Plugin
Sfoglia, filtra e installa i plugin di DeepSeek-Harness.
8 plugin trovati
Vibe-Mathematics
chongcyrus/vibe-mathematics
Framework multi-agente di risoluzione e verifica di problemi matematici: brainstorming → iterazione del risolutore → dibattito tra verificatori multipli → base di conoscenza verificata, con ripresa dai checkpoint e intervento manuale/automatico (disponibile anche come preset agente).
dsh-tradingagents
megatronyy/dsh-tradingagents
Analisi multi-agente per azioni A-share dietro /trading-agent, portata da TradingAgents-AShare, esegue 14 ruoli (analisti, dibattito bull/bear, revisione rischi, trader) tramite il routing LLM dell'harness.
formalswarm
fashionmascherine-svg/formalswarm
Runs a thesis-antithesis-seal debate whose verdict is computed from real command exit codes and case counts, never from agent prose.
dsh-council
fufengyuan/dsh-council
Council of High Intelligence: riunisci un think tank di 18 figure storiche per un dibattito strutturato da più prospettive, con un pannello laterale DAG live, cronologia di esecuzione SQLite e impostazioni predefinite per progetto tramite .council.yaml.
gateway-agent
elmaxid/gateway-agent
Delega attività, revisioni e dibattiti a modelli alternativi (DeepSeek, MiniMax, Ollama, ecc.) tramite la CLI gateway-companion, senza MCP.
dsh-debate
shlouai/dsh-debate
Answers hard questions by staging a formal debate: two adversarial subagents argue opposite sides with research, then the main agent rules on the transcript through the debate_open, debate_round, and debate_verdict tools.
dsh-multi-role-debate (dsh-multi-role-debate)
jiang12345-code/dsh-multi-role-debate
Multi-role parallel argumentation for DeepSeek Harness: runs the real Codex and Claude CLIs in parallel, has the DSH model (Judge) write a third-party verdict back into the conversation, plus Obsidian-style single-agent direct chat and per-role model config.
dsh-debate
lkshjd/dsh-debate
Multi-agent isolated debate for DeepSeek Harness: debaters of different temperaments research independently (no herding), then cross-examine, then judges converge to a conclusion. Runs as a background job with per-agent progress, token accounting, wave-parallel execution, and session-referenced resume.