SSE Stream Viewer

Kaccha streaming jawab paste kijiye aur text, tool calls aur usage wapas paiye.

Kacchi stream mein poora jawab hota hai. Wo browser mein padhi jaati hai aur kabhi upload nahi hoti.

Jis loop se ye bana uska kharch nikaaliye

SSE Stream Viewer ke baare mein

Jab koi streaming call bigadti hai to haath mein event aur data lines ki ek deewaar hoti hai, aur jo teen baatein aap jaanna chahte hain wo usi mein bikhri hoti hain: model ne asal mein kya kaha, wo kyun ruka, aur usne kitna kharch kiya. Text tukdon mein aata hai, tool ke arguments adhoore JSON ke roop mein jo aakhri tukda aane par hi padha jaata hai, aur usage pehli aur aakhri event ke beech banti hoti hai. Yahaan ye sab dobara joda jaata hai. Stream ke dono aam roop sambhaale jaate hain, kyunki vyavhaar mein log dono paste karte hain, aur jo events padhi nahi jaatin wo chupchaap chhodi nahi, bataayi jaati hain.

Features

• Deltas se text kram mein dobara joda gaya
• Adhoore JSON se tool ke arguments dobara banaye aur sajaye gaye
• Rukne ka kaaran, model aur usage nikali jaati hai
• Block-delta roop aur choices/delta roop dono
• Dohrayi gayi data lines maanak ke anusaar jodi jaati hain
• Jeevit rakhne wali comments event nahi gini jaatin
• Error events chhodi nahi, dikhayi jaati hain
• Har event kholi ja sakti hai, naam ke hisaab se ginti bhi

SSE Stream Viewer kaise use karein

1. Kacchi stream paste kijiye
2. Dobara juda text padhiye
3. Rukne ka kaaran aur usage dekhiye
4. Kuchh ab bhi samajh na aaye to alag-alag events kholiye

Example

Input

delta: {"type":"text_delta","text":"Order 4471 "}

Output

Order 4471 shipped on 30 August.

Chhah events mein bikhre tukde, kram mein dobara jode gaye.

Common errors aur troubleshooting

• Koi text nikalta hi nahi. — Ho sakta hai content aane se pehle hi stream khatam ho gayi — events ki list aur rukne ka kaaran dekhiye. Inkaar ya koi error event valid par khaali stream deti hai.
• Tool ke arguments adhoore kat gaye hain. — Wo kai events mein adhoore JSON ke roop mein aate hain. Stream beech mein kati ho to JSON kabhi band nahi hota, aur padhe hue ke bajaye kaccha tukda dikhta hai.
• Kuchh events padhi nahi jaatin. — Aam taur par copy karte samay event ke beech se kati stream. Aisi events chhodi nahi, apne kramaank ke saath bataayi jaati hain.
• Usage ke aankde adhoore lagte hain. — Wo stream mein bante hain: input tokens shuru mein, output tokens ant mein. Jis stream mein aakhri events nahi, usme aadhe nahi honge.

Aksar pooche jaane wale sawaal

server-sent events kya hai?

Ek saral streaming format: field aur value ki lines, events khaali lines se alag. Streaming API jawab yahi istemaal karte hain, isiliye bigdi call aapke haath mein yahi deewaar chhod jaati hai.

Stream ke kaun-se roop chalte hain?

Block wala roop, jahaan text content blocks ke bhitar type wale deltas ke roop mein aata hai, aur choices/delta roop. Dono apne aap pehchane jaate hain, kyunki log wahi paste karte hain jo unke client ne darj kiya.

Arguments tukdon mein kyun aate hain?

Wo adhoore JSON ke roop mein bheje jaate hain taaki client call poori hone se pehle dikhana shuru kar sake. Iska matlab ye ki tukde aakhri ke aane par hi padhe jaate hain, aur kati hui stream invalid JSON chhod jaati hai.

Kya kharch dikh sakta hai?

Usage ka block maujood ho to nikala jaata hai, cache readings samet. Wo batata hai ki call ne kya kharch kiya; uske aage kharch wale tools sambhaalte hain.

Kya stream kahin bheji jaati hai?

Nahi. Stream mein poora jawab hota hai aur aksar wo prompt bhi jisse wo bana, aur saara padhna browser mein hi hota hai.

Related tools

• Chat Transcript Converter — Messages array ko padhne layak transcript mein badliye, aur wapas.
• JSONL Viewer aur Validator — JSONL file line-dar-line padhiye, chhaantiye aur jaanchiye.
• Tool Schema Checker — Definition paste kijiye aur jaaniye wo kyun reject ho rahi hai.
• Agent Loop Cost Estimator — Turn-dar-turn kharch nikaaliye, dobara bheje history aur cache ke saath.
• Context Window Planner — Window ko prompt, tools, history aur jawab mein baantiye.
• MCP Tool Schema Builder — Tool tay kijiye aur JSON Schema paiye, strict mode jaanch ke saath.

AI aur agents ke aur tools

• Skill Builder — SKILL.md likhiye, frontmatter saath-saath check hota hai.
• Skill Linter — SKILL.md paste kijiye aur dekhiye use kisi kaam se judne se kya rok raha hai.
• Subagent Builder — Subagent ki definition file banaiye, uske tools aur model samet.
• Slash Command Builder — Arguments aur tools list ke saath apna command banaiye.
• Project Instructions Builder — Aisi CLAUDE.md likhiye jo har session mein apni jagah kamaye.
• Plugin Manifest Builder — plugin.json banaiye aur uski apekshit folder banawat dekhiye.
• Hooks Builder — settings.json ke hooks banaiye aur kabhi na chalne wale matchers pakadiye.
• Permission Rule Tester — Allow, ask aur deny rules asli commands par aazmaiye.
• MCP Server Config — stdio, HTTP ya SSE ke liye MCP server entry banaiye.
• MCP Tool Schema Builder — Tool tay kijiye aur JSON Schema paiye, strict mode jaanch ke saath.
• Tool Schema Checker — Definition paste kijiye aur jaaniye wo kyun reject ho rahi hai.
• Prompt Template Filler — Ek prompt ko rows ki table par failaiye aur kami dekhiye.
• System Prompt Linter — Contradictions, bekaar instructions aur bina bhare placeholders dhoondhiye.
• Prompt Diff — Do versions shabd-dar-shabd milaiye, token antar ke saath.
• Few-shot Example Builder — Input-output jodon ko tag, JSON ya chat roop ke examples banaiye.
• XML Prompt Section Builder — Sections ko aise tags mein lapetiye jinse content bahar na nikle.
• JSONL Viewer aur Validator — JSONL file line-dar-line padhiye, chhaantiye aur jaanchiye.
• Chat Transcript Converter — Messages array ko padhne layak transcript mein badliye, aur wapas.
• Dataset Validator — chat roop ka JSONL dataset un errors ke liye jaanchiye jo training gira dein.
• Context Window Planner — Window ko prompt, tools, history aur jawab mein baantiye.
• Prompt Cache Planner — Blocks aise kram mein rakhiye ki cache hone wala prefix sabse lamba ho.
• Text Chunker — Document ko asli seemayon par tootne wale overlap tukdon mein baantiye.
• Agent Loop Cost Estimator — Turn-dar-turn kharch nikaaliye, dobara bheje history aur cache ke saath.
• Batch Request Builder — Rows ki table ko alag pehchaan wale bulk job mein badliye.

AI aur agents ke sabhi tools dekhein · Saare tools