Compression

Fit more signal into fewer tokens: compaction, summarization, token-efficient formats, repo packers, caching.

13 listed entries

Packs a whole repository into one AI-friendly file, ready to feed to a model.

tool

Website captured Oct 19

Token-Oriented Object Notation: a compact, human-readable encoding of JSON for LLM prompts.

tool

Website captured Oct 19

Fast BPE tokenizer for OpenAI models: count tokens before they hit the context window.

tool

Turns any GitHub repo into a prompt-friendly text digest (swap 'hub' for 'ingest' in the URL).

tool

Website captured Oct 19

CLI that converts a codebase into one prompt with a source tree, templates and token counting.

tool

Website captured Oct 19

Microsoft's prompt-compression method: drops low-information tokens; the repo reports up to 20x compression.

tool

Website captured Oct 19

The finite attention an LLM spends across its context; every new token draws it down, so context must be curated.

term

When a conversation nears the window limit, summarize it and restart a fresh window with the summary.

process

Automatically clearing stale tool calls and results from the context as it nears its limit (a Claude Developer Platform feature).

term

The share of prompt tokens served from the model's key-value cache; higher means lower latency and cost.

term

Keep the prompt prefix stable and the context append-only so cached tokens get reused.

process

Mark the stable prefix (system prompt, tools, reference docs) for caching so repeated calls reuse it.

process

Drop old tool results from the history once they've been used: the lightest form of compaction.

process

How they connect