☯ isheng-eqi
AI-native — I don't write code. I direct AI to produce it. Reverse-engineering AI agents. Redefining the human-tool relationship.
Hermes — 10 commits · commits
Agent Core #97167 fix(agent): tool-pair sanitizers (salvage #55845 + #78063 + #59434) #60120 fix(interrupt/auth): prevent /stop swallow (incl. Bedrock) and empty-provider credential corruption #83009 fix(process): reject non-positive wait timeouts (salvage #60004)
Scheduling #59524 fix(cron): durable run-claim prevents one-shot double-execution #59447 fix(cron): reject past one-shot in update_job fallback + resume_job
Gateway / Voice #73517 fix(discord): voice correctness — callback wiring, warm-up clip fix, threading, drain, configurable timeouts
#6373 ConcurrencyLimit max_queued validation · #6375 Tool max_retries validation · #6377 ToolOutput max_retries validation · #6383 Agent tool_timeout validation
#38447 drop redundant len(tag_ids)==0 check · #38448 raise clear error on unsupported execute_code language ·
#767 dify-plugin-daemon — fallback to lowercase proxy env vars
#292 thread-safe node input reads + JSON parse error handling
Janus — AI agent framework, four-role architecture agentcanary — Agent security red-team toolkit claude-code-reverse-engineering — Full binary reverse of Claude Code (215MB) elder-pill — Android medication reminder with deep alarm integration media-crawler-easy — Multi-platform social media scraper pure-chat — Zero-injection, zero-prompt, multi-API chat
Reverse-engineering AI agents · Agent architecture design · Open-source contribution From formlessness, form. Taiji yields two poles.