Tines The single, secure environment for agents, apps, and automations.
Promoted
Otter Code
Code on your terms. Local AI coding agent.
9 followers
Code on your terms. Local AI coding agent.
9 followers
Otter Code is a terminal-native AI agent: run otter in any repo and keep your whole workflow in the shell. Plug in OpenAI-, Anthropic-, or Gemini-compatible APIs, or Qwen OAuth for a free daily quota. Skills and SubAgents handle deep repo context and multi-step tasks—not just chat. Orange-themed CLI, one-line install: npm i -g @heronsamuel/otter-code. For developers who want control without leaving the terminal.
"Skills and SubAgents" — what does that actually mean in practice? Like if I'm debugging a memory leak that spans 6 files across different modules, does a SubAgent spin up to handle each file independently and then reconcile? Or is "SubAgent" more of a marketing term for a long context window doing sequential tool calls?
@sounak_bhattacharya well they have seperate context but it depends on ur settings, if ur using the same model for subagents, i assume kv cache poining one of ur agents could certainly poison outputs on other instances.
@sounak_bhattacharya at the moment the subagents feature would be identical to qwen codes subagent feature here is more documentation https://qwenlm.github.io/qwen-code-docs/en/users/features/sub-agents/