v0.2.0 - Token Profiler

What's New

• ctxfold --profile — analyze any input: where the characters go
  (keys/syntax/values/whitespace, affixes/constants, template boilerplate)
  and what folding would save, without transforming anything
• Programmatic API: profile(text, opts) returns structured data;
  renderProfile() produces the text report
• When nothing folds, the profiler explains why (quoted CSV, nested JSON,
  too few records, prose) instead of silently no-opping
• examples/profile-demo.js — zero-setup demo, no API key required
• 14 new invariant tests (42 total): composition sums exactly to input size,
  profiler ratios always equal compress()'s, honesty flags enforced per format

Details

The profiler makes the same claims the benchmark table does, enforced by
tests: composition is character-exact, foldable figures come from actually
running the encoder, and readability claims mirror measured results — JSON
and logs validated direct-readable, CSV flagged pipeline-only.

First minor version: new capability, no changes to the existing
compress/decompress/validate API.

v0.1.4 - CSV Readability Measured

What's New

• Added examples/gpt-csv-equivalence.js — readability harness for the CSV
  encoder, same methodology as the JSON/logs checks
• Result: folded CSV is not direct-readable (0/24 on gpt-4o-mini, 6–9/24
  across runs on gpt-4o, vs 24/24 raw) — CSV folding is now documented as
  pipeline-mode
• Root cause: CSV folding factors data values; JSON/logs fold only structure
• Benchmark table updated with measured CSV numbers; unvalidated footnote gone

Details

Every cell in the benchmark table is now backed by a model-scored measurement.
Lossless-or-no-op already governed the data contract; this release extends the
same honesty to readability claims.

v0.1.3 - Validation & Decoder Hardening

What's New

• Added schema-drift validate() function for runtime schema checking
• Hardened decoders against malformed input
• New --validate CLI flag to enforce schema compliance during encoding/decoding

Details

Improves reliability and provides developers with tools to catch schema mismatches early.

v0.1.2 - Dictionary Coding & Accuracy Validation

What's New

• Added opt-in dictionary coding for context compression
• Implemented three-arm LLM accuracy experiment to validate compression tradeoffs
• Results: demonstrated savings gains with measurable accuracy delta
• Users can now balance compression vs. model performance

Details

Introduces --dictionary flag for controlled compression with experimental validation framework.