coroboros/agent-skills / skills/humanize-en
humanize-en
VerifiedRemove AI writing tells from English prose while preserving meaning, structure, code, links and quotations. Use for humanize this, remove AI slop and polish the English on inline text or prose files. Optional -f BRAND-VOICE.md applies brand rules with explicit mechanical and semantic coverage. Inherit authorized rewrite scope; audit/propose remains read-only. Structural README work belongs to write-clear-readme.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 18.1 KB |
| eval-corpus/brand-voice/brand-allcaps-01.json | config | 1.2 KB |
| eval-corpus/brand-voice/brand-clean-01.json | config | 1.3 KB |
| eval-corpus/brand-voice/brand-lexicon-01.json | config | 911 B |
| eval-corpus/brand-voice/brand-pronouns-01.json | config | 978 B |
| eval-corpus/brand-voice/brand-rewrite-rules-01.json | config | 946 B |
| eval-corpus/brand-voice/brand-rhetoric-01.json | config | 1.1 KB |
| eval-corpus/samples/clean-prose-01.json | config | 417 B |
| eval-corpus/samples/docs-intro-01.json | config | 631 B |
| eval-corpus/samples/hero-landing-01.json | config | 637 B |
| eval-corpus/samples/release-notes-01.json | config | 571 B |
| evals/evals.json | config | 2.8 KB |
| references/output-formats.md | markdown | 1.7 KB |
| references/patterns.md | markdown | 18.3 KB |
| references/schemas.md | markdown | 10.7 KB |
| references/voice.md | markdown | 2.8 KB |
| scripts/__init__.py | script | 0 B |
| scripts/brand_prescan.py | script | 28.0 KB |
| scripts/eval_patterns.pyexecutable | script | 5.3 KB |
| scripts/prescan.pyexecutable | script | 11.0 KB |
| scripts/utils.py | script | 1.6 KB |
| scripts/validate.py | script | 7.0 KB |
scripts/eval_patterns.py
View on GitHub#!/usr/bin/env python3
"""
eval_patterns.py — score prescan against the eval corpus.
Usage:
eval_patterns.py # universal corpus (eval-corpus/samples/*.json)
eval_patterns.py --brand # brand corpus (eval-corpus/brand-voice/*.json)
eval_patterns.py --sample <path.json> # single sample (auto-detects brand)
eval_patterns.py --corpus <dir> # custom corpus directory
Runs the deterministic scan on each sample's `input` and compares the
detected patterns against `expected_hits`. Universal samples reference
patterns by integer ID (1-32); brand samples reference them by string slug
("brand:all_caps_emphasis", etc.) and may inline a `brand_voice_yaml` to
exercise brand-aware detection.
Exit codes:
0 all samples pass
1 at least one fails (under- or over-detection)
2 argument or I/O error
Requires Python 3.7+. No third-party dependencies.
"""
from __future__ import annotations
import argparse
import sys
import tempfile
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent))
from utils import read_json, write_json # noqa: E402
from prescan import scan as prescan_scan # noqa: E402
SCRIPT_DIR = Path(__file__).parent
SKILL_DIR = SCRIPT_DIR.parent
DEFAULT_CORPUS = SKILL_DIR / "eval-corpus" / "samples"
BRAND_CORPUS = SKILL_DIR / "eval-corpus" / "brand-voice"
def evaluate_sample(sample):
"""Run prescan on sample['input'] and compare against expected_hits.
When the sample carries `brand_voice_yaml`, also run brand-aware detection
and compare both universal pattern numbers (ints) and brand pattern slugs
(strings) against the expected set.
Returns a dict shaped per schemas.md § eval result. The `pass` field is
True when missing == [] AND extra == []."""
expected = sorted({h["pattern"] for h in sample.get("expected_hits", [])},
key=lambda x: (isinstance(x, str), x))
has_brand = bool(sample.get("brand_voice_yaml"))
detected_hits = prescan_scan(sample["input"], attach_source=has_brand)
if has_brand:
from brand_prescan import load_brand_rules, scan_brand
tmp_path = None
try:
with tempfile.NamedTemporaryFile(
mode="w", suffix=".md", delete=False, encoding="utf-8"
) as tmp:
tmp.write("---\n" + sample["brand_voice_yaml"] + "---\n")
tmp.write("\n# Test\n\n## 1. Core voice attributes\n\nstub\n")
tmp_path = tmp.name
rules = load_brand_rules(tmp_path)
detected_hits.extend(scan_brand(sample["input"], rules))
finally:
if tmp_path is not None:
try:
Path(tmp_path).unlink()
except FileNotFoundError:
pass
detected = sorted({h["pattern"] for h in detected_hits},
key=lambda x: (isinstance(x, str), x))
missing = [p for p in expected if p not in detected]
extra = [p for p in detected if p not in expected]
return {
"id": sample["id"],
"expected_patterns": expected,
"detected_patterns": detected,
"missing": missing,
"extra": extra,
"pass": not missing and not extra,
}
def discover_samples(corpus_dir):
"""Return a sorted list of sample file paths."""
return sorted(Path(corpus_dir).glob("*.json"))
def main():
parser = argparse.ArgumentParser(description=(__doc__ or "").split("\n\n")[0])
parser.add_argument("--sample", help="Single sample file to evaluate.")
parser.add_argument(
"--corpus",
default=None,
help=f"Corpus directory (default: {DEFAULT_CORPUS} or "
f"{BRAND_CORPUS} when --brand is set)",
)
parser.add_argument(
"--brand",
action="store_true",
help="Run the brand-voice corpus instead of the universal corpus. "
"Equivalent to --corpus eval-corpus/brand-voice.",
)
parser.add_argument(
"--out",
help="Write the JSON report to a file instead of stdout.",
)
args = parser.parse_args()
if args.corpus is None:
args.corpus = str(BRAND_CORPUS if args.brand else DEFAULT_CORPUS)
try:
if args.sample:
sample_paths = [Path(args.sample)]
else:
sample_paths = discover_samples(args.corpus)
if not sample_paths:
print(
f"error: no samples found in {args.corpus}",
file=sys.stderr,
)
return 2
except (OSError, ValueError) as exc:
print(f"error: {exc}", file=sys.stderr)
return 2
results = []
for path in sample_paths:
try:
sample = read_json(str(path))
results.append(evaluate_sample(sample))
except (OSError, ValueError, KeyError) as exc:
print(f"error reading {path}: {exc}", file=sys.stderr)
return 2
passed = sum(1 for r in results if r["pass"])
total = len(results)
report = {
"samples": results,
"summary": {
"total_samples": total,
"passed": passed,
"failed": total - passed,
"pass_rate": (passed / total) if total else 0.0,
},
}
write_json(report, path=args.out)
return 0 if passed == total else 1
if __name__ == "__main__":
sys.exit(main())
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |