CLI Interface

Nedura's CLI is a full terminal user interface (TUI) — not a web UI. It features multiline editing, slash-command autocomplete, conversation history, interrupt-and-redirect, and streaming tool output. Built for people who live in the terminal.

First-time setup

One command — nedura setup --portal — and you're ready to nedura chat. See Nous Portal.

tip

Nedura also ships a modern TUI with modal overlays, mouse selection, and non-blocking input. Launch it with nedura --tui — see the TUI guide.

Running the CLI

# Start an interactive session (default)
nedura

# Single query mode (non-interactive)
nedura chat -q "Hello"

# Single query from a file or stdin — nothing is shell-interpreted, so
# arbitrary text (quotes, $(...), backticks) arrives verbatim
nedura chat --query-file prompt.txt
nedura chat --query-file - < prompt.txt

# With a specific model
nedura chat --model "anthropic/claude-sonnet-4"

# With a specific provider
nedura chat --provider nous        # Use Nous Portal
nedura chat --provider openrouter  # Force OpenRouter

# With specific toolsets
nedura chat --toolsets "web,terminal,skills"

# Start with one or more skills preloaded (rendered once into the session's prompt)
nedura chat -s hermes-agent-dev,github-auth

# Resume previous sessions
nedura --resume <session_id>  # Resume a specific session by ID (-r)
nedura --resume "my thread"   # Resume by title (latest "#N" continuation in the lineage)
nedura -c "my thread" --create-if-missing  # Resume that titled session, creating it if absent
nedura --continue             # Resume the most recent session (bare -c)
nedura --resume latest        # Same as bare -c
nedura --resume latest --in ./dir  # Resume ./dir's latest session, staying in ./dir

# Verbose mode (debug output)
nedura chat --verbose

# Isolated git worktree (for running multiple agents in parallel)
nedura -w                         # Interactive mode in worktree
nedura -w -z "Fix issue #123"     # Single query in worktree

Worktree cleanup

nedura -w sessions create disposable worktrees under <repo>/.worktrees/. With nedura --tui -w, closing the terminal detaches from the gateway-owned session. A clean worktree is removed with its branch on quit unless something still needs it. It is retained when it has uncommitted changes or unpushed commits, when the gateway still has a turn queued or running there, when a session in it has an active /loop, /heartbeat or /goal, when an enabled cron job runs in it or is bound to such a session, or when another viewer (a Desktop window, say) still holds that session. Resuming a session uses its working directory: if quit removed the worktree, nedura --tui --resume <id>, nedura --tui -c and the classic nedura --resume/-c recreate it at the same path before attaching (on its branch if that still exists, else on a new branch of the same name from the repo's current HEAD). Only worktrees that nedura --tui -w itself created are recreated. The quit-time check reads the session store and then removes the tree; a viewer that attaches and submits in the moment between the two is not seen, so that turn can find its directory gone. A sweep runs automatically at nedura -w startup and every 6 hours from the cron ticker: idle clean+merged trees are removed, and idle trees with uncommitted changes or unpushed commits are archived, then removed (restorable with nedura worktree restore <name>; see Stale worktree sweep). Reclaim or inspect explicitly:

nedura worktree list              # audit: age, size, verdict, reason per tree + archives
nedura worktree list --json       # machine-readable audit (trees, external trees, branches, archives)
nedura worktree prune             # run the sweep now + delete merged branches
nedura worktree prune --dry-run   # print every archive/remove/skip decision and why; changes nothing
nedura worktree restore <name>    # recreate an archived tree (add --path DIR to put it elsewhere)
nedura worktree prune --older-than 7   # only reap trees idle for 7+ days
nedura worktree prune --trees-only     # leave local branches alone
nedura worktree prune --branches-only  # leave worktrees alone

Worktrees registered outside .worktrees/ (created by hand or by another tool) are reported read-only in list output and are never removed. The one exception is metadata: registrations whose directory no longer exists are dropped via git worktree prune (no files are touched). --older-than DAYS only ever narrows what gets reaped, including what the stale-worktree sweep archives and removes during that run: a tree idle for less than DAYS is never touched.

Inside a session, /worktree prune [--dry-run] does the same (and never touches the tree the session is running in).

Safety guarantees (all modes, any age):

• Uncommitted changes and unique unpushed commits are never deleted without a restorable archive (refs/hermes/archive/<tree>); with worktrees.archive_dirty: false they are never deleted at all — commits that were rebase/squash-merged upstream are detected via git cherry patch-equivalence and count as merged, which is what lets the dominant "merged PR, tree preserved forever" leak finally reclaim.
• Repositories without a remote are judged against the local trunk (main/master, else the branch checked out in the main worktree): only trees and branches whose commits are reachable from — or patch-equivalent to — that trunk are reclaimed. With no trunk to compare against, every tree and branch is preserved.
• Pushed open-PR lanes free their disk without losing anything: when a clean tree's branch head exactly matches what origin holds (checked with one git ls-remote per sweep), the checkout is redundant — the tree is removed but its branch ref is kept, so the lane is one git worktree add .worktrees/<name> <branch> away from restored. If the remote can't be reached, the tree is preserved.
• Trees in use are never touched: a running nedura session's lock, or any live process whose working directory (or, on Linux, an open file) is inside the tree. Each tree is checked again right before it is archived or removed, and again after its archive is built; a tree that changed meanwhile is kept.
• Untracked-only scratch (PR body drafts, notes) is archived to ~/.nedura/archive/worktree-prune/ before its tree is removed — never destroyed.
• Nothing irreproducible is deleted unarchived: a tree is kept when it holds an embedded repository (even a gitignored one), a commit only a per-worktree ref (refs/worktree/*, refs/bisect/*) reaches, an untracked file over the archive size cap. Ignored files (.env copies, caches, build output) are not archived and are deleted with the tree; a nested git repository inside an ignored folder keeps it. See ignored files.
• Branch deletion is content-gated, not name-gated: any local branch whose commits are all on upstream is safe to delete; branches with unique work, checked-out branches, and main/master/develop are always kept.

The same sweep also runs from the cron scheduler (at most once every 6 hours per profile, in the background) over every repository the profile created a worktree in, so gateway-only machines — where nobody launches nedura -w for days — no longer accumulate scratch trees between CLI sessions.

When .worktrees/ grows past 10 trees or 5 GB, startup prints a one-line notice pointing at these commands.

Plugin management

The nedura plugins commands manage native Nedura plugins and portable Agent Plugins v1 packages through the same opt-in workflow:

nedura plugins install owner/repository --no-enable
nedura plugins list
nedura plugins enable <plugin-name>
nedura plugins disable <plugin-name>
nedura plugins update <plugin-name>
nedura plugins remove <plugin-name>

Portable packages remain disabled until explicitly enabled. Nedura currently loads portable Agent Skills and stdio MCP entries. See the plugin developer guide for the exact supported subset and trust boundary.

Interface Layout

The classic CLI is a gateway client

nedura chat (and the bare nedura prompt when display.interface is cli) no longer runs the agent inside your terminal process. It discovers or starts the profile's gateway (nedura gateway ensure), creates or resumes the session there, and attaches over the gateway's local WebSocket — the same session a Desktop window, the TUI, or a Telegram topic can open at the same time. Closing the terminal detaches (Detached; accepted work continues at the gateway.); it does not cancel the turn. Your working directory, --model, --provider, --reasoning, --toolsets, --max-turns, --skills, --checkpoints, --yolo, --accept-hooks, --pass-session-id, --ignore-rules, --base-url/--api-key, and --safe-mode/--ignore-user-config are frozen into the session at creation; --resume may repeat the same values but cannot change them. --resume <id-or-title> and -c <title> name the session; the gateway resolves the title (exact id first, then the latest "<title> #N" continuation, followed to its live tip) among the sessions you may read, and reports No session found matching '…' (exit 1) otherwise. Bare -c and --resume latest resume this workspace's most recent CLI session through the gateway. -c <title> --create-if-missing creates a session with that title when none exists and resumes it when one does, so programmatic callers (Bot Mode's Bot Chat turns) get a deterministic thread. Options that only make sense for an in-process agent — --image, --worktree/-w, --no-restore-cwd, --run-budget, --verbose/-v, --compact, and --create-if-missing without -c <name> — are refused up front (Unsupported gateway CLI options: …, exit 2); the refusal lists, per flag, where that capability lives now (the TUI flag, the config.yaml key, or the nedura subcommand), and nothing falls back to a local agent. A --safe-mode / --ignore-user-config launch reads no profile default model, so it must name one; the refusal prints a complete example such as nedura chat --safe-mode --provider openrouter --model anthropic/claude-sonnet-4 -q "hello". In this mode the slash surface is /stop, /approve <id> <choice>, /answer <id> <text>, /discard <admission_id>, /branch [title], /model <model> [--provider name], /compress [here [N] | <focus>] [--preview], /yolo [on|off], /help and /quit; other slash commands print Unsupported gateway CLI command; use /help. Use the TUI (nedura --tui) for the full slash registry.

If a running turn loses its owner while your finite CLI query is queued, the CLI exits with code 3. Your accepted input stays queued. Resume the printed session interactively, inspect the unknown turn, and use /discard <admission_id> to acknowledge it before continuing. Do not resend the queued input.

Layout styles

display.cli_layout picks how the classic CLI draws. The default, modern, prints everything in the scrollback as plain rows that start with one glyph and never depend on the terminal width, so resizing the window can't garble earlier output:

❯ add a multiply function and run the tests
∴ Thought for 3s · Planning the change
📖 Read calc.py
🔧 Patch  0.2s
│ calc.py
│ @@ -1,2 +1,6 @@
│ +def multiply(a, b):
│ +    return a * b
✗ $ pytest -q  0.4s [exit 1]
◆ Nedura
Added multiply; pytest isn't installed here, so I ran the tests directly: 2 passed.
─ 15.0s · edited 1 file +4 -0 · read 2 files · 8:27 AM
── ⠹ Running tests · 14s ─────────────────────────── Add multiply to calc.py ─
❯ type to redirect this turn · Tab queues it for after
──────────────────────────────────────────────────────────────────────────────
 gpt-5-mini · ctx [░░░░░░░░] 4% · ~/project (main)     enter interrupt · tab queue · ctrl+c stop

• A 40-column bar opens each user message and answer (accent) and each reasoning block (dim); reasoning streams live as in classic and ends with a ∴ Thought for 3s · <title> line. ❯ your message, a finished tool call led by its emoji (📖 Read, 💻 $; ✗ in red when it failed, with its exit code; a refused skill_manage write keeps its emoji and shows the reason dim, since the agent retries it itself), │ an edit's diff under its row, ◆ the answer, and ─ the turn's summary line.
• The composer's top rule carries the live activity (spinner, what's running, elapsed time, streamed tokens) on the left and the session title on the right; no spinner row is printed into the scrollback.
• The status line sits under the composer: the same fields as the classic status bar (model, context, cache hit, latency, tokens/s, background work, goal, duration, YOLO…) plus the working directory (with git branch) on the left; the keys that do something right now on the right.
• The welcome banner is the same in both layouts.

Set display.cli_layout: classic to keep the boxed reasoning and answer panels and the status bar above the input.

The welcome banner shows your model, terminal backend, working directory, available tools, and installed skills at a glance.

Status Bar

In the classic layout a persistent status bar sits above the input area, updating in real time (the modern layout's status line under the composer reads the same data; an explicit display.status_bar.fields list chooses its left side too):

 ◆ claude-sonnet-4-20250514 │ 12.4K/200K │ [██████░░░░] 6% │ $0.06 │ 15m

Element	Description
Model name	Current model (truncated if longer than 26 chars)
Token count	Context tokens used / max context window; ~ marks an estimate
Context bar	Visual fill indicator with color-coded thresholds
Cost	Estimated session cost (or n/a for unknown/zero-priced models). Rates come from Nedura's bundled official price table, then the provider's /models listing; on a direct first-party API (OpenAI, xAI, Anthropic, Google, DeepSeek, Xiaomi) a model missing from both is priced at the vendor's list price from models.dev. Proxies, relays and custom endpoints serving the same model id stay n/a rather than inherit that price.
🗜️ N	Context compression count — how many times the running session has been auto-compressed. Appears once the first compression fires.
▶ N	Active background tasks — how many /bg prompts are still running in the current session. Appears whenever at least one task is in flight.
Duration	Elapsed session time
Session title	Once the session has a title, it appears as a gold badge pinned to the far-right edge. Long titles truncate before displacing the essential model and context fields.
⚠ YOLO	YOLO mode warning — shown whenever NEDURA_YOLO_MODE is on (either nedura --yolo at launch or /yolo toggled mid-session). Mirrors the banner-line warning so you can't forget you're in auto-approve mode.

A ~ before a context count or percentage means it includes a local estimate. This also applies to gateway /status and /context, the TUI, and the Desktop context gauge. An unchanged provider-usage reading has no ~; a provider anchor plus unpriced new messages does. /context reports the selected source. Category, free-space, skill, and toolset breakdowns are always local estimates, even when the overall occupancy comes from provider usage. These display labels do not change compaction decisions or make extra provider requests.

The bar adapts to terminal width — full layout at ≥ 76 columns, compact at 52–75, minimal (model + duration, plus the YOLO badge when active) below 52.

Context color coding:

Color	Threshold	Meaning
Green	< 50%	Plenty of room
Yellow	50–80%	Getting full
Orange	80–95%	Approaching limit
Red	≥ 95%	Near overflow — consider /compress

Use /usage for a detailed breakdown including per-category costs (input vs output tokens).

On the openai-codex provider, /usage also shows any banked usage-limit resets on your ChatGPT account ("You have N resets banked - use /usage reset to activate"). /usage reset redeems one banked reset, fully restoring your 5-hour and weekly limits. Nedura refuses to redeem while your limits aren't exhausted (a banked reset restores the full allowance, so spending it early wastes it) — pass /usage reset --force to redeem anyway.

Session Resume Display

When resuming a previous session (nedura -c or nedura --resume <id>), a "Previous Conversation" panel appears between the banner and the input prompt, showing a compact recap of the conversation history. See Sessions — Conversation Recap on Resume for details and configuration.

Keybindings

On macOS, F6/F7 mean the physical function keys, not the media/system controls shown on the top row. Hold Fn (the globe key on newer keyboards) while pressing the function key, or enable Use F1, F2, etc. keys as standard function keys in System Settings → Keyboard → Keyboard Shortcuts → Function Keys. The reliable terminal fallbacks are Ctrl+T for F6 and Ctrl+R for F7.

Key	Action
Enter	Send message
Alt+Enter, Ctrl+J, or Shift+Enter	New line (multi-line input). Shift+Enter requires a terminal that distinguishes it from Enter — see below. On Windows Terminal, Alt+Enter is captured by the terminal (fullscreen toggle); use Ctrl+Enter or Ctrl+J instead.
Alt+V	Paste an image from the clipboard when supported by the terminal
Ctrl+V	Paste text and opportunistically attach clipboard images
Ctrl+B	Start/stop voice recording when voice mode is enabled (voice.record_key, default: ctrl+b)
Ctrl+G	Open the current input buffer in $EDITOR (vim/nvim/nano/VS Code/etc.). Save and quit to send the edited text as the next prompt — ideal for long, multi-paragraph prompts.
Ctrl+X Ctrl+E	Emacs-style alternate binding for the external editor (same behavior as Ctrl+G).
Ctrl+S	Stash the prompt. Parks the current draft and clears the composer so you can send something else first. Press Ctrl+S again on an empty composer to bring the draft back (cursor at the end, attached images restored). Repeated presses build a stack rather than overwriting, so an earlier draft is never silently lost — with two or more stashed, Ctrl+S opens a browse panel (↑/↓ to navigate, Enter to restore, D to discard, Esc or Ctrl+S to close). A 📌 N badge in the status bar shows how many drafts are parked. Multi-line drafts round-trip exactly, including blank lines. The stash lives in memory for the session only — nothing is written to disk, since drafts often contain secrets.
Ctrl+C	Clear a typed draft first (even while the agent works); with an empty prompt, interrupt the agent (double-press within 2s to force exit)
Ctrl+T / F6	Open the full-screen live work monitor (subagents and background processes) without losing the composer draft. The live dock appears automatically above the status bar; arrows select a worker or process, Enter shows its recent log, s steers a worker, and x requests stop with confirmation. See Monitoring subagents.
Ctrl+R / F7	Toggle the live work dock between its multi-row preview and a single summary line without moving composer focus. Ctrl+R is the reliable fallback when macOS reserves the function-key row. Besides subagents and background processes, the dock shows a standing /goal (active, parked or paused, with turns used) on its top row and the prompts waiting in /queue on its bottom rows.
Ctrl+D	Exit (on an empty prompt; with text it deletes under the cursor). While the gateway runs a turn, it exits and leaves the turn running: the gateway finishes it, ends the session afterwards, and nedura --resume shows the reply. A turn that asks a question (approval or clarify) meanwhile waits for its timeout or for you to resume. Use Ctrl+C to stop the turn instead.
Ctrl+Z	Suspend Nedura to background (Unix only). Run fg in the shell to resume.
Tab	Accept auto-suggestion (ghost text) or autocomplete slash commands. While the agent is working in the modern layout, Tab on a typed message queues it for the next turn
!<command>	Shell mode — run a shell command yourself without spending a model turn (e.g. !git status, !pytest -x). See below.

Multiline paste preview. When you paste a multi-line block, the CLI echoes a compact single-line preview ([pasted: 47 lines, 1,842 chars — press Enter to send]) instead of dumping the whole payload into the scrollback. The full content is still what gets sent; this is just display polish.

! Shell Mode

Start a line with ! to run it as a shell command instead of sending it to the agent:

> !git status
> !ls -la
> !pytest -x tests/hermes_cli

• Zero cost. The model is never invoked — no API call, no tokens, no latency.
• Nothing enters the conversation. The command and its output are not added to history, so your context stays clean and the prompt cache is untouched.
• Runs on your machine, in the session working directory. With the default local terminal backend !pwd matches what the agent would see. A remote or sandboxed terminal.backend (ssh, docker, …) is not used for ! commands — they always run on the host where Nedura itself runs, so !hostname names your machine while the agent's terminal tool names the backend. Ask the agent (or open a shell on the target) to run something inside the backend. Path completion in the composer, by contrast, does follow the configured backend and lists the target's filesystem.
• Approvals still apply. A dangerous command (rm -rf, writes to ~/.nedura/config.yaml, etc.) goes through the same approval prompt the agent's terminal tool uses. ! is a cost/latency shortcut, not a security bypass.
• Non-zero exits are shown. A failing command prints ! exited <code> after its output.
• ! on its own prints a one-line usage reminder.

Shell mode is CLI-only. Gateway platforms (Discord, Telegram, Slack) and cron runs ignore it — those users already have their own shells.

Markdown stripping in final responses. The CLI strips the most verbose markdown fences and **bold** / *italic* wrappers from final agent replies so they render as readable terminal prose rather than raw source. Code blocks and lists are preserved. This does not affect gateway platforms or tool results — they keep their markdown for native rendering.

Slash Commands

Type / to see the autocomplete dropdown. Nedura supports a large set of CLI slash commands, dynamic skill commands, and user-defined quick commands.

Common examples:

Command	Description
/help	Show command help
/model	Show or change the current model
/tools	List currently available tools
/skills browse	Browse the skills hub and official optional skills
/bg <prompt>	Run a prompt in a separate background session
/btw <question>	Ask a side question about the current conversation without interrupting it
/skin	Show or switch the active CLI skin
/voice on	Enable CLI voice mode (press Ctrl+B to record)
/voice tts	Toggle spoken playback for Nedura replies
/reasoning high	Increase reasoning effort
/title My Session	Name the current session
/status	Show session info — model/profile/tokens/duration — followed by a local Session recap block (recent turn counts, top tools used, files touched, latest user prompt + assistant reply). Pure local compute; no LLM call.
/context [all]	Visual context-usage breakdown — glyph block grid + per-category token table (system prompt / tools / skills / memory / conversation / free space). /context all adds per-skill and per-toolset costs.
/sessions	Open an interactive session picker right inside the classic CLI (same surface the TUI uses). Type to filter, arrow keys to navigate, Enter to resume.

For the full built-in CLI and messaging lists, see Slash Commands Reference.

For setup, providers, silence tuning, and messaging/Discord voice usage, see Voice Mode.

tip

Commands are case-insensitive — /HELP works the same as /help. Installed skills also become slash commands automatically.

Quick Commands

You can define custom commands that run shell commands instantly without invoking the LLM. These work in both the CLI and messaging platforms (Telegram, Discord, etc.).

# ~/.nedura/config.yaml
quick_commands:
  status:
    type: exec
    command: systemctl status hermes-agent
  gpu:
    type: exec
    command: nvidia-smi --query-gpu=utilization.gpu,memory.used --format=csv,noheader
  restart:
    type: alias
    target: /gateway restart

Then type /status, /gpu, or /restart in any chat. See the Configuration guide for more examples.

Preloading Skills at Launch

If you already know which skills you want active for the session, pass them at launch time:

nedura -s hermes-agent-dev,github-auth
nedura chat -s github-pr-workflow -s github-auth

Nedura loads each named skill into the session prompt before the first turn. The same flag works in interactive mode and single-query mode.

Persistent auto-load via config

To have the same skills active at the start of every new session — CLI, TUI, gateway, cron and API sessions alike — set skills.auto_load in config.yaml:

skills:
  auto_load:
    - hermes-agent-dev
    - github-pr-workflow

Each entry is a skill name. The list is resolved once when a session's system prompt is first built and the rendered bytes are reused for the life of the conversation (model switches, compression), so prompt caching stays intact; config edits take effect in the next session. Missing or disabled skills log a warning and are skipped. -s names that overlap the list are loaded once.

--ignore-rules (equivalently NEDURA_IGNORE_RULES=1) skips auto-load together with AGENTS.md, SOUL.md, .cursorrules and memory injection; explicit -s skills still load. The setting is profile-scoped: each profile's config.yaml controls its own list.

Skill Slash Commands

Every installed skill in ~/.nedura/skills/ is automatically registered as a slash command. The skill name becomes the command:

/gif-search funny cats
/axolotl help me fine-tune Llama 3 on my dataset
/github-pr-workflow create a PR for the auth refactor

# Just the skill name loads it and lets the agent ask what you need:
/excalidraw

Personalities

Set a predefined personality to change the agent's tone:

/personality pirate
/personality kawaii
/personality concise

Built-in personalities include: helpful, concise, technical, creative, teacher, kawaii, catgirl, pirate, shakespeare, surfer, noir, uwu, philosopher, hype.

To go back to the default (no overlay), use /personality none — default and neutral work too.

You can also define custom personalities in ~/.nedura/config.yaml:

personalities:
  helpful: "You are a helpful, friendly AI assistant."
  kawaii: "You are a kawaii assistant! Use cute expressions..."
  pirate: "Arrr! Ye be talkin' to Captain Nedura..."
  # Add your own!

Multi-line Input

There are two ways to enter multi-line messages:

1. Alt+Enter, Ctrl+J, or Shift+Enter — inserts a new line
2. Backslash continuation — end a line with \ to continue:

❯ Write a function that:\
  1. Takes a list of numbers\
  2. Returns the sum

Ctrl+J and backslash continuation are enabled by default, matching Claude Code / Codex / OpenCode multiline shortcuts. On supported terminals such as iTerm2, Nedura also requests extended key reporting so Shift+Enter arrives as a distinct newline key. If your terminal sends LF for plain Enter and you need the legacy Ctrl+J-as-submit fallback, opt out:

# ~/.nedura/config.yaml
display:
  cli_multiline_shortcuts: false

info

Pasting multi-line text is supported — use any of the newline keys above, or simply paste content directly.

In terminals using the Kitty keyboard protocol, Alt+Enter on the numeric keypad also inserts a newline, including next to a collapsed paste. Modified keypad navigation keys follow their non-keypad equivalents.

Shift+Enter compatibility

Most terminals send the same byte sequence for Enter and Shift+Enter by default, so applications cannot distinguish them. Nedura recognises Shift+Enter only when the terminal sends a distinct sequence via the Kitty keyboard protocol or xterm's modifyOtherKeys mode.

Terminal	Status
Kitty, foot, WezTerm, Ghostty	Distinct Shift+Enter enabled by default
iTerm2 (recent), Alacritty, VS Code terminal, Warp	Supported once the Kitty protocol is enabled in settings
Windows Terminal Preview 1.25+	Supported once the Kitty protocol is enabled in settings
macOS Terminal.app, stock Windows Terminal (stable)	Not supported — Shift+Enter is indistinguishable from Enter

Where the terminal cannot distinguish them, Alt+Enter and Ctrl+J continue to work by default. On Windows Terminal specifically, Alt+Enter is captured by the terminal (toggles fullscreen) and never reaches Nedura — use Ctrl+Enter (delivered as Ctrl+J) or Ctrl+J directly for a newline.

Redirecting the Agent Mid-Turn

While the agent is working, you can send a correction without starting a new turn:

• Type a new message + Enter — redirects the active turn using your correction
• Ctrl+C — interrupt the current operation (press twice within 2s to force exit); if the prompt holds a draft, the first press clears it and the turn keeps running
• Completed tool work and reasoning already shown stay in context
• A running tool reaches its safe boundary before the correction is applied

Busy Input Mode

The display.busy_input_mode config key controls what happens when you press Enter while the agent is working:

Mode	Behavior
"interrupt" (default)	Your message redirects the active turn. Model generation restarts with displayed reasoning and completed work preserved. A running foreground terminal command is moved to the background (not killed — you get a completion notification) so your message is read immediately; other running tools finish first
"queue"	Your message is queued as the next turn after the agent finishes. The queue lives on the gateway (crash-durable, visible to every attached client).
"steer"	Your message is injected into the current run via /steer, arriving at the agent after the next tool call — no interrupt, no new turn

# ~/.nedura/config.yaml
display:
  busy_input_mode: "steer"   # or "queue" or "interrupt" (default)

"queue" mode prepares a separate follow-up turn. "steer" always waits for the next tool-result boundary. The default "interrupt" mode responds sooner during model generation while avoiding cancellation of a running tool; a long foreground terminal command (a build, a poller) is handed to the background so the agent sees your message right away instead of after the command exits. Use /stop when you want to cancel the turn and its foreground work. Unknown values fall back to "interrupt".

"steer" has two automatic fallbacks: if the agent hasn't started yet, or if images are attached, the message falls back to "queue" behavior so nothing is lost.

Whatever the mode, /queue <prompt> queues a follow-up turn explicitly, and /queue list, /queue rm N, /queue edit N … and /queue move A B act on the pending queue immediately, even mid-run. The live work dock lists what is waiting.

When the CLI runs on the gateway's session (the default local setup), a prompt queued while a turn runs, by /queue or by queue-mode Enter, goes into the gateway's own queue for the session. It survives a gateway restart and runs once, in the order you entered it. Image follow-ups still wait in the CLI. That queue is shared: /queue list, rm, edit and clear show and change every prompt queued on the session, including ones another window or an automation queued. The gateway runs its queue in order and cannot reorder it, so /queue move refuses to move a prompt the gateway holds. A slash command typed while busy after your own /queue prompt runs after that prompt's turn starts; it waits at most a couple of seconds, so a paused queue never blocks typing.

You can also change it inside the CLI:

/busy queue
/busy steer
/busy interrupt
/busy status

First-touch hint

The first time you press Enter while Nedura is working, Nedura prints a one-line reminder explaining the /busy knob. It only fires once per install; onboarding.seen.busy_input_prompt in config.yaml records that it was shown. Delete that key to see the tip again.

Suspending to Background

On Unix systems, press Ctrl+Z to suspend Nedura to the background — just like any terminal process. The shell prints a confirmation:

Nedura has been suspended. Run `fg` to bring Nedura back.

Type fg in your shell to resume the session exactly where you left off. This is not supported on Windows.

Tool Progress Display

The CLI shows animated feedback as the agent works:

Thinking animation (during API calls):

  ◜ (｡•́︿•̀｡) pondering... (1.2s)
  ◠ (⊙_⊙) contemplating... (2.4s)
  ✧٩(ˊᗜˋ*)و✧ got it! (3.1s)

Tool execution feed:

  ┊ 💻 terminal `ls -la` (0.3s)
  ┊ 🔍 web_search (1.2s)
  ┊ 📄 web_extract (2.1s)

Cycle through display modes with /verbose: off → new → all → verbose. This command can also be enabled for messaging platforms — see configuration.

Tool Preview Length

The display.tool_preview_length config key controls the maximum number of characters shown in tool call preview lines (e.g. file paths, terminal commands). The default is 0, which means no limit — full paths and commands are shown.

# ~/.nedura/config.yaml
display:
  tool_preview_length: 80   # Truncate tool previews to 80 chars (0 = no limit)

This is useful on narrow terminals or when tool arguments contain very long file paths.

Session Management

Resuming Sessions

When you exit a CLI session, a resume command is printed:

Resume this session with:
  nedura --resume 20260225_143052_a1b2c3

Session:        20260225_143052_a1b2c3
Duration:       12m 34s
Messages:       28 (5 user, 18 tool calls)

Resume options:

nedura --continue                          # Resume the most recent CLI session
nedura -c                                  # Short form
nedura -c "my project"                     # Resume a named session (latest in lineage)
nedura -c "my project" --create-if-missing # Same, creating the titled session if none exists
nedura --resume 20260225_143052_a1b2c3     # Resume a specific session by ID
nedura --resume "refactoring auth"         # Resume by title
nedura --resume latest                     # Resume the most recent session (same as -c)
nedura --resume latest --in ./my-project   # Latest session for ./my-project's workspace
nedura -r 20260225_143052_a1b2c3           # Short form

Resuming restores the full conversation history from SQLite. The agent sees all previous messages, tool calls, and responses — just as if you never left.

Use /title My Session Name inside a chat to name the current session, or nedura sessions rename <id> <title> from the command line. Use nedura sessions list to browse past sessions.

Session Storage

CLI sessions are stored in Nedura's SQLite state database under ~/.nedura/state.db. The database keeps:

• session metadata (ID, title, timestamps, token counters)
• message history
• lineage across compressed/resumed sessions
• full-text search indexes used by session_search

Some messaging adapters also keep per-platform transcript files alongside the database, but the CLI itself resumes from the SQLite session store.

Context Compression

Long conversations are automatically summarized when approaching context limits:

# In ~/.nedura/config.yaml
compression:
  enabled: true
  threshold: 0.50    # Compress at 50% of context limit by default

# Summarization model configured under auxiliary:
auxiliary:
  compression:
    model: ""  # Leave empty to use the main chat model (default). Or pin a cheap fast model, e.g. "google/gemini-3-flash-preview".

When compression triggers, middle turns are summarized while the first 3 and last 20 turns are always preserved.

Background Sessions

Run a prompt in a separate background session while continuing to use the CLI for other work:

/bg Analyze the logs in /var/log and summarize any errors from today

Nedura immediately confirms the task and gives you back the prompt:

🔄 Background task #1 started: "Analyze the logs in /var/log and summarize..."
   Task ID: bg_143022_a1b2c3

How It Works

Each /bg prompt spawns a completely separate agent session in a daemon thread:

• Isolated conversation — the background agent has no knowledge of your current session's history. It receives only the prompt you provide.
• Same configuration — the background agent inherits your model, provider, toolsets, reasoning settings, and fallback model from the current session.
• Non-blocking — your foreground session stays fully interactive. You can chat, run commands, or even start more background tasks.
• Multiple tasks — you can run several background tasks simultaneously. Each gets a numbered ID.

Results

When a background task finishes, the result appears as a panel in your terminal:

╭─ ◆ Nedura (background #1) ──────────────────────────────────╮
│ Found 3 errors in syslog from today:                         │
│ 1. OOM killer invoked at 03:22 — killed process nginx        │
│ 2. Disk I/O error on /dev/sda1 at 07:15                      │
│ 3. Failed SSH login attempts from 192.168.1.50 at 14:30      │
╰──────────────────────────────────────────────────────────────╯

If the task fails, you'll see an error notification instead. If display.bell_on_complete is enabled in your config, the terminal bell rings when the task finishes.

Use Cases

• Long-running research — "/bg research the latest developments in quantum error correction" while you work on code
• File processing — "/bg analyze all Python files in this repo and list any security issues" while you continue a conversation
• Parallel investigations — start multiple background tasks to explore different angles simultaneously

info

Background sessions do not appear in your main conversation history. They are standalone sessions with their own task ID (e.g., bg_143022_a1b2c3).

Quiet Mode

By default, the CLI runs in quiet mode which:

• Suppresses verbose logging from tools
• Enables kawaii-style animated feedback
• Keeps output clean and user-friendly

For debug output:

nedura chat --verbose