better-dsh 0.2.4 → 0.2.5

package/docs/80_upstream-docs-and-skills/dsh-dev-skill/pi-and-omp/omp/skills.md

@@ -0,0 +1,187 @@
+<!--
+source: https://omp.sh/docs/skills
+fetched: 2026-09-06
+-->
+
+# Skills
+
+> Give omp a reusable playbook for a kind of work, then let it load that guidance only when it is relevant.
+
+A skill changes **how omp approaches a task**: it supplies instructions, conventions, and optional supporting files for a particular domain. It does not add permissions or an executable tool by itself.
+
+To use an installed skill, start with the task you actually want done:
+
+```text
+Review this Postgres migration for locking and rollback risks.
+```
+
+At session start, omp advertises each available skill's name and description. When your request matches a description, the agent loads that skill's full playbook before working. You do not need special syntax for normal use.
+
+## Find and use an installed skill
+
+In the interactive TUI, run:
+
+```text
+/extensions
+```
+
+Open **Skills** to see each discovered skill's description, status, scope, provider, and source path. Select a skill and press <kbd>Space</kbd> or <kbd>Enter</kbd> to enable or disable it. Changes are persisted; run `/reload-plugins` afterward so the current session refreshes its skills and `/skill:` commands.
+
+When you want to require a particular skill rather than rely on matching, invoke it explicitly:
+
+```text
+/skill:postgres Review migrations/042_add_status.sql for lock risk.
+```
+
+Skill names appear in slash-command completion after you type `/skill:`. A skill token can also appear in ordinary prose:
+
+```text
+Review migrations/042_add_status.sql with /skill:postgres and propose a safer rollout.
+```
+
+Explicit invocation is available by default. It can be disabled with `skills.enableSkillCommands: false`; natural task matching still works. A hidden skill is also absent from automatic matching, but remains available through `/skill:<name>` when skill commands are enabled.
+
+## Install a skill
+
+A standalone skill is just a directory containing `SKILL.md`. Copy it into one of omp's native locations:
+
+```text
+~/.omp/agent/skills/<name>/SKILL.md   # user: every project
+.omp/skills/<name>/SKILL.md           # project: this repository/tree
+```
+
+For example, installing a checked-out skill for your user can be as small as:
+
+```sh
+mkdir -p ~/.omp/agent/skills/postgres
+cp ./postgres-skill/SKILL.md ~/.omp/agent/skills/postgres/SKILL.md
+```
+
+Keep any files referenced by the playbook in that same `postgres/` directory. Discovery is one level deep: `skills/postgres/SKILL.md` is found, while `skills/databases/postgres/SKILL.md` is not.
+
+Skills can also arrive inside plugins. Browse a configured marketplace, then install the plugin at user or project scope:
+
+```sh
+omp plugin discover [marketplace]
+omp plugin install --scope user name@marketplace
+# or: omp plugin install --scope project name@marketplace
+```
+
+See [Marketplace](/docs/marketplace) for adding catalogs, inspecting packages, pinning, upgrading, and uninstalling plugins. After copying, installing, upgrading, or editing a skill, start a new session or run `/reload-plugins`. Plugin modules, hooks, or tools bundled beside a skill can still require a full restart.
+
+## Scope and trust
+
+A user skill is available in every project. A project skill is discovered only when omp starts within that project tree; omp also walks parent directories up to the repository boundary. This makes project skills useful for repository-specific workflows, but it also means a repository you open can contribute instructions.
+
+Treat a skill like code:
+
+- inspect `SKILL.md` and its referenced scripts, templates, and files before enabling it;
+- check the origin shown in `/extensions`, especially in a newly cloned repository;
+- install third-party skills at project scope when they are needed by only one project;
+- remember that a playbook may instruct the agent to run commands using the session's existing permissions.
+
+Skills with the same name do not merge. An explicitly configured custom directory overrides default discovery locations. Otherwise omp keeps the highest-priority provider: native `.omp`, OMP/plugin packages, Claude, Agent/Codex-compatible directories, OpenCode, GitHub, then managed auto-learned skills. Within a walked project source, the closest directory wins. Avoid duplicate names; `/extensions` shows which copy is active and which is shadowed.
+
+## Enable and filter discovery
+
+Use one-run flags when you want a temporary boundary:
+
+```sh
+omp --no-skills
+omp --skills 'git-*,postgres'
+```
+
+`--skills` accepts comma-separated glob patterns and keeps only matching names. For persistent controls, edit `~/.omp/agent/config.yml`:
+
+```yaml
+skills:
+  enabled: true
+  enableSkillCommands: true
+  includeSkills:
+    - "git-*"
+    - postgres
+  ignoredSkills:
+    - "*-experimental"
+  customDirectories:
+    - ~/work/shared-skills
+```
+
+An empty `includeSkills` list includes all names. `ignoredSkills` excludes matching names even if they also match the allowlist. Each custom directory must contain one directory per skill, such as `~/work/shared-skills/postgres/SKILL.md`.
+
+The `/extensions` toggle persists an individual disabled skill as `skill:<name>` in the top-level `disabledExtensions` setting. Prefer the dashboard for individual changes; the equivalent YAML is:
+
+```yaml
+disabledExtensions:
+  - skill:untrusted-skill
+```
+
+Provider-specific switches are also available under `skills`: `enablePiUser`, `enablePiProject`, `enableAgentsUser`, `enableAgentsProject`, `enableClaudeUser`, `enableClaudeProject`, and `enableCodexUser`. All default to `true`.
+
+## Compatible locations
+
+In addition to `.omp`, omp discovers these one-skill-per-directory layouts by default:
+
+| Scope | Locations |
+| --- | --- |
+| User | `~/.agent/skills/`, `~/.agents/skills/`, `~/.claude/skills/`, `~/.codex/skills/`, `~/.config/opencode/skills/` |
+| Project | `.agent/skills/`, `.agents/skills/`, `.claude/skills/`, `.codex/skills/`, `.opencode/skills/`, `.github/skills/` |
+| Managed | `~/.omp/agent/managed-skills/` |
+
+Project `.agent`, `.agents`, and `.claude` locations are found while walking from the working directory toward the repository root. Use `/extensions` rather than guessing when several compatible tools have installed the same name.
+
+## Create the smallest skill
+
+Create `.omp/skills/release-check/SKILL.md` in a project:
+
+```md
+---
+description: Use when preparing or reviewing a release; verify versioning, changelog, build artifacts, and publication status.
+---
+
+# Release check
+
+1. Identify the version and release tag.
+2. Verify the changelog describes the shipped behavior.
+3. Check the build artifacts before publishing.
+4. Report blockers before changing a tag or publishing.
+```
+
+That is a complete current skill. The directory name, `release-check`, becomes its invocation name, so it can be requested with `/skill:release-check` after `/reload-plugins` or in a new session.
+
+## Authoring reference
+
+### Frontmatter
+
+| Field | Required | Current behavior |
+| --- | --- | --- |
+| `description` | Yes for `.omp`, custom-directory, plugin, and GitHub skills | Shown before the body loads and used to decide whether the skill applies. Write concrete task verbs, nouns, and scope. |
+| `name` | No | Overrides the directory name used for matching and `/skill:<name>`. Prefer the directory name unless compatibility requires an override. |
+| `hide` | No | Omits the skill from the automatic skill listing while preserving explicit invocation. |
+| `disable-model-invocation` | No | Agent Skills-compatible spelling with the same effect as `hide`. |
+
+Although some compatible formats accept fields such as `globs` or `alwaysApply`, omp does not currently use them to select or automatically inject a skill. Put the trigger conditions in `description`. For guidance that must always be present, use [Context files](/docs/context-files) instead.
+
+A useful description names:
+
+- the actions: writing, reviewing, debugging, migrating;
+- the objects: Postgres migrations, snapshot tests, release artifacts;
+- the boundary when needed: a directory, file type, or subsystem.
+
+```yaml
+description: Use when adding or reviewing Vitest tests in src/importer; covers fixtures, snapshots, and integration setup.
+```
+
+### Body and supporting files
+
+Write direct, operational instructions. State when the workflow applies, the required sequence, important safety boundaries, and what a successful result contains. Keep detailed references, templates, or scripts beside `SKILL.md` and refer to them with paths relative to the skill directory:
+
+```text
+release-check/
+├── SKILL.md
+├── references/
+│   └── registry-checks.md
+└── scripts/
+    └── verify-artifacts.sh
+```
+
+The body is loaded only on demand, so moving long reference material out of the always-visible description keeps ordinary sessions small. Supporting files do not execute automatically; the playbook must explain when and how they should be used.

package/docs/80_upstream-docs-and-skills/dsh-dev-skill/pi-and-omp/omp/slash.md

@@ -0,0 +1,328 @@
+<!--
+source: https://omp.sh/docs/slash
+fetched: 2026-09-06
+-->
+
+# Slash commands
+
+> The complete in-session command reference: syntax, built-ins, subcommands, aliases, dynamic commands, precedence, and reload behavior.
+
+## Run a command
+
+In the main prompt editor, type `/` to open autocomplete. Keep typing to filter, use the arrow keys to move, and press <kbd>Tab</kbd> or <kbd>Enter</kbd> to accept a completion. <kbd>Escape</kbd> closes the menu.
+
+The smallest useful check is:
+
+```text
+/context
+```
+
+Press <kbd>Enter</kbd> to run it. Arguments follow the command after a space, and command families take a subcommand first:
+
+```text
+/rename investigate checkout timeout
+/usage reset active
+/mcp test github
+```
+
+In the forms below, `[value]` is optional, `<value>` is required, and `a|b` means choose one. Quoted arguments keep spaces together. A colon can separate a top-level command from its first argument, which is why both `/force write` and `/force:write` work.
+
+There is no top-level `/help` built-in. Learn commands from `/` autocomplete and its inline argument hints. Type a command family plus a space to see its subcommands. `/mcp help`, `/ssh help`, and `/marketplace help` print family-specific usage; `/hotkeys` shows keyboard shortcuts. Invalid arguments also print the accepted form.
+
+This page describes the interactive terminal. Commands run only from the main session editor, not while a subagent chat is focused. During a collaboration, guests may run only `/dump`, `/export`, `/copy`, `/hotkeys`, `/settings`, `/leave`, `/collab`, `/exit`, and `/quit` locally; every other slash command is host-only.
+
+## Built-in commands
+
+These names ship in the core registry and take precedence over dynamically discovered commands.
+
+### Modes, models, and turn control
+
+| Command | Visible effect and availability |
+| --- | --- |
+| `/settings` | Opens the settings menu. |
+| `/setup [providers]` Alias: `/providers` | Opens provider setup for sign-in and web-search providers. No other subcommand is accepted. |
+| `/plan [prompt]` | Toggles plan mode; an inline prompt starts planning immediately. Requires `plan.enabled`, cannot start during goal mode, and is mutually exclusive with vibe mode. See [Plan mode](/docs/plan). |
+| `/plan-review` | Reopens the latest plan review. Available only while plan mode is active. |
+| `/vibe [prompt]` | Toggles read-only vibe mode and optionally starts it with a prompt. It cannot start during plan or goal mode. See [Vibe mode](/docs/vibe). |
+| `/goal [objective]` | Opens the goal editor when inactive or the management menu when active; an inline objective starts a goal directly. Requires `goal.enabled` and cannot start during plan or vibe mode. See the subcommands below and [Goal mode](/docs/goal). |
+| `/guided-goal [rough objective]` | Starts an interview in chat, then configures goal mode from the result. It has the same `goal.enabled`, plan-mode, and vibe-mode constraints as `/goal`. |
+| `/loop [count\|duration] [prompt]` | Toggles automatic resubmission after each yield. A positive integer limits iterations; durations accept compact or spaced forms such as `90s`, `10 min`, or `1h30m`. Plain prose is an unbounded inline prompt. Run `/loop` again to stop; <kbd>Escape</kbd> cancels the current iteration. |
+| `/queue <message>` | Delivers a follow-up after the active turn yields; if idle, starts it immediately. `-> message` and `=> message` are equivalent editor shorthand. |
+| `/model` Alias: `/models` | Opens the model hub for role and provider selection. In the TUI this command does not take a model ID argument. |
+| `/switch` | Opens the compact session-only model picker, the same action as <kbd>Alt</kbd>+<kbd>P</kbd>; it does not persist a role setting. |
+| `/fast [on\|off\|status]` | Bare form toggles priority service (`service_tier=priority` for OpenAI or fast speed for Anthropic). Enabling can be unavailable for the current model. `toggle` is also accepted. |
+| `/extended-context [on\|off\|status]` | Bare form toggles the persisted premium long-context setting; `toggle` is also accepted. |
+| `/computer [on\|off\|status]` | Bare form toggles the native computer-use tool for this session. `status` reports backend, permissions, exposure, and model; enabling can be unavailable in the session. `toggle` is also accepted. |
+| `/vision <on\|off\|auto\|status>` | Sets a session override for the `inspect_image` delegation tool. `auto` follows the configured mode and hides delegation for models with native image input. |
+| `/prewalk` | Arms the authenticated `@smol` role for the next todo-gated edit or write, even if omp was not launched with `--prewalk`. See [Prewalk](/docs/prewalk). |
+| `/advisor [on\|off\|status\|dump [raw]\|configure]` | Bare form toggles the advisor; `/advisor toggle` is equivalent. Explicit forms are `/advisor on`, `/advisor off`, `/advisor status`, `/advisor dump [raw]`, and `/advisor configure`. `dump` copies a compact transcript unless `raw` is supplied; `configure` opens the TUI editor. A configured advisor still needs a model assigned to the `advisor` role. See [Advisor](/docs/advisor). |
+| `/browser [headless\|visible]` | Bare form toggles the browser between headless and visible and restarts it. Requires `browser.enabled`. `hidden` means headless; `show` and `headful` mean visible. |
+| `/force:<tool-name> [prompt]` Also `/force <tool-name> [prompt]` | Forces one tool choice for the next agent turn. An inline prompt starts that turn; without it, the next message uses the choice. The tool must be active. |
+| `/live` | Starts Codex-backed realtime voice mode; run it again to stop. Interactive TUI only. |
+| `/pause` | Freezes the main agent, subagents, and advisor at their next safe boundary. In-flight calls finish; <kbd>Escape</kbd>, <kbd>Enter</kbd>, or <kbd>Space</kbd> resumes them. |
+
+#### `/goal` subcommands
+
+| Form | Effect |
+| --- | --- |
+| `/goal set <objective>` | Sets or replaces the persistent objective. A bare objective after `/goal` is the short form. |
+| `/goal show` | Shows the objective, status, and budget. |
+| `/goal pause` | Pauses the current goal. |
+| `/goal resume` | Resumes a paused goal. |
+| `/goal drop` | Removes the goal from this session. |
+| `/goal budget <N\|off>` | Changes or removes its token budget. |
+
+### Security
+
+`/security` requires `security.enabled`. A bare `/security` lists stored scans. The command starts local scan work as background operations, so use `status` rather than waiting on the prompt. Cloud forms also require a stored OpenAI Codex OAuth account. See [Security](/docs/security) for the scan model and artifacts.
+
+| Form | Effect |
+| --- | --- |
+| `/security plan [options]` | Creates an immutable plan and prints its ID and fingerprint. Options: repeated `--path <path>`, `--exclude <path>`, and `--knowledge-base <path>`; `--working-tree`; `--diff <base> <head>`; `--output <dir>`; `--archive-existing`; `--credential <id>`. |
+| `/security scan [<plan-id>\|options]` | Starts an existing `secplan_…`, or creates a plan from the same options and starts it. Prints scan and operation IDs. |
+| `/security status [operation-id]` | Shows one operation or lists operations. |
+| `/security cancel <operation-id>` | Requests cancellation of a running operation. |
+| `/security scans` | Lists stored project scans. |
+| `/security show <scan-id\|security://…>` | Renders a scan or another security resource. |
+| `/security import <sarif-file\|bundle-directory>` | Imports SARIF or a Codex Security bundle into the project store. |
+| `/security export <scan-id> --output <path> [--format bundle\|sarif\|report]` | Writes a canonical bundle, SARIF document, or report. |
+| `/security validate <finding-uri>` or `/security validate <scan-id> <finding-id>` | Starts an agent turn that validates one finding with native security tools. |
+| `/security compare <before-scan-id> <after-scan-id>` | Prints finding-lineage comparison data. |
+| `/security disposition <scan-id> <finding-id> <open\|false_positive\|accepted_risk\|fixed\|wont_fix> [rationale]` | Updates a finding. Every state except `open` requires a rationale. |
+| `/security cloud scans [--credential <id>]` | Lists Codex Security cloud configurations. This accepted subcommand is not currently offered by nested autocomplete. |
+| `/security cloud start --repo-id <id> --repo-url <url> --environment <id> [--lookback <days\|all>] [--credential <id>]` | Starts a cloud scan and consumes cloud scan allowance. |
+| `/security cloud status <configuration-id> [--credential <id>]` | Shows cloud scan statistics. |
+| `/security cloud pull <configuration-id> [--credential <id>]` | Imports cloud findings into the local project store. |
+
+### Sessions, history, and context maintenance
+
+| Command | Visible effect and availability |
+| --- | --- |
+| `/session [info]` | Opens current-session information in the TUI. |
+| `/session delete` | Deletes the current persisted session and returns to the selector. Unavailable while streaming or for an in-memory session. |
+| `/session pin [account]` | Pins the current provider to a stored OAuth account; without an account, opens the account selector. This is different from top-level `/pin`. |
+| `/new` | Starts a new session without deleting the current one. |
+| `/fresh` | Closes provider-side stream state and starts fresh provider state without changing the local transcript. Unavailable while streaming. |
+| `/clear` | Clears conversation context in place while keeping the session. Unavailable while streaming. |
+| `/drop` | Deletes the current session and starts a new one. |
+| `/resume [session-id\|@claude\|@codex]` | Opens the local session picker, opens a Claude/Codex import picker, or resumes the matching session ID. Interactive TUI only. |
+| `/pin [session-id]` | Toggles the current or named session at the top of the resume list. |
+| `/rename <title>` | Gives the current session a user-set title. |
+| `/move <path>` | Moves the session to another existing directory, reloads directory-scoped settings and discovered capabilities, and makes that path the working directory. Unavailable while streaming. |
+| `/add-dir <path>` | Adds an existing additional workspace root to this session. Unavailable while streaming. |
+| `/remove-dir <path>` | Removes an additional workspace root. The working directory itself cannot be removed; use `/move`. |
+| `/dirs` | Lists the working directory and every additional workspace root. |
+| `/tree` | Opens the in-place session tree and switches to an existing branch. |
+| `/branch` | Selects an earlier message and creates another branch in the same session file. |
+| `/fork` | Selects an earlier message and creates a separate session file from it. |
+| `/compact [focus]` | Manually compacts context using configured methods and optional focus instructions. See [Compaction](/docs/compaction). |
+| `/compact soft [focus]` | Forces a local summary with the active model for this run. |
+| `/compact remote [focus]` | Tries OpenAI-compatible server compaction, then a local summary. |
+| `/compact snapcompact` | Archives history into dense bitmap images without an LLM summary; it rejects focus text. |
+| `/shake [elide\|images\|thinking]` | Removes heavy context without summarizing. Bare form/`elide` strips tool results and large blocks; the other modes strip image or thinking blocks. |
+| `/handoff [focus instructions]` | Generates a handoff document and commits it as a compaction entry in the same session. Requires an idle session with enough messages; it does not automatically end the turn or create a new session. See [Handoff](/docs/handoff). |
+| `/btw <question>` | Asks an ephemeral side question with current context; the exchange is not added to the persisted conversation. |
+| `/tan <work>` | Starts a full background agent for tangential work. Interactive TUI only. |
+| `/retry` | Retries the last failed agent turn. It works only while idle and only when there is a failed turn to retry. |
+| `/omfg <complaint>` | Converts a complaint about recurring behavior into a TTSR rule. Interactive TUI only. |
+| `/cleanse [request] [--all]` | Detects and fixes project diagnostics with weighted parallel subagents. Interactive TUI only. |
+
+#### `/todo` subcommands
+
+A bare `/todo` displays the current phased task list.
+
+| Form | Effect |
+| --- | --- |
+| `/todo edit` | Opens a Markdown round trip in `$EDITOR`. |
+| `/todo copy` | Copies the list as Markdown. |
+| `/todo export [path]` | Writes Markdown; default path is `TODO.md`. |
+| `/todo import [path]` | Replaces the list from Markdown; default path is `TODO.md`. |
+| `/todo append [phase] <task…>` | Appends a task; fuzzy-matches the phase or creates one. |
+| `/todo start <task>` | Fuzzy-matches and marks a task in progress. |
+| `/todo done [task\|phase]` | Marks the fuzzy match completed; no target applies to all. |
+| `/todo drop [task\|phase]` | Marks the fuzzy match abandoned; no target applies to all. |
+| `/todo rm [task\|phase]` | Removes the fuzzy match; no target applies to all. |
+| `/todo help` Alias: `/todo ?` | Prints todo usage. |
+
+### Status and inspection
+
+| Command | Visible effect and availability |
+| --- | --- |
+| `/jobs` | Shows running and recently settled background jobs. Settled jobs remain for about five minutes. |
+| `/usage [show]` | Shows provider usage, rate limits, resets, and account attribution. |
+| `/usage reset [account\|active]` | Spends a saved Codex rate-limit reset. With no argument, the TUI opens an account selector. |
+| `/stats [--port <port>] [--host <host>]` | Syncs session files and launches the local statistics dashboard. |
+| `/context` | Shows the estimated token budget by source for the current turn. |
+| `/tools` | Shows tools currently visible to the agent and distinguishes active from available tools. |
+| `/changelog [full]` | Shows recent entries; `full` shows the complete changelog. |
+| `/hotkeys` | Shows the live shortcut list. See [Keybindings](/docs/keybindings) for the complete remapping reference. |
+| `/extensions` Alias: `/status` | Opens Extension Control Center to inspect and enable or disable discovered capabilities. Interactive TUI only. |
+| `/agents` | Opens Agent Control Center for task agents, models, prewalk, and advisor configuration. Interactive TUI only. |
+| `/debug` | Opens the configured debugger selector. Interactive TUI only. |
+
+### Memory
+
+A bare `/memory` is `/memory view`. The `mm` family is interactive-TUI-only and requires an active Hindsight backend with mental models enabled.
+
+| Form | Effect |
+| --- | --- |
+| `/memory view` | Shows the memory payload currently injected into the session. |
+| `/memory stats` | Shows backend statistics when the backend provides them. |
+| `/memory diagnose` | Runs backend diagnostics when supported. |
+| `/memory clear` Alias: `/memory reset` | Clears persisted memory data and artifacts, then refreshes the system prompt. |
+| `/memory enqueue` Alias: `/memory rebuild` | Enqueues consolidation maintenance. |
+| `/memory mm [list]` | Lists mental models on the active bank. |
+| `/memory mm show <id>` | Shows one mental model. |
+| `/memory mm refresh [id]` | Refreshes auto-refresh models bank-wide or refreshes one ID. |
+| `/memory mm history <id>` | Shows the model's change history. |
+| `/memory mm seed` | Creates missing built-in mental models. |
+| `/memory mm delete <id>` Alias: `/memory mm remove <id>` | Deletes one mental model. |
+| `/memory mm reload` | Re-pulls the cached mental-model injection block. |
+
+### Export, sharing, and live collaboration
+
+| Command | Visible effect and availability |
+| --- | --- |
+| `/export [--themes] [path]` | Writes the session as HTML. `--themes` bundles the active TUI light/dark themes; a path cannot contain spaces. |
+| `/dump` | Copies the plain-text transcript and writes the current LLM request JSON to a temporary file when possible. The JSON can contain raw context or secrets. |
+| `/share` | Uploads an encrypted session and prints a share URL, using the configured share server/store or secret-gist path. Large sessions can be trimmed. |
+| `/copy` | Opens a conversation-copy picker. |
+| `/copy code` | Copies the last code block. |
+| `/copy cmd` Alias: `/copy command` | Copies the last shell command or Python evaluation. |
+| `/collab [start] [relay-url]` | Starts live hosting, or shows the existing writable link. Uses `collab.relayUrl` when no URL is supplied; a scheme-less relay defaults to `wss://`. Interactive TUI only. |
+| `/collab view [relay-url]` | Starts hosting if needed and shows a read-only watcher link. |
+| `/collab status` | Shows the active link and participants, or guest status. |
+| `/collab stop` | Stops hosting. |
+| `/join <link>` | Joins a writable or read-only collaboration link. You must stop hosting or leave another room first. Interactive TUI only. |
+| `/leave` | Leaves as a guest or stops the room as its host. |
+
+### Providers, MCP, SSH, and plugins
+
+| Command | Visible effect and availability |
+| --- | --- |
+| `/login [provider]` | Opens OAuth provider selection or starts the named provider. During a manual callback flow, `/login <redirect-url>` submits the callback. Interactive TUI only. |
+| `/logout [provider]` | Opens logout selection or revokes the named OAuth provider. Interactive TUI only. |
+
+#### `/mcp` subcommands
+
+A bare `/mcp` shows its help. The no-argument `/mcp add` wizard is interactive-TUI-only.
+
+| Form | Effect |
+| --- | --- |
+| `/mcp add` | Opens the server setup wizard. |
+| `/mcp add <name> [--scope project\|user] [--url <url> --transport http\|sse] [--token <token>] [-- <command…>]` | With only a name, opens the wizard prefilled. Supplying `--url` or `-- <command…>` quickly adds a remote HTTP/SSE or local stdio server. |
+| `/mcp list` | Lists configured and discovered servers. |
+| `/mcp remove <name> [--scope project\|user]` Alias: `/mcp rm <name> [--scope project\|user]` | Removes the project entry by default. |
+| `/mcp test <name>` | Tests a connection. |
+| `/mcp reauth <name>` | Runs OAuth authorization again. |
+| `/mcp unauth <name>` | Removes stored OAuth state. |
+| `/mcp enable <name>` | Enables a server. |
+| `/mcp disable <name>` | Disables a server. |
+| `/mcp smithery-search <keyword> [--scope project\|user] [--limit <1-100>] [--semantic]` | Searches Smithery and opens deployment selection. |
+| `/mcp smithery-login` | Logs in and caches the Smithery API key. |
+| `/mcp smithery-logout` | Removes the cached Smithery key. |
+| `/mcp reconnect <name>` | Reconnects one server. |
+| `/mcp reload` | Reloads server configuration and rebinds runtime tools. |
+| `/mcp resources` | Lists resources from connected servers. |
+| `/mcp prompts` | Lists prompts from connected servers. A prompt is invoked as `/<server>:<prompt> [name=value…]`. |
+| `/mcp notifications` | Shows notification capabilities and subscription state. |
+| `/mcp help` | Prints MCP usage. |
+
+#### `/ssh` subcommands
+
+A bare `/ssh` shows its help.
+
+| Form | Effect |
+| --- | --- |
+| `/ssh add <name> --host <host> [--user <user>] [--port <port>] [--key <key-path>] [--desc <description>] [--compat] [--scope project\|user]` | Adds an SSH host. `--compat` selects compatibility mode. |
+| `/ssh list` | Lists configured hosts. |
+| `/ssh remove <name> [--scope project\|user]` Alias: `/ssh rm <name> [--scope project\|user]` | Removes the project entry by default. |
+| `/ssh help` | Prints SSH usage. |
+
+#### `/marketplace` and `/plugins`
+
+| Form | Effect |
+| --- | --- |
+| `/marketplace` or `/marketplace install` | Opens the interactive install browser. Outside the TUI, an explicit `name@marketplace` is required. |
+| `/marketplace add <source>` | Adds a source such as `owner/repo`. |
+| `/marketplace remove <name>` Alias: `/marketplace rm <name>` | Removes a source. |
+| `/marketplace update [name]` | Refreshes one catalog or all catalogs. |
+| `/marketplace list` | Lists configured sources. |
+| `/marketplace discover [marketplace]` | Lists available plugins, optionally from one source. |
+| `/marketplace install [--force] [--scope user\|project] <name@marketplace>` | Installs and reloads the selected plugin. |
+| `/marketplace uninstall` | Opens the TUI uninstall selector. |
+| `/marketplace uninstall [--scope user\|project] <name@marketplace>` | Uninstalls one plugin. |
+| `/marketplace installed` | Lists installed plugins, scope, and shadowing state. |
+| `/marketplace upgrade` | Upgrades every outdated plugin. |
+| `/marketplace upgrade [--scope user\|project] <name@marketplace>` | Upgrades one plugin. |
+| `/marketplace help` | Prints marketplace usage. |
+| `/plugins [list]` | Lists npm and marketplace plugins and their enabled state. |
+| `/plugins enable [--scope user\|project] <name@marketplace>` | Enables an installed marketplace plugin. |
+| `/plugins disable [--scope user\|project] <name@marketplace>` | Disables an installed marketplace plugin. |
+| `/reload-plugins` | Refreshes discovery caches, skills, file slash commands, task agents, capability state, and MCP servers in the active session. See reload limits below. |
+
+### Exit
+
+| Command | Effect |
+| --- | --- |
+| `/exit` | Exits the interactive application. |
+| `/quit` Alias: `/q` | Exits the interactive application. |
+
+## Commands that are not core built-ins
+
+Autocomplete combines several command systems. Their icons and source labels identify where they came from.
+
+### Bundled workflow commands
+
+Three useful commands ship outside the core registry, so they can be overridden:
+
+| Command | Source and effect |
+| --- | --- |
+| `/review [GitHub PR URL\|pr://owner/repo/number] [instructions]` | Bundled TypeScript command. A PR reference starts that review directly; otherwise the TUI offers base-branch, uncommitted, commit, or custom review. See [Review](/docs/review). |
+| `/green` | Bundled TypeScript command that expands to a workflow for watching and repairing CI until the current branch is green. |
+| `/init [request]` | Bundled Markdown file command that expands to the repository-initialization workflow. A discovered file command named `init` replaces it. |
+
+### Skills
+
+When `skills.enabled` and `skills.enableSkillCommands` are on, every discovered skill appears as:
+
+```text
+/skill:<name> [arguments]
+```
+
+The command loads that skill's instructions for the turn. Skill commands are user-attributed prompts, not core built-ins. Use `/extensions` to see their source and state. See [Skills](/docs/skills).
+
+### Extension and executable commands
+
+Extensions may register in-session commands with their own UI or program logic. User/project TypeScript command modules are discovered from `commands/<name>/index.{ts,js,mjs,cjs}` under the active agent/config directories, and MCP prompt commands are added by connected servers as `/<server>:<prompt>`. Built-in names are reserved. omp's bundled `/review` and `/green` are the lowest-priority executable commands; a user or project command module with the same name replaces the bundled one. See [Authoring extensions](/docs/extension-authoring).
+
+### File slash commands
+
+Markdown under a discovered `commands/` capability directory becomes `/<filename>` and expands to a prompt. Native locations are the current project's `.omp/commands/*.md` and the active profile's agent `commands/*.md`; compatible Claude Code, agent-dir, Codex, OpenCode, and installed-plugin providers are also discovered.
+
+For duplicate file-command names, the capability provider with the higher priority wins. Native `.omp` commands have the highest provider priority, and project `.omp/commands` wins over the active-profile user directory. Plugin and compatibility providers follow. The bundled `/init` is used only when discovery did not already claim `init`.
+
+### Prompt templates
+
+Markdown prompt templates also appear as `/<filename>`, but they are a separate, last-resort expansion system. Project templates live under `.omp/prompts/**/*.md`; user templates live in the active agent profile's `prompts/**/*.md`. User templates load before project templates, so a user template wins a same-name project template. See [Prompt templates](/docs/prompt-templates) for arguments and interpolation.
+
+### Resolution order
+
+When the same token exists in more than one system, omp resolves it in this order:
+
+1. core built-in name or alias;
+2. `/skill:<name>`;
+3. extension-registered command;
+4. TypeScript custom command, then an MCP prompt command;
+5. discovered Markdown file slash command;
+6. prompt template.
+
+Autocomplete omits lower-priority prompt templates that are already claimed. Use distinctive names instead of relying on shadowing.
+
+### Reload after changing commands
+
+`/reload-plugins` refreshes skills, Markdown file slash commands, task-agent definitions, capability caches, and MCP connections in the active TUI. Restart omp after adding or changing prompt templates, executable command modules, extension modules, hooks, or custom tools; those initialized surfaces are not fully rebuilt in place. Moving the session with `/move` refreshes directory-scoped capabilities, but a new session is still the reliable way to load changed prompt templates or executable code.
+
+## Slash commands versus magic keywords
+
+[Magic keywords](/docs/magic-keywords) are ordinary words in a prompt that add hidden steering for that turn; they are not `/` commands and do not appear in command autocomplete. Likewise, `/review` is a bundled workflow command rather than a core built-in. Use [Review](/docs/review), [Advisor](/docs/advisor), [Vibe](/docs/vibe), [Security](/docs/security), and [Prewalk](/docs/prewalk) for their focused workflows; this page is the inventory and syntax reference.