This index maps the implemented public surfaces to their detailed guides.
For exact installed-version flags and subcommands use slmcode --help and
slmcode <command> --help. For the complete configuration field catalog use
slmcode config schema; configuration explains precedence and
persistence. The slmcode docs command reads project documents, not this website.
| Capability | Entry points | Guide |
|---|---|---|
| Install, upgrade and offline use | Installers, init, update, version |
Installation, offline installation, quick start |
| Interactive coding | Default command, tui, chat |
TUI and chat, user guide |
| Run or watch a project | run, watch |
CLI, recipes |
| Browser interface and live events | studio; run setup, Live floor, board, settings |
Studio |
| Plans, tasks and human steering | plan, board, task, compose, plan/clarify/escalate gates |
CLI, pipeline, teams |
| Review, stage or discard edits | diff, apply, reject, commit |
CLI review commands, permissions |
| Continue or inspect work | session, status, task show |
CLI, user guide |
| Capability | Entry points | Guide |
|---|---|---|
| Discover and configure a served model | configure, config, auth, doctor, readiness |
Providers, configuration, troubleshooting |
| Model context and runtime calibration | calibrate, model profiles |
Calibration, decoding |
| Agents, skills and reusable building blocks | agent, skills, blocks, stack |
Agents, skills, blocks, customization |
| Team composition and specialist ownership | Dynamic/strict teams, team pins, compose |
Teams, pipeline |
| Frontend workflows | Frontend method, skills and specialist roles | Frontend |
| External tools and lifecycle integrations | MCP configuration, hooks trust/list commands |
Tools, customization, permissions |
| Structured responses and tool discipline | Role schemas, constrained decoding, bounded tools | Decoding, tools, conventions |
| External model predictions | laya_* config; native Laya or OpenAI-compatible server |
Decision models |
| Shell completion | completion |
CLI completions |
| Capability | Entry points | Guide |
|---|---|---|
| Token budgets, repository context and retrieval | context, docs, context packs, repository map |
Context engineering, architecture |
| Persistent memory and forgetting | memory |
Self-improvement and memory |
| Adaptive policies and regression history | evolve, metrics |
Self-improvement, CLI |
| Repository knowledge graph | graph |
Knowledge graph |
| Evaluation and bounded experiments | eval, autoresearch |
Testing, autoresearch, SLM learnings |
Adaptive harness policies and stored lessons are distinct from training neural weights. The optional external decision model does not automatically train or replace itself. Its role and limits are documented on the decision-model page.
See the dated release qualification record for verified checks and qualification limits from the latest audit.
- Testing: automated gates, fake-server binary acceptance, live-model release checks and how to interpret their evidence.
- Permissions: scope protection, shell controls, review mode, trusted hooks and Studio access.
- Troubleshooting and FAQ: setup and runtime recovery.
- Architecture, contributing and conventions: implementation and development workflow.
- Migration and changelog: changed behavior and compatibility. Maintainers should also follow the repository's release checklist.