OfficialJev Cookbooks
Twitch Chat Filter Cookbook
A live chat filter that asks three questions per message — category, toxicity, and spoilers — then turns the answers into show, highlight, dim, hide, or blur in plain code. The 29-message example hides seven, blurs two, and highlights six.
- Format
- Docs
- Published by
- TypeSafe AI
- Last verified
- Oct 2026
- Languages
- Python
Highlights
- One request per message holds three questions: a five-way category Choice, a toxic Noul, and a spoiler Noul, judged against the stream title and game.
- decide() maps answers to show, highlight, dim, hide, or blur; changing a rule re-runs only decide() with no new request, as 14 changed messages in the example show.
- On 29 hand-written Elden Ring messages the filter hid 7, blurred 2 spoilers, and highlighted 6 questions; requests took a median of about 0.4 seconds.
- The demo is a Jev-powered Chrome extension filtering a live Twitch chat, and messages run eight at a time with ThreadPoolExecutor.
Quickstart
python
RULES = {"question": "highlight", "hype": "show", "conversation": "show",
"backseat": "hide", "spam": "hide"}
TOXIC_AT, SPOILER_AT = 0.7, 0.7
def decide(result: dict, rules: dict = RULES) -> str:
if result["toxic"] >= TOXIC_AT:
return "hide (toxic)"
action = rules[result["category"]]
if action == "hide":
return f"hide ({result['category']})"
if result["spoiler"] >= SPOILER_AT:
return "blur (spoiler)"
return actionWatch out
A message waits about 0.4 seconds for its answer, so a toxic line is visible until then, and the page says one request at a time cannot keep up with a busy chat. Jev was trained mainly on English at publication, and the spoiler check has no record of the streamer's progress. Numbers come from jev-1.13.0 on 2026-09-25.
Questions about this entry
Short answers drawn from the linked project or page.
One. The request carries all three questions — category, toxic, and spoiler — against a state holding the stream title, game, and the message, and the page runs up to eight of these requests concurrently.
The cookbook's examples have no bad words: 'go left, there's a treasure box behind the tree' is unrequested gameplay advice and a plot twist is a spoiler. The three questions catch both without keyword rules.
A Choice over question, hype, conversation, backseat, and spam; a Noul for whether the message is insulting, hateful, or harassing; and a Noul for whether it reveals a plot point or ending of the game being played.
decide() reads only the answers: at or above the 0.7 toxic threshold the message is hidden; a category whose rule is hide is hidden; at or above the 0.7 spoiler threshold it is blurred; everything else gets its category's rule, such as highlight for questions.
Yes. The answers describe the message and decide() turns them into actions, so swapping the rules dict re-runs only decide() — the example's quieter rules changed 14 of 29 messages with 0 new requests.
Each message waits about 0.4 seconds for its answer, one request at a time cannot keep up with a busy chat, Jev was trained mainly on English at publication, close probabilities near 0.5 need a policy, and the spoiler check does not know the streamer's progress.
Replace STREAM and CHAT, or pass a different stream to classify() per channel, and install with pip install 'typesafe-sdk>=0.7.1,<0.8.0' ipython 'cooksafe>=0.2.0,<0.3.0'. json_cache.json replays the published numbers without a key.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
YouTube: Building a contextual brand-safety scorer on Jev from scratch.
X: A website slop detector that checks 35 tells in 243 ms
X: Breaking, golden nugget, or slop
X: A filter for reply-guy comments
More Jev resources like this
A support-ticket cookbook that lets Jev decide which requests are worth remembering, then wraps each call so saved tickets ride along in the state. Six later messages route correctly with memory against two without.
Jev CookbooksDocs#official#cookbook#memory
Official
Screen every message going into and out of an LLM app: a Noul describes possible hazards, a Score rates severity, and code thresholds the probabilities.
Jev CookbooksDocs#official#cookbook#guardrails
Official
A contextual brand-suitability scorer that judges each feed item against an advertiser's own versioned policy, returning a category, an exclusion flag, and a confidence for routing borderline items to review.
Jev Integrations & Adapters#community#typescript#advertising
Community
Decide which of 450 candidate product pairs from two beer catalogues describe the same entity: one three-level Score per pair splits 40 merges, 50 curator reviews, and 360 non-matches.
Jev CookbooksDocs#official#cookbook#entity-matching
Official
A repeatability study: an 8-question moderation rubric run 15 times per condition shows a mean probability standard deviation of 0.0098, and a 0.60 uncertainty gate lifts agreement to 99.2%.
Jev CookbooksDocs#official#cookbook#consistency
Official
Classify documents through deep patent, retail product, biomedical, and source-code hierarchies using parallel beam search over Choice probabilities.
Jev CookbooksDocs#official#cookbook#classification
Official
From the Jev guides
Original write-ups on what this entry is for.
Retrieval is a sequence of bounded judgments. How a decision model slots into passage filtering, semantic find, reranking, and citation checks, and what stays in deterministic code.
Jev Cookbooks#retrieval#reranking#rag
Read article
Jev bills input tokens and gives output away, and questions in one request run in parallel. The design consequence is simple: ask more per call, then decide in code.
Jev Cookbooks#batching#patterns#search
Read article
From the Jev community
Posts from builders shipping with Jev right now.