Save your main context to raise quality and cut cost — building Context Drop

Description and use

Context Drop isolates large logs and screenshots from the main Claude Code conversation, reducing token costs and preventing tool‑call corruption.

Problem

Using Claude Code with heavy context, ultracode fan‑outs, and large logs caused tool‑call corruption, silence, and high token costs.

Outcome

Context Drop reduced token costs by up to 100× for large logs, eliminated tool‑call corruption, and restored model responsiveness after marathon sessions.

Workflow

• 1. Identify heavy context usage (large logs, screenshots, many agents).
• 2. Install Context Drop desktop tool.
• 3. Configure Context Drop to route raw context to a subagent.
• 4. Use /compact only for the main conversation, not for raw data.
• 5. Monitor token usage and model stability.
• 6. Adjust ultracode and effort settings per task.
• 7. Iterate to keep main context under a safe size.

Evidence

If I had to name the one factor most consistently present, it's context bloat.

Source

The result: the conversation snapped back toward its original size immediately after the compact

Source

By the estimate on record, reading a 20k-token log once means paying for about 85k tokens over the next 30 turns.

Source

Context Drop is a small desktop tool I built to cut exactly one habit at the root: pouring raw context directly into the main conversation.

Source