Skip to content

try_in_5_minutes.md

Latest commit

 

History

History
110 lines (76 loc) · 3.07 KB

try_in_5_minutes.md

File metadata and controls

110 lines (76 loc) · 3.07 KB
Raw

Try LynkMesh in 5 Minutes

This guide shows the shortest public-safe path to try LynkMesh on the included synthetic fixture.

LynkMesh is a deterministic MeshContext evidence protocol for AI coding agents. It turns codebases into graph-backed, inspectable context artifacts that can support AI-assisted code understanding workflows.

This is an early validation workflow, not benchmark proof.

Prerequisites

  • Python 3.11+
  • Git

Step 1 - Clone and set up

git clone https://github.com/ommukhlis-spec/lynkmesh-open.git
cd lynkmesh-open

Set the deterministic hash seed (required by the pipeline):

Linux / macOS:

export PYTHONHASHSEED=0

Windows (Command Prompt):

set PYTHONHASHSEED=0

Windows (PowerShell):

$env:PYTHONHASHSEED = "0"

Step 2 - Verify the environment

python -m lynkmesh doctor

Expected: Result: ready.

Step 3 - Run the tests

python -m pytest test/semantic/contracts test/unit/cli -q

Expected: 139+ passed.

Step 4 - Build a MeshContext AI Context Pack

The included synthetic fixture is evals/before_after/fixtures/mini_auth_shop_php - a minimal PHP project with auth, products, routing, and middleware.

Compact profile (default):

python -m lynkmesh pack evals/before_after/fixtures/mini_auth_shop_php --pretty

Expanded profile:

python -m lynkmesh pack evals/before_after/fixtures/mini_auth_shop_php --profile expanded --pretty

Supported profiles: compact, balanced, expanded.

Step 5 - Run a token benchmark across all profiles

python -m lynkmesh benchmark evals/before_after/fixtures/mini_auth_shop_php --profiles compact,balanced,expanded --pretty

Note: the benchmark command uses --profiles (plural, comma-separated), not --profile.

What you just ran

Step Command What it does
2 doctor Reports local environment diagnostics (no graph build, no file writes, no network).
3 pytest Runs public contract and CLI unit tests.
4 pack Builds a deterministic AI Context Pack from the synthetic fixture.
5 benchmark Builds a deterministic token estimate across all three profiles.

The LynkMesh commands shown here are local-first: no network access and no LLM inference.

Troubleshooting

PYTHONHASHSEED must be '0' error: The pipeline requires PYTHONHASHSEED=0 for deterministic output. Set it as shown in Step 1.

error: path does not exist: Ensure you are running commands from the repository root (the directory containing __main__.py).

No module named lynkmesh: If you installed from source, add the parent directory to PYTHONPATH:

export PYTHONPATH="$(pwd)/..:$PYTHONPATH"    # Linux / macOS
set PYTHONPATH=%cd%\..;%PYTHONPATH%           # Windows

Next steps