How one Anthropic user runs agents 24/7 with fewer tokens
Al_Grigor · x · 2026-07-22
A practitioner says they maxed out Anthropic only once in the past month — when Fable launched — and used the moment to share context-engineering tactics for running agents 24/7 on one subscription.
- Use progressive disclosure: build hierarchies in AGENTS.md, LLM wikis, and Obsidian so models rarely need full files.
- Plan with the strongest model first, then let cheaper models execute the plan; upfront planning is cheaper than repeated execution.
- For new features or anything that does not already exist, use an LLM wiki as the context layer to speed up bootstrapping.
- The thread frames token efficiency as a workflow problem, not just a model-choice problem.
More from coding & agent
- Four teams independently shipped the same “LLM wiki” pattern after Karpathy’s gist — garrytan · 2026-07-22
- Greptile reviews GPT code with Claude, and Claude code with GPT — garrytan · 2026-07-22
- Claude placed a live crypto trade through Bitpanda MCP, raising latency questions — Bitches172882 · 2026-07-22
- A builder says losing coding-model access would cut productivity sharply — serrjoa · 2026-07-22
- Security agents need harsher isolation because models will cheat, search for hints and peek anywhere — banteg · 2026-07-22
- Hermes Agent v0.19.0 adds subscription controls, reasoning modes and transcript exports — Teknium · 2026-07-22