but (also )?
+we'?re not (just )?(talking about|looking at|dealing with)
+gone are the days
+(here|this)'?s the (thing|kicker|catch|twist)
+```
+
+Rhetorical-question variant (regex-resistant; check by hand): a one-line question immediately answered by a one-word or one-clause sentence. *"The result? Chaos."* / *"Sound familiar?"*
+
+## 2. Puffery and inflated vocabulary
+
+Single words that spike in LLM output. Each is fine in isolation; two or more per page is a finding. The fix is the plain word or the concrete fact the word was hiding.
+
+```
+\b(delve|delving)\b
+\btapestry\b
+\b(testament|stands as)\b
+\bseamless(ly)?\b
+\b(pivotal|paramount|crucial)\b
+\bunderscore(s|d)?\b
+\b(landscape|realm|sphere) of\b
+\bnavigat(e|ing) the\b
+\bfoster(s|ing)?\b
+\bleverage(s|d)?\b
+\bmeticulous(ly)?\b
+\bintricate\b
+\bboasts\b
+\bgame.?chang(er|ing)\b
+\b(seismic|monumental|transformative) (shift|change)\b
+\bunwavering\b
+\bcommendable\b
+\belevate(s|d)? (the|your)\b
+\bshowcas(e|es|ing)\b
+\bresonate(s|d)?\b
+\bcompelling\b
+\brich (cultural )?(heritage|history|tradition)\b
+\bvibrant\b
+\bplays? a (vital|key|crucial|pivotal) role\b
+\bdeep(er)? dive\b
+\bunlock(s|ing)? (the|your)\b
+\bharness(es|ing)? the\b
+\bembark(s|ed|ing)? on\b
+\bever.?(evolving|changing)\b
+\bfast.?paced (world|environment)\b
+\bin today'?s\b
+\bat the end of the day\b
+\bwhen it comes to\b
+\bcutting.?edge\b
+\brobust\b
+\bholistic\b
+\bsynergy\b
+\bempower(s|ing|ment)?\b
+```
+
+## 3. Hedging, both-sidesing, throat-clearing
+
+The tell is reflexive balance: every claim gets a softener, every opinion gets a counterpoint. Commit or cut.
+
+```
+it'?s (worth|important) (to note|noting|to remember|to consider)
+(that|it) (being )?said,
+while (it'?s|this is) (true|important)
+arguably
+in many ways
+to some (extent|degree)
+on the other hand
+at its core
+in essence
+essentially,
+ultimately,
+in conclusion
+in summary
+to sum(marize| up)
+overall,
+in the end,
+needless to say
+as (we|you) (can see|know|all know)
+let'?s (dive|unpack|explore|take a (look|closer look))
+whether you('re| are) [^.;]{2,60} or 
+```
+
+## 4. False ranges and rule-of-three
+
+**False range** — a "from X to Y" with no actual spectrum between X and Y:
+
+```
+from [^.;]{3,50} to [^.;]{3,50}
+```
+
+Triage by hand: if you can name a meaningful midpoint, it's a real range and stays. If X and Y are just two loosely related examples, name them plainly or cut one.
+
+**Rule of three** — LLMs default to triplets to make thin analysis look thorough. Regex only catches the simplest shape; check lists by hand too.
+
+```
+\b\w+, \w+, and \w+[.!?]
+\b(\w+ \w+), (\w+ \w+), and (\w+ \w+)
+```
+
+Triage: keep the strongest item, cut the rest — or keep all three only if each carries distinct information, and then break the rhythm.
+
+## 5. Punctuation and formatting
+
+Em dash: not banned — humans use it. Findings are about **density** and the contrast move:
+
+- More than ~1 em dash per 150 words.
+- Two em dashes in one sentence.
+- `— not X, but Y` (already in section 1).
+- Em dash used for punchy emphasis where a comma works: `[a-z] — [a-z][^—]{1,25}\.$`
+
+Other formatting tells (check by hand; most regexes here are layout-dependent):
+
+- **Bold scattered through prose** like a textbook highlighting itself: `\*\*[^*]{2,40}\*\*` appearing more than ~once per 3 paragraphs of body prose.
+- **"Term: definition" bullets**: `^[-*] +\*\*[^*]+:?\*\*:? ` — the signature LLM list shape.
+- **Emoji headers/bullets** (🚀, ✅, 💡): needs PCRE, not `-E` — `LC_ALL=C.UTF-8 grep -Pn '^\s*[-*#]+\s.*[\x{1F300}-\x{1FAFF}\x{2600}-\x{27BF}]' draft.txt`.
+- **Headers on short texts** — section headers on anything under ~400 words.
+- **The tidy skeleton** — intro that previews three points, three matched sections, conclusion that restates them. Resolves too neatly; real writing has loose ends.
+- **Numbered lists where a paragraph would do.**
+- Curly quotes/apostrophes in a context where the author types straight ones (mixed within one text is the stronger tell).
+
+## 6. Cadence and statistical shape
+
+No regex; measure or eyeball.
+
+- **Uniform sentence length** (the single strongest current tell): a run of 3+ consecutive sentences within ±4 words of each other, paragraph after paragraph of 18–24-word sentences. Quick measurement on a file:
+
+  ```bash
+  tr '\n' ' ' < draft.txt | sed 's/[.!?] /\n/g' | awk '{print NF}'
+  ```
+
+  Human prose mixes 4-word sentences with 30-word sentences. Variance should be obvious at a glance.
+- **Uniform sentence shape**: every sentence opens subject-first; no fragments, no questions, no inversions.
+- **Uniform paragraph length**: every paragraph 3–4 sentences.
+- **Low specificity**: "many companies", "studies show", "experts agree", "recent research", "various factors" — generic where a human who knew the material would name names, numbers, dates. (Fix only with real specifics; never invented ones.)
+- **No friction**: nothing colloquial, no aside, no opinion held without a softener, nothing that risks being disagreed with.
+
+## 7. Genre-specific instant tells
+
+Covered in detail in [voices.md](voices.md); the headline items:
+
+- **Reddit/forums**: bold mid-comment, bullet-pointed comments, "Hope this helps!", perfectly balanced takes.
+- **Tweets/X**: "🧵", "Let that sink in", line-broken one-clause-per-line cadence, ending on a question to drive engagement.
+- **LinkedIn**: one-sentence paragraphs stacked vertically, "Agree?", the not-X-but-Y move (its natural habitat).
+- **Academic**: "delve", "novel insights", puffed significance claims ("crucial implications for the field"), citation-free superlatives.
+- **Email**: "I hope this email finds you well", restating the recipient's question back at them, three-paragraph symmetry for a one-line answer.