| You want to | Read |
|---|---|
| Install the binary and run it on one change | README |
| Know what the released commands check and what they do not | README: what is not checked and R55 |
| Try it with one reviewer before making it a gate | Adoption guide |
| Build from source, check prerequisites and license notices | Native core |
| Understand the architecture or add a language | Multi-language deconvolution |
| See how checks and releases are built | CI and release workflows |
| Verify a download or set up the signing key | Release signing |
| Follow the experiments, including the failed ones | Dated research log |
| Judge whether the agnostic mutation driver can become part of the Rust binary | Feasibility report |
Use the frozen-manifest comparison (prepare, compare) |
Native workflow and scoped readiness report |
The native core runs on Linux x86_64 and aarch64, Windows x86_64, and macOS on
Intel and Apple Silicon; each push to main is checked on all five in
Actions.
The Python prototype (lesscode-baseline) was retired at commit fd0b5b9.
Its guides (CLI, runtime setup,
analyzer setup, profiles,
technical overview) are kept as archived reference
and carry a notice at the top.
Rows that mention the Python package, uv or profiles belong to the retired prototype. Some rows point to experiment records that are kept in the owner's working tree and may not be present in a fresh clone.
| Goal | Guide |
|---|---|
| Inspect the unified 4-language state-boundary deconvolution architecture | Multi-language deconvolution architecture and extension guide |
| Inspect preserved Python baseline history and commit references | Preserved Python baseline index |
| Inspect the first CodeQL query prototype and its failed callback check | CG-G0: extraction, frozen oracle and replay |
| Review the proposed graph-driven, multi-pass constraint-modeling research | Research blueprint and staged experiments |
| Inspect real paired replacements, including a marginal Facdrone result | mvibe: 23 lines removed and Facdrone CSV: rejected |
| Inspect local Rust/Python/TypeScript baselines and a rejected library shortcut | mmhmem/mvibe measurements, versions, chart and COBESY experiment track |
| Run Rust and Python tests in parallel | cargo-nextest and pytest-xdist |
Find decision sites, write mutants and run a project's tests against them (on main, not yet released) |
The agnostic driver |
| Run the same Linux checks locally (native Dagger CLI with Podman) and in Actions | Dagger pipeline and native platform evidence |
| Build and run the native comparison CLI | Native setup and working paired example |
| Inspect the native decision and maturity boundary | Readiness gate and 32-criterion coverage |
| Inspect the completed Python self-trial and exact reviewed adoption | Self-application record, failures and provenance |
| Inspect the real typed capability trial | TypeScript results and limits |
| Understand the original native stack proposal | Historical recommendation |
| Understand earlier trials and the current SMT-augmented LLM experiment | Compression trial progress |
| Coordinate framework knowledge without dependency proliferation | Repository capability driver and next experiment protocol |
| Compare local discovery with a shared capability inventory | Completed pilot, paired results and chart |
| Compare local and connected-responsibility rewrites | Graph trial: behavior preserved, compression rejected |
| Test whether a feature path contains redundant representations | Analyzer-to-report trace and rejected deletion premises |
| Measure compression across a capability rather than individual functions | ATS bilingual renderer: scoped replacement, chart and checks |
| Repeat scope-based compression on active typed code | CPU/NPU backend trial and two-trial synthesis |
| Install the Python package with uv | Runtime setup |
| Provision and probe the pinned analyzers | Analyzer setup |
| Run the included unchanged fixture | README example |
| Compare your own snapshots | Profile configuration, then CLI reference |
| Try the process with a reviewer | Adoption guide |
| Drive compression trials from real code opportunities | Reusable procedure and agent skill |
| Choose symbolic-execution and guided-input tooling per language | Symbolic execution tooling assessment |
| Understand the input and run a real family prototype | Reader-family trial plan and results |
| Inspect a smaller classifier and source-derived SMT check | Outcome hotspot and next LLM compression pass |
| Inspect a real simplification of Lesscode | Self-evaluation example and charts |
| Inspect solver-assisted compression, its negative control and next measurements | Code-flow synthesis report and charts |
| See selected Facdrone parts and the first self-application rejection | Measured boundaries and test impact |
| Inspect actual Lesscode synthesis coverage and whole-replacement statistics | Lesscode self-application report |
| Research substantial simplification after modularization | Facdrone empirical roadmap |
| Inspect a measured architectural obstruction to simplification | Facdrone solution review and boundary probe |
Read the technical overview for the detailed writeup:
- Architecture and execution
- Measurement definitions
- Decision rules
- Artifact storage and recovery
- Controlled experiments
- Trust boundaries and limits
The CLI reference specifies invocation and exit semantics. The profile guide explains what you must configure and which rules the current implementation can execute.
The measurement guide distinguishes LLOC, complexity, convolution, role-revealing names and actual maintenance effort. Scoped package instructions guide future coding agent edits before a refactoring pass is needed.
- Current native scoped readiness and fresh Windows receipt
- ATS replay, NPU replay and manipulations, TypeScript trial
- Python self-trial and adopted exact file
The following Python baseline receipts are historical evidence for their recorded revisions.
- Native CI procedure and artifact access
- Native audit receipt
- Fresh suite and installed-package receipt
- Final feature validation
- All 56 acceptance criteria and their coverage
The receipts describe specific tested revisions. Raw CI artifacts have finite retention; the CI guide explains how to reproduce them.
The versioned records and serializers in
contracts.py define the JSON boundary.
The approved design and
requirements record the
baseline contract. The roadmap separates delivered
evaluation from proposed agent rules, skills and rewrites.
Analyzer dependency licenses map locked packages to
the original notices retained in tools/notices/; dependencies do not all share
one license.
Documentation design notes record the COBESY framing, composition plan and adoption assumptions behind these guides.