Generate agent-ready VOICE.md packs and CI benchmarks for website copy.
Project description
site2voice
Turn any website into agent-ready copy context. Then test whether the agent followed it.
pipx install site2voice
site2voice init https://example.com
site2voice bench https://example.com draft.md --strict
No LLM dependency. No browser dependency. Just a small Python CLI that extracts measurable voice signals and turns them into files coding agents can use.
Why
Agents are getting better at building interfaces, but they still guess the words: headings, CTAs, navigation labels, claim boundaries, and product tone.
DESIGN.md gives agents visual taste. site2voice gives them copy taste.
Voice Packs
The repo includes downloadable voice packs for well-known product and editorial sites: Apple, Stripe, Linear, Vercel, Notion, Figma, Shopify, GitHub, OpenAI, Anthropic, Hypebeast, Highsnobiety, Monocle, Wallpaper, and EYESMAG.
curl -L https://raw.githubusercontent.com/SihyeonJeon/site2voice/main/packs/stripe/VOICE.md -o VOICE.md
See packs.
See also the Stripe pack comparison,
which shows the same prompt run with and without VOICE.md.
What You Get
.site2voice/
VOICE.md human-readable writing brief
voice.json machine-readable voice profile
agent-prompt.md drop-in instruction for Codex, Claude Code, Cursor, etc.
The profile includes:
- heading shape, sentence rhythm, paragraph length, CTA shape;
- repeated vocabulary and CTA verbs;
- an output contract with target ranges for generated copy;
- deterministic tone labels;
- unsupported-claim detection;
- copied-span detection.
Benchmark
site2voice bench scores candidate copy against a source voice profile and can
fail CI when the copy drifts.
site2voice bench examples/editorial-home.html examples/after-copy.md --strict
Example fixture:
| Candidate | Result | Overall | Lexicon | Copy safety |
|---|---|---|---|---|
after-copy |
PASS | 83.8 | 70.0 | 93.2 |
before-copy |
FAIL | 36.6 | 0.0 | 100.0 |
The score rewards measurable voice alignment without rewarding verbatim copying.
Commands
# Generate one Markdown brief
site2voice https://example.com --out VOICE.md
# Generate an agent-ready context pack
site2voice init https://example.com --dir .site2voice
# Write JSON for custom pipelines
site2voice https://example.com --format json --out voice.json
# Gate candidate copy in CI
site2voice bench https://example.com draft.md \
--fail-under 75 \
--min-copy-safety 85 \
--min-claim-safety 75
What It Is Not
- Not an official brand guideline.
- Not a DESIGN.md visual-token extractor.
- Not a crawler for private pages or authenticated apps.
- Not a tool for copying another site's prose.
Develop
python3 -m pip install -e .
make test
make bench-ci
Docs
- Benchmark
- Voice packs
- CI usage
- Agent workflows
- Voice patterns
- JSON schema
- Market position
- Research
- Source candidates
- Awesome eligibility
- Roadmap
License
MIT
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file site2voice-0.5.0.tar.gz.
File metadata
- Download URL: site2voice-0.5.0.tar.gz
- Upload date:
- Size: 55.7 kB
- Tags: Source
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
a611d5435fa60d069c30ccca112cf2cd468628ff8f8a30cdde6fedc48e270280
|
|
| MD5 |
2e67e8ebf611d7913ec5863f940ff0ff
|
|
| BLAKE2b-256 |
999ab6dfb6f839b6375e727b74c5ea89310b75f8de0f570dcb4f51630f6f76b0
|
Provenance
The following attestation bundles were made for site2voice-0.5.0.tar.gz:
Publisher:
publish.yml on SihyeonJeon/site2voice
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
site2voice-0.5.0.tar.gz -
Subject digest:
a611d5435fa60d069c30ccca112cf2cd468628ff8f8a30cdde6fedc48e270280 - Sigstore transparency entry: 1581067668
- Sigstore integration time:
-
Permalink:
SihyeonJeon/site2voice@6bfba352207cf72e0cfdc425969efc917b9cfaf4 -
Branch / Tag:
refs/tags/v0.5.0 - Owner: https://github.com/SihyeonJeon
-
Access:
public
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
publish.yml@6bfba352207cf72e0cfdc425969efc917b9cfaf4 -
Trigger Event:
release
-
Statement type:
File details
Details for the file site2voice-0.5.0-py3-none-any.whl.
File metadata
- Download URL: site2voice-0.5.0-py3-none-any.whl
- Upload date:
- Size: 15.8 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
1e9a76fdf4d47a86c8e24d70d6fb24d58945815ee0fc6b39f1d03af733423b4c
|
|
| MD5 |
b96184ba0cce0ce7d8890d5d01eae399
|
|
| BLAKE2b-256 |
79f38408c72790c15b3e41aac0e43092d3c80351677273228e2d54de2e7e614c
|
Provenance
The following attestation bundles were made for site2voice-0.5.0-py3-none-any.whl:
Publisher:
publish.yml on SihyeonJeon/site2voice
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
site2voice-0.5.0-py3-none-any.whl -
Subject digest:
1e9a76fdf4d47a86c8e24d70d6fb24d58945815ee0fc6b39f1d03af733423b4c - Sigstore transparency entry: 1581068035
- Sigstore integration time:
-
Permalink:
SihyeonJeon/site2voice@6bfba352207cf72e0cfdc425969efc917b9cfaf4 -
Branch / Tag:
refs/tags/v0.5.0 - Owner: https://github.com/SihyeonJeon
-
Access:
public
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
publish.yml@6bfba352207cf72e0cfdc425969efc917b9cfaf4 -
Trigger Event:
release
-
Statement type: