gcf-python

Python implementation of GCF (Graph Compact Format) — the most token-efficient wire format for LLMs. A drop-in alternative to JSON and TOON for any structured data.

79% fewer input tokens than JSON. 75% fewer output tokens. 52% smaller than TOON. 100% LLM comprehension at 500 symbols, where JSON fails at 66.7%.

Docs: gcformat.com · Playground · GCF vs TOON

Install

pip install gcf-python

Zero dependencies. Pure Python. Python 3.9+. Includes CLI. Don't want to change code? Use the MCP proxy for zero-code adoption.

CLI

gcf encode < payload.json    # JSON to GCF
gcf decode < payload.gcf     # GCF to JSON
gcf stats  < payload.json    # token comparison with visual bar

Payload: 50 symbols, 20 edges

  JSON  ██████████████████████████████  4,200 tokens
  GCF   ████████░░░░░░░░░░░░░░░░░░░░░░  1,150 tokens

  Savings: 73% fewer tokens with GCF

Library

Quick Start

from gcf import encode, Payload, Symbol, Edge

p = Payload(
    tool="context_for_task",
    token_budget=5000,
    tokens_used=1847,
    symbols=[
        Symbol(qualified_name="pkg.AuthMiddleware", kind="function", score=0.78, provenance="lsp_resolved", distance=0),
        Symbol(qualified_name="pkg.NewServer", kind="function", score=0.54, provenance="lsp_resolved", distance=1),
    ],
    edges=[
        Edge(source="pkg.NewServer", target="pkg.AuthMiddleware", edge_type="calls"),
    ],
)

output = encode(p)

Output:

GCF tool=context_for_task budget=5000 tokens=1847 symbols=2
## targets
@0 fn pkg.AuthMiddleware 0.78 lsp_resolved
## related
@1 fn pkg.NewServer 0.54 lsp_resolved
## edges
@0<@1 calls

Decode

from gcf import decode

p = decode(input_text)
print(p.tool, len(p.symbols), "symbols", len(p.edges), "edges")

Session Deduplication

Track transmitted symbols across multiple tool responses. Previously-sent symbols become bare references instead of full declarations:

from gcf import encode_with_session, Session, Payload, Symbol

sess = Session()

out1 = encode_with_session(payload1, sess)  # full declarations
out2 = encode_with_session(payload2, sess)  # reused symbols as "@N  # previously transmitted"

By the 5th call in a session: 92.7% token savings vs JSON.

Delta Encoding

When the consumer already has a prior context pack, send only what changed:

from gcf import encode_delta, DeltaPayload, Symbol, Edge

delta = DeltaPayload(
    tool="context_for_task",
    base_root="aaa111",
    new_root="bbb222",
    removed=[Symbol(qualified_name="pkg.OldFunc", kind="function")],
    added=[Symbol(qualified_name="pkg.NewFunc", kind="function", score=0.85, provenance="rwr")],
    delta_tokens=30,
    full_tokens=200,
)

output = encode_delta(delta)

81.2% savings on re-queries where the pack changed slightly.

Generic Encoding

Encode any Python value (not just graph payloads) into GCF tabular format:

from gcf import encode_generic

output = encode_generic({
    "employees": [
        {"id": 1, "name": "Alice", "department": "Engineering", "salary": 95000},
        {"id": 2, "name": "Bob", "department": "Sales", "salary": 72000},
    ],
})

Output:

## employees [2]{id,name,department,salary}
1|Alice|Engineering|95000
2|Bob|Sales|72000

Works on dicts, lists, and primitives. Lists of uniform dicts get tabular rows. Nested dicts use ## key section headers.

API

Types

Comprehension Eval

Rigorous 3-way benchmark (GCF vs TOON vs JSON) at 500 symbols, 200 edges. Six structured extraction questions sent to an LLM:

JSON failed on counting tasks. GCF and TOON both achieved perfect accuracy. GCF does it in 32% fewer tokens.

Token Efficiency (TOON's Own Benchmark)

Running TOON's benchmark harness with GCF inserted (their datasets, their tokenizer):

GCF wins on every dataset except deeply nested config (75 tokens on a 618-token payload). On semi-uniform data, GCF uses 44% fewer tokens than TOON.

Reproducible: blackwell-systems/toon@gcf-comparison

Links

• Documentation
• Playground
• Specification
• Go library
• TypeScript library
• MCP Proxy (zero-code adoption)
• GCF vs TOON
• TOON benchmark fork

License

MIT