zero-slop 2.8.2 → 2.8.4

package/README.md

@@ -5,10 +5,10 @@
   <img alt="tests" src="https://img.shields.io/badge/tests-passing-227B5B">
   <img alt="dependencies" src="https://img.shields.io/badge/runtime%20dependencies-0-227B5B">
   <img alt="privacy" src="https://img.shields.io/badge/learning-private-227B5B">
-  <img alt="version" src="https://img.shields.io/badge/version-2.8.2-72528F">
+  <img alt="version" src="https://img.shields.io/badge/version-2.8.4-72528F">
   <a href="https://hol.org/guard/plugins"><img alt="Listed in the HOL plugin registry" src="https://img.shields.io/badge/HOL%20registry-listed-2C6E8F"></a>
   <a href="https://github.com/hashgraph-online/awesome-ai-plugins#tools--integrations"><img alt="Listed in awesome-ai-plugins" src="https://img.shields.io/badge/awesome--ai--plugins-listed-2C6E8F"></a>
-  <a href="https://zero-slop.ai/try/"><img alt="This README scores 12.0 out of 100 on the slop score, inside the human range" src="https://img.shields.io/badge/slop%20score-12.0%2F100-0f7d55"></a>
+  <a href="https://zero-slop.ai/try/"><img alt="This README scores 11.6 out of 100 on the slop score, inside the human range" src="https://img.shields.io/badge/slop%20score-11.6%2F100-0f7d55"></a>
 </p>
 
 Less slop, more pop in your writing.
@@ -114,7 +114,7 @@ The badge above is this README's own, from `npx zero-slop score README.md`.
 Put yours up the same way:
 
 ```md
-[![Slop score](https://img.shields.io/badge/slop%20score-12.0%2F100-0f7d55)](https://zero-slop.ai/try/)
+[![Slop score](https://img.shields.io/badge/slop%20score-11.6%2F100-0f7d55)](https://zero-slop.ai/try/)
 ```
 
 Colours are the scorer's bands: `0f7d55` under 25, `b8860b` under 60, `b0502c`
@@ -122,7 +122,7 @@ above.
 
 ## The slop that Zero Slop catches
 
-290 weighted patterns and a 96-term lexicon, including:
+294 weighted patterns and a 96-term lexicon, including:
 
 1. Binary contrasts: "It's not X. It's Y."
 2. Throat-clearing openers: "Here's the thing," "Let me be clear"
@@ -228,33 +228,35 @@ proof of general accuracy.
 
 ### Speed
 
-One busy Apple silicon Mac. Meter: 1,000 documents in 2.5035 seconds (399.4 per
-second), 15,201 words in 0.4301 seconds, worst stress case 3.2518 seconds. Reading
-pass, untimed on record until now: 0.8304 seconds for the same 1,000 (1,204.2 per
-second), 0.1359 for the same large document, linear to 96,000 words. Learning pass,
-8,000 words: 0.2027 seconds. Across 24 interleaved runs against 2.7.7 we measured
-0.08% lower median throughput; that is machine noise, not a speed claim. Editing
+One busy Apple silicon Mac. Meter: 1,000 documents in 1.9537 seconds (511.8 per
+second), 15,201 words in 0.3179 seconds, worst stress case 2.4137 seconds.
+Reading pass: 0.7741 seconds for the same 1,000 (1291.9 per
+second), 0.1051 for the same large document, linear to 96,000 words. Learning
+pass, 8,000 words: 0.1597 seconds. Across 24 interleaved runs against 2.7.7 we measured
+1.29% lower median throughput; that is machine noise, not a speed claim. Editing
 time is excluded.
 
 ### Reading-pass accuracy
 
 The reading pass budgets antithesis pairs by frequency, so the count has to be right
-before the budget means anything. It had never been measured. On 58 labelled pairs in
+before the budget means anything. On 75 labelled pairs in
 [`bench/antithesis/`](bench/antithesis/):
 
 | Reading pass | 2.8.0 | now |
 |---|---:|---:|
-| Recall, all shapes | 40.0% | 90.0% |
+| Recall, all shapes | 40.0% | 91.2% |
 | Recall, shapes in reach | 44.4% | 100% |
 | Precision | 80.0% | 100% |
 | False positives | 3 | 0 |
 
-![Antithesis detection before and after, on 58 labelled pairs](assets/bench-antithesis.png)
+2.8.3 added the families the 58-pair corpus never tested, where 2.8.2 scores 67.5%
+precision.
 
-Bare subject swap and the weak isocolon stay out of reach and count against recall: the
-first is a judgment call in `references/tells.md`, the second is identical to ordinary
-parallel prose on every lexical statistic. Maintainer labels on constructed pairs, so
-this is a regression floor, not field accuracy.
+![Antithesis detection before and after, on 75 labelled pairs](assets/bench-antithesis.png)
+
+Bare subject swap and the weak isocolon stay out of reach and count against recall: both
+are identical to ordinary prose on every lexical statistic. Maintainer labels on
+constructed pairs, so this is a regression floor, not field accuracy.
 
 ### Current models
 

package/SKILL.md

@@ -2,7 +2,7 @@
 name: zero-slop
 license: MIT
 metadata:
-  version: "2.8.2"
+  version: "2.8.4"
   author: manavmishra
 description: Turn drafts into sharp, natural prose or inspect them without rewriting. Zero Slop runs inside the user's existing AI assistant; Claude, GPT, or another compatible model reads and edits in context while local tools point to exact phrases and protect the source. Use when the user asks to humanize or de-slop writing, inspect AI-sounding patterns, fix text that reads like ChatGPT, polish outward-facing prose, draft social or LinkedIn content, or apply a final quality check to prose the agent generated. The workflow preserves facts, voice, and format and learns privately from repeated, reason-labelled human edits.
 ---
@@ -91,8 +91,8 @@ local and AI responsibilities distinct:
    and directly fix stumbles, repetition, weak transitions, and awkward flow.
 7. **Verifier — local tools plus the AI assistant.** Check the exact final text
    against the source for the writing score, facts, meaning, qualifiers, voice,
-   format, and structure. Any repair returns through roles 5 and 6 before role 7
-   runs again.
+   format, and structure. This is a release gate, not advice: a warning blocks
+   delivery. Any repair returns through roles 5 and 6 before role 7 runs again.
 8. **Fresh-eyes finalizer — a new AI pass.** Read the verified text as a first-time
    reader, apply only safe final polish, and approve it without changes. A role 8
    edit restarts roles 5 through 8; the finalizer never bypasses verification.
@@ -192,7 +192,7 @@ Run the heuristic surface scorer on the draft:
 python3 <skill-root>/scripts/slopscore.py --explain <file>   # any cwd; or pipe via stdin
 ```
 
-Every channel runs on every draft: the pattern meter (290 weighted tells plus
+Every channel runs on every draft: the pattern meter (294 weighted tells plus
 a 96-term lexicon and 26 context-gated riders), rhythm and burstiness,
 long-form word variety, followability, formatting
 densities, and register. Each one is interpretable: pattern-meter hits come
@@ -933,7 +933,7 @@ the AI model already running in the assistant or rewrite this `SKILL.md`.
 - `references/tells.md` — the master taxonomy (113 tells, 6 families) with fixes.
   It is the human-readable catalogue; `data/patterns.json` is its machine
   implementation. Together with the reviewed shared overlay, the current
-  release carries 290 weighted regexes because some tells need more than one.
+  release carries 294 weighted regexes because some tells need more than one.
 - `references/rewrite-moves.md` — the positive program: the six ladder rungs
   expanded, with before/after pairs and voice calibration.
 - `references/platforms.md` — LinkedIn, X/Twitter, email, blog, newsletter,

package/data/corpus/performed-register/README.md

@@ -1,6 +1,6 @@
 # performed-register — should-flag corpus
 
-Thirty-five spans a human editor flagged as slop on sight while the scorer
+Thirty-six spans a human editor flagged as slop on sight while the scorer
 passed every one clean (9.5–10.9, zero pattern hits). All come from
 zero-slop.ai blog drafts written in August 2026 — first-party text, so no
 external registry entry applies. They share one register: **performed-writer

package/data/corpus/performed-register/mechanical/social-stage-stack.txt

@@ -0,0 +1,3 @@
+The channel feels less like a workshop and more like a stage. I know I'm not
+alone. Not in a planning meeting. Not in a team review. But in a private group.
+That is where the real conversations are starting to happen.

package/data/learned-log.md

@@ -390,3 +390,14 @@ example that motivated it.
   alone: paragraph-order dependence, unsupported novelty, self-labelled
   significance, moral-adjective category errors, recap-flattery, and
   wall-of-text replies.
+- 2026-09-03 — v2.8.4 social-stage false negative. A long promotional social
+  post scored 15.0 because its only machine-readable signal was one candour
+  phrase; repeated validation, staged negative fragments, a less-like/more-like
+  comparison, and a generic "real conversations" close were left entirely to
+  the contextual reader. Added four narrow anchors whose combined weight finds
+  that stack without making any one ordinary phrase decisive. Also fixed the
+  reading pass so curly contractions, irregular past-tense verbs, and Unicode
+  dash bullets are not misreported as verbless prose. The reproduced miss moves
+  from 15.0 to 50.2, the edited version remains below the 25-point line, every
+  known-human control remains below it, and the exact finishing pipeline now
+  treats a verifier warning as a blocked release rather than advisory copy.

package/data/patterns.json

@@ -162,6 +162,15 @@
    "first_seen": "2026-08-03",
    "last_confirmed": "2026-08-15"
   },
+  {
+   "name": "social-validation",
+   "cat": "performed",
+   "rx": "\\b(?:i (?:know|realize) i(?:'|\u2019)m not alone|you(?:'|\u2019)re not (?:alone|imagining it)|if you(?:'|\u2019)ve (?:felt|noticed) [\\w '\u2019-]{2,36}, you(?:'|\u2019)re not (?:alone|imagining it))\\b",
+   "hints": ["not alone", "not imagining it"],
+   "w": 2.5,
+   "first_seen": "2026-09-03",
+   "last_confirmed": "2026-09-03"
+  },
   {
    "name": "faux-insight",
    "cat": "performed",
@@ -226,6 +235,15 @@
    "first_seen": "2026-08-03",
    "last_confirmed": "2026-08-03"
   },
+  {
+   "name": "staged-negative-ladder",
+   "cat": "contrast",
+   "rx": "(?:^|[.!?]\\s+|\\n)[\u2013\u2014*+\u2022-]*\\s*not in (?:an? |the )?[\\w '\u2019-]{2,32}\\.\\s*(?:[\u2013\u2014*+\u2022-]+\\s*)?not in (?:an? |the )?[\\w '\u2019-]{2,32}\\.\\s*(?:[\u2013\u2014*+\u2022-]+\\s*)?but in (?:an? |the )?[\\w '\u2019-]{2,48}",
+   "hints": ["not in a", "not in the"],
+   "w": 5,
+   "first_seen": "2026-09-03",
+   "last_confirmed": "2026-09-03"
+  },
   {
    "name": "this-isnt-about",
    "cat": "contrast",
@@ -515,6 +533,15 @@
    "first_seen": "2026-08-14",
    "last_confirmed": "2026-08-14"
   },
+  {
+   "name": "less-like-more-like",
+   "cat": "contrast",
+   "rx": "\\b(?:feel(?:s|ing)?|seem(?:s|ing)?) less like (?:an? |the )?[\\w '\u2019-]{2,48}(?:,| and)? (?:an? )?(?:little )?more like (?:an? |the )?[\\w '\u2019-]{2,48}",
+   "hints": ["less like", "more like"],
+   "w": 4,
+   "first_seen": "2026-09-03",
+   "last_confirmed": "2026-09-03"
+  },
   {
    "name": "thats-not-thats",
    "cat": "contrast",
@@ -1177,6 +1204,15 @@
    "first_seen": "2026-08-14",
    "last_confirmed": "2026-08-14"
   },
+  {
+   "name": "real-conversation-closer",
+   "cat": "performed",
+   "rx": "\\bwhere the (?:real|genuine|meaningful) (?:conversation|conversations|connection|connections) (?:is|are) (?:starting to |finally )?(?:happen|happening|taking place)\\b",
+   "hints": ["real conversation", "genuine conversation", "meaningful conversation"],
+   "w": 3,
+   "first_seen": "2026-09-03",
+   "last_confirmed": "2026-09-03"
+  },
   {
    "name": "recipe-for",
    "cat": "cliche",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "zero-slop",
-  "version": "2.8.2",
+  "version": "2.8.4",
   "description": "An Agent Skill that scores AI-sounding prose locally, guides the host AI through an editorial rewrite, and checks changed source details. The local tools run offline with zero dependencies.",
   "bin": {
     "zero-slop": "bin/zero-slop.mjs"
@@ -47,6 +47,7 @@
     "!scripts/build_plugin.py",
     "!scripts/build_skill_zip.py",
     "!scripts/check_svg.py",
+    "!scripts/check_release_surfaces.py",
     "!scripts/contextual.py",
     "!scripts/contextual-signals.md",
     "!references/contextual-signals.md",

package/scripts/register.py

@@ -95,9 +95,13 @@ NEGATION = re.compile(r"\b(?:not|never|cannot)\b|\w+n['’]t\b", re.I)
 # figure's usual carrier and is absent from ordinary negated prose ("The server
 # did not respond. We restarted it.").
 COPULA = re.compile(r"\b(?:is|are|was|were|be|been|am)\b|\w+['’]s\b|\w+n['’]t\b", re.I)
-RX_MARKED_OPEN = re.compile(r"^\W*not\b", re.I)
+RX_MARKED_OPEN = re.compile(
+    r"^\W*not\s+(?!all\b|every\b|only\b|just\b|much\b|many\b|most\b|enough\b"
+    r"|yet\b|quite\b|nearly\b|entirely\b|\w+ly\b)", re.I)
 RX_MARKED_CLOSE = re.compile(
-    r"\b(?:was|were|is|are|did|does|do|has|have|had|will|can|could|would)\s+not\W*$", re.I)
+    r"\b(?:was|were|is|are|did|does|do|has|have|had|will|can|could|would)\s+not\W*$"
+    r"|\b(?:wasn|weren|isn|aren|didn|doesn|don|hasn|haven|hadn|won|can|couldn|wouldn)"
+    r"['\u2019]t\W*$", re.I)
 # The two template shapes the meter already anchors (this-is-what-looks-like,
 # no-x-had-to). The meter scores them as spans; the register pass has to COUNT
 # them, because tells.md budgets the family by frequency and a span hit is not
@@ -138,7 +142,20 @@ def _common_prefix(a: list[str], b: list[str]) -> int:
 # splitter or prose_of -- both are shared by every family, and moving either one
 # pushed four to eleven documents over budget on rate alone -- this family
 # rejects the glued span itself. Nothing else sees the change.
-RX_SCAFFOLD = re.compile(r"^\s*(?:[#>|]|[-*+]\s|\d+\.\s)|\*\*|\n\s*(?:[#>|]|[-*+]\s|\d+\.\s)")
+RX_SCAFFOLD = re.compile(r"^\s*(?:[#>|]|[-*+]\s|\d+\.\s|\*\*)|\n\s*(?:[#>|]|[-*+]\s|\d+\.\s|\*\*)")
+
+
+# Cardinals and calendar words. Their presence in the part of a pair that
+# differs marks a specification rather than a rhetorical figure.
+ENUMERATED_VALUE = frozenset("""
+one two three four five six seven eight nine ten eleven twelve twenty thirty
+forty fifty sixty seventy eighty ninety hundred thousand million billion
+first second third fourth fifth
+monday tuesday wednesday thursday friday saturday sunday
+january february march april may june july august september october november
+december hour hours minute minutes day days week weeks month months year years
+am pm noon midnight
+""".split())
 
 
 def antithesis_pairs(prose: str) -> list[str]:
@@ -189,7 +206,24 @@ def antithesis_pairs(prose: str) -> list[str]:
         # role never certifies its own output." shares "role" and is ordinary
         # prose. The frame is a repeated opening, a copula on both sides, or two
         # words in common.
-        if negated and na <= 10 and nb <= 8 and (prefix >= 1 or len(shared) >= 2):
+        # `prefix >= 1` counted a shared stopword as a frame, so "It does not
+        # run on Windows. It runs on Linux and macOS." qualified on "it" alone.
+        # Requiring a CONTENT frame instead was too strict: "We do not guess. We
+        # measure." shares only "we" and is the figure.
+        #
+        # What separates them is the payoff, not the frame. The figure lands its
+        # twist in a breath -- "We measure." "She proved it." -- while ordinary
+        # negated prose just carries on at normal length. So a stopword frame is
+        # allowed, but only when the second half is that short.
+        content_prefix = any(w not in ANTITHESIS_STOP for w in ta[:prefix])
+        shared_content = {w for w in shared if w not in ANTITHESIS_STOP}
+        # A repeated CONTENT opening is anaphora -- "The report ... The report
+        # ..." -- which the isocolon branch below already refuses for exactly
+        # this reason. Here it was being read as a frame, i.e. as evidence for
+        # the figure rather than against it.
+        strong_frame = len(shared_content) >= 2 and not content_prefix
+        weak_frame = prefix >= 1 or len(shared) >= 2
+        if negated and na <= 10 and ((strong_frame and nb <= 8) or (weak_frame and nb <= 5)):
             out.append(f"{first} {second}"); consumed = i + 1; continue
         # A copula on both sides is the weakest of the three frames, so it only
         # counts when the halves are staccato-short. "Passwords are never stored
@@ -217,6 +251,20 @@ def antithesis_pairs(prose: str) -> list[str]:
         shared_open = any(w not in ANTITHESIS_STOP for w in ta[:prefix])
         if shared_open or head == tail or len(head) < 2 or len(tail) < 2:
             continue
+        # Enumeration wears the same clothes as isocolon: one frame, both
+        # arguments swapped, the same word overlap. "The free tier includes ten
+        # seats. The pro tier includes fifty seats." is structurally identical
+        # to "A junior engineer reads the error. A senior engineer reads the
+        # stack trace." -- four shared words and 0.67 overlap in both.
+        #
+        # What separates them is what varies. A specification varies a VALUE:
+        # a quantity, a weekday, a time. The figure varies a CONCEPT. So a
+        # cardinal or a calendar word in the part that differs means this is a
+        # table written as prose, and the pass stays quiet.
+        differing = set(head) | set(tail)
+        if differing & ENUMERATED_VALUE or any(
+                any(ch.isdigit() for ch in w) for w in differing):
+            continue
         if len(shared) >= 2 and len(shared) / min(len(head), len(tail)) >= 0.5:
             out.append(f"{first} {second}")
             consumed = i + 1
@@ -226,7 +274,7 @@ def antithesis_pairs(prose: str) -> list[str]:
 # "X, not Y." and "A rather than B." The corrective appositive. Each instance is
 # usually careful writing, which is why no pattern list contains it.
 RX_SUBTRACTIVE = re.compile(
-    r"[^.\n]{3,90}?,\s+not\s+[^.\n]{3,60}[.\n]"
+    r"[^.\n]{3,90}?[,\u2014\u2013]\s*(?:not|never)\s+[^.\n]{3,60}[.\n]"
     r"|[^.\n]{3,70}\brather than\b[^.\n]{3,50}[.\n]",
     re.I,
 )
@@ -332,6 +380,26 @@ FINITE_VERB = re.compile(
     r"|gate[sd]?|gives?|runs?|gets?|gave|gone)\b"
     r"|\b\w+(?:s|ed|es)\b", re.I)
 
+# Curly apostrophes are ordinary prose, not token boundaries. A contracted
+# auxiliary makes the clause finite even when the main verb is irregular or a
+# participle ("I’ve found", "you’re seeing"). The old tokeniser split those
+# into ``I``/``ve`` and ``you``/``re`` and then reported complete sentences as
+# fragments. Keep this conservative: a possible possessive false negative is
+# safer than telling a writer that a grammatical sentence has no verb.
+CONTRACTED_FINITE = re.compile(r"\b[A-Za-z][\w-]*['’](?:m|re|ve|s|d|ll)\b", re.I)
+# Only forms that are rarely ordinary nouns or adjectives belong here. Ambiguous
+# forms such as ``cut``, ``left``, ``read``, ``set``, and ``thought`` made real
+# fragments look grammatical ("Nothing left in the queue"). Contractions are
+# handled separately above, so this small list only closes clear past-tense
+# holes in uncontracted sentences.
+IRREGULAR_FINITE = frozenset("""
+arose became began blew broke brought bought caught chose came dealt did drew
+drank drove ate fell fed felt fought found flew forgot forgave froze gave got
+grew had heard held hid kept knew laid led lent lost made meant met paid rode
+ran said saw sold sent shook showed sang sat slept spoke spent stood stole
+struck swam took taught told threw understood woke wore won wrote was were went
+""".split())
+
 # The -s half of that catch-all cannot tell a verb from a plural noun, and a
 # plural noun in a fragment made the whole fragment invisible: "Same compound,
 # three identifiers." and "Same assay, two units." both read as verbed on
@@ -386,6 +454,11 @@ def _looks_like_noun(sent: str, start: int) -> bool:
 
 
 def _has_finite_verb(sent: str) -> bool:
+    if CONTRACTED_FINITE.search(sent):
+        return True
+    if any(word.lower() in IRREGULAR_FINITE
+           for word in re.findall(r"[A-Za-z][\w'’-]*", sent)):
+        return True
     for match in FINITE_VERB.finditer(sent):
         word = match.group(0)
         # An explicit auxiliary or an -ed form is a verb wherever it appears.
@@ -408,10 +481,10 @@ def _sentences(prose: str) -> list[str]:
 def verbless_fragments(prose: str) -> list[str]:
     out = []
     for sent in _sentences(prose):
-        words = re.findall(r"[A-Za-z][\w'-]*", sent)
+        words = re.findall(r"[A-Za-z][\w'’-]*", sent)
         if not (3 <= len(words) <= 12):
             continue
-        if sent.rstrip().endswith(":") or sent.lstrip().startswith(("-", "*", "#", "|")):
+        if sent.rstrip().endswith(":") or sent.lstrip().startswith(("-", "–", "—", "•", "*", "#", "|")):
             continue
         if IMPERATIVE_OPENER.match(sent):
             continue
@@ -514,7 +587,7 @@ def prose_of(text: str) -> str:
     return "\n".join(
         line
         for line in text.split("\n")
-        if not line.strip().startswith(("|", "![", ">", "    "))
+        if not line.strip().startswith(("|", "![", ">", "-", "–", "—", "•", "*", "    "))
     )
 
 

package/scripts/slopscore.py

@@ -900,7 +900,7 @@ def render_portfolio(result):
 # broetry is a slop tell, not a machine tell: LinkedIn writers invented it
 # years before GPT-3, and it demonstrably performs on the platform. Whether to
 # trade reach for a human voice is the author's call, not the meter's.
-STRUCT_MARK = re.compile(r"^\s*(?:[-*+•>#]|\d+[.)]|\|)")
+STRUCT_MARK = re.compile(r"^\s*(?:[-–—*+•>#]|\d+[.)]|\|)")
 DIALOGUE_OPEN = re.compile("^[\"“‘']")