Status: stable language release
Version label: v1.1.0
Python distribution version: 1.1.0
Audience: users, implementers, host integrators, and contributors
Purpose: define the public commitments and non-commitments for CogLang v1.1.0
CogLang v1.1.0 is a stable language release and reference implementation for a graph-first intermediate language designed for LLM-generated graph queries and updates.
It is intended to sit between a language model and a graph-backed host, so generated operations can be parsed, validated, inspected, traced, and denied before they are allowed to affect a host system.
This release aligns three public version surfaces:
- GitHub release tag:
v1.1.0 - Python distribution version:
1.1.0 - CLI
language_release:v1.1.0
CogLang v1.1.0 commits to the following public surfaces:
- The canonical M-expression language form documented in
CogLang_Specification_v1_1_0_Draft.md. - The public CLI entry point
coglang. - The documented command surface exposed through
coglang info,coglang manifest,coglang bundle,coglang smoke, andcoglang release-check. - The executable conformance cases documented in
CogLang_Conformance_Suite_v1_1_0.mdand covered by the test suite. - Error-as-value behavior for documented language operators.
- Profile and capability boundaries documented in
CogLang_Profiles_and_Capabilities_v1_1_0.md. - Operator status and call-surface summaries documented in
CogLang_Operator_Catalog_v1_1_0.md. - A local reference host bridge and Host Runtime Contract v0.x materials for reviewable host integration experiments.
- English-first public documentation, with Chinese companion translations where provided.
- Apache-2.0 licensing for the public repository.
CogLang v1.1.0 does not claim the following:
- It is not a general-purpose programming language.
- It is not a schema definition language.
- It is not a replacement for Cypher, SPARQL, GQL, or any native graph database query language in that system's native setting.
- It does not claim benchmark superiority over other graph languages.
- It does not expose Python internal modules as a stable public API unless that surface is explicitly documented.
- It does not freeze a complete multi-host runtime standard.
- It does not promise a mature extension ecosystem, LSP integration, or IDE support in this release.
The stable v1.1.0 commitment applies to:
- language syntax and canonicalization behavior documented for the current release
- documented operator semantics and conformance examples
- public CLI command names and documented JSON payload shapes
- release metadata paths used by
manifest,bundle, andrelease-check - public documentation boundaries and non-goals
The stable v1.1.0 commitment does not apply to:
- private helper functions
- internal Python module layout
- experimental or reserved operators beyond their documented status
- future Host Runtime Contract v0.2+ decisions
- unreleased ecosystem tooling
The stable release keeps the pre-release language target and closes the public release metadata gap:
COGLANG_LANGUAGE_RELEASEchanges fromv1.1.0-pretov1.1.0.- The Python package version changes from
0.1.0to1.1.0. - The active release notes document changes from
CogLang_Release_Notes_v1_1_0_pre.mdtoCogLang_Release_Notes_v1_1_0.md. - PyPI publication uses Trusted Publishing and does not require a long-lived PyPI API token.
- GitHub-only pre-release tags remain historical source releases and should not be backfilled to PyPI.
For the stable release artifact:
pip install coglang
coglang manifest
coglang release-checkFor source development:
pip install -e .[dev]
python -m pytestA release artifact should pass:
python -m pytestpython -m build- fresh virtual environment wheel installation
coglang manifestcoglang release-checkcoglang smoke
The public release workflow publishes only stable v* tags whose tag name matches the Python distribution version.
CogLang v1.1.0 is a stable language and CLI release for auditable LLM-generated graph operations, with clear public boundaries around what is stable, what remains host-contract work, and what is intentionally out of scope.