Overview · How work moves through smith, from ticket to launch card to run to landing on main, and who clicks what.
Glossary · Every word the app's pages explain on hover, one line each with a link to the page that holds the detail, and the proposed new display names for the engine's own words.
Window · A tour of every screen in the app, from the home page and chat to the project tabs, the ticket, asset, run board and commit pages, and the foot line, with where each is explained.
Assets · What an asset is, how assets reach the desk, how git versions are numbered and move at sync, how a launch card suggests assets, and what plugging as reference, target or instructions does
Containment · How smith contains the processes it starts, systemd units on Linux and job objects on Windows, the uncontained development profile, checks_require_containment and run_query host
Feedback loops and retry limits · Why a run goes back; the revise loop, check repair hops, the intake repair, the hop cap, attempt retries and tool retries, each with a default, where it is set and what happens at the limit.
Gates and the acceptance list · How a run stops at a human gate; approve, send back or reject, what the next pass reruns, how the acceptance list is graded and edited, and how a conductor proposes it.
Landing and syncing work · How approved work reaches main; land commits the sealed handoff, merge conflicts and partial lands, then sync pushes or opens a PR, runs checks, and restarts smith only on its own source.
Land and its rebase conflicts · What land does with approved work. A rebase conflict is sent back to work once per run; a second fails with merge_conflict. A rebased land re-runs the checks.
Launch cards · A launch card proposes runs and go starts them; what draft records, what go re-reads and freezes, the queue, lineup, choosing a template, and the 409 no_checks refusal.
Models and presets · How the harness picks a model for each role from its runtimes and tag registry, the default tags per role, model pins, the six shipped presets and the conductor's model and effort
Nodes and builtins · The nodes a template is made of, with node kinds and fields, ports and cords, every builtin and its config keys, the ticket-completion node ids in order, and the model profiles.
Runs, their journal and their saved context · A run's statuses, its journal and every event in it, the hop cap, its workspace and branch, continue, retry and run again, what each run tool answers, and the context it saves at go.
Skills · A skill is an asset with a SKILL.md. How it is plugged as instructions, what claude loads, why codex and grok refuse it, why skills grant no tools, how suggestions work and what is not built
Templates · A run template's file format and header fields, the eight shipped templates and their purposes, where operator templates live, override and reset, and what validation refuses.
Tickets · Columns, kinds and fields of a ticket, how comments override the body, what completing records, and the conductor tools that edit tickets and read the launch queue.
Autopilot · Autopilot hands the manager chat the queue for a while. The five power stops, max_parallel and failures_in_row, park, skip and retry, the no-autopilot and operator labels, and the report.
Cards · Every proposal card in a chat, grouped by what the click does, what the operator sees and whether autopilot may press it, plus the cards no tool drafts, such as blocked and permission.
Chats · How desk chats work. Projects own chats for good, or a chat is general. Covers the manager and its pin, side chats and the delegate contract, reports, handoffs, and archive, move and trash.
The conductor · The conductor runtime (claude or codex), how its brief is built, auto mode and permission cards, turn tags, run news, wake and show_task. The brief summarises this page.
Discord · One Discord bot for every project, where its token comes from, how a project binds a server, the operator and mates audiences, channels, roles and member cards, and the discord page.
Memory and search · Memory file mode and store mode and which one a desk uses, the search tool and its kinds, and what ticket-search embeddings are for, the providers and how to stop the Ollama probe.
Memory layers · The desk, project and area memory layers, where a memory_write with no scope lands, the MEMORY.md line, conductor.project_memory_dir, and which layer edit and delete look in.
Permission rules in desk chats · Which permission rules a manager or general chat starts with, and when a changed rule applies.
Quarantine and problem reports · Why a ticket an install filed is untrusted until the operator accepts it, the nonce fence, how report_issue drafts a card and sends, scrubbing, the outbox, diagnostics zip and crash files.
Roles · The nine built-in roles, what base.md gives every role, user roles in the roles folder (role.md and kit.json), the kit fields, project role overrides and team notes.
Team notes · Team notes hold facts about the tools and the place, at two levels (the install's and each project's). Which chats and runs read which, and how a manager writes each.
Checks · What a check is, its fields (id, executable, args, cwd, timeout, shot), how it runs and what environment it gets, and which test output smith parses.
Config edits · Which config files propose_config_edit can change, how the strict loaders refuse unknown keys, and what smith config check does.
desk.json · Every key of the desk settings file desk.json, with its type, default, whether a change applies live or at restart, and the settings page row that edits it.
harness.json · Every key of the runs settings file (harness.json), where it is read from, how it is decoded, and each key's type, default and meaning, with the nested shapes.
Projects · A project is two halves, a shared record in the tracker and a local file on this machine, with no field in both; which fields each holds, which value wins, and where runs build.
Backup and restore · What a snapshot holds, where backups go, how often and how many are kept, why one was taken, and how a restore is requested and applied at the next start.
The smith command line · Every smith command and flag in one line each, from the cobra definitions in internal/cli, with the desk's memory limit and the legacy task-tree commands.
Environment variables · Every SMITH_ variable and GOMEMLIMIT that smith reads or sets, what reads it, and what it changes.
Smith MCP control plane · The smith mcp server for conductor LLMs, its operating model, tool inventory, live patches and the ticket harness.
Where smith keeps its files · The shell data dir, the core data dir, the desk dir, the work home, run copies and check caches on Linux and Windows, and what is safe to delete.
First-run setup · The ten setup steps, what each one checks and what it stores, when the page opens by itself, and how a fresh install restores from a backup.
Troubleshooting launch errors · Launch error messages, what causes each, and where to read more; covers the retained-context folder errors, the couldn't start line, refusals and broken check environments.
Updates and rollback · How the desktop app finds, downloads, retries and installs an update, and how rollback works.
Windows specifics · How checks run on Windows (no shell, six-variable environment, PATH and PATHEXT lookup), job objects, retained context, the installer, and that smith does not handle line endings.