Skip to content

CLI Command Surface

Every command, subcommand, alias, and flag the CLI exposes, derived from the live command tree (not maintained by hand). For task-oriented, copy-pasteable usage see the curated COMMANDS.md.

nuvira

Nuvira — multi-agent AI coding CLI (local models & cloud APIs)

  • flags: --debug, --task <task>, --version

nuvira admin

Admin governance policy for Auto routing (P6 M6.5) — allow/deny providers & models, hard cost cap, PII privacy, unblock control

nuvira admin policy

Show the current governance policy + enforcement status

  • flags: --json

nuvira admin allow

Add providers to the allow-list (empty = all providers allowed)

nuvira admin deny

Add providers to the deny-list (wins over the allow-list)

nuvira admin allow-model

Add models to the allow-list — a provider survives only if one of its candidate models is listed

nuvira admin deny-model

Add models to the deny-list (wins over the allow-list)

nuvira admin max-cost

Set the admin hard max cost per call (USD); joins routing.maxCostUsd (stricter wins)

nuvira admin pii-min

Set the minimum privacy score (0-1) required when a task matches a PII pattern (default 1.0 = local-only)

nuvira admin unblock

Control whether nuvira models unblock may override REGISTRY-learned blocks (false = admin-hard)

nuvira admin clear

Remove one governance rule (the policy becomes permissive on that field)

nuvira admin role

Manage RBAC roles over the admin surface (P6 M6.1 — requires admin role)

nuvira admin role add

Assign a role to a user (first assignment exits legacy single-user mode)

nuvira admin role remove

Remove a user's role assignment (if none remain, legacy permissive mode resumes)

nuvira admin role list

List all role assignments

nuvira admin whoami

Show the current identity and its effective RBAC permissions

nuvira admin cron

Scheduled tool invocations (J2 — cron/jobs.py)

nuvira admin cron add

Add a cron job: nuvira admin cron add nightly-build "0 3 * * *" build

  • flags: --args <args>, --channel <channel>, --dry-run

nuvira admin cron list

List cron jobs with schedule + next run

nuvira admin cron remove

Remove a cron job

nuvira admin cron run

Run a cron job NOW (invoke its tool immediately)

nuvira chat

Start an interactive chat session with AI

  • flags: --dev, --file <file>, --model <model>, --no-cache, --provider <provider>

nuvira edit

Edit a file using AI assistance

  • flags: --auto-route, --dry-run, --instruction <instruction>, --model <model>, --provider <provider>, --review

nuvira plan

Generate an implementation plan for a codebase task

  • flags: --model <model>, --provider <provider>, --task <task>, --verbose

nuvira config

Manage Nuvira configuration

nuvira config set

Set a configuration value

nuvira config get

Get a configuration value

nuvira config list

List all providers and their status

nuvira config init

Initialize configuration interactively

nuvira config vault

Secret vault management (Phase A1)

nuvira config vault status

Show the active vault tier and migration state

nuvira config vault log

Show recent vault access-log entries (K3 tamper-evident audit)

  • flags: --limit <limit>

nuvira config vault migrate-keys

Move plaintext provider API keys from buffconfig.json into the vault

nuvira config gateway

Manage gateway platform transports (tokens written to ~/.nuvira/.env)

nuvira config gateway list

Show every platform transport and its env-var status

nuvira config gateway set

Configure a platform transport (interactive wizard, or --set VAR=value)

  • flags: --set <set>

nuvira config gateway remove

Remove a platform transport from the env file

  • flags: --yes

nuvira config gateway allow

Allow a user/group to trigger the agent on a platform (written to gateway.policies in config)

nuvira config gateway disallow

Remove a user/group from the allowed list of a platform

nuvira config gateway reply

Set how unapproved senders are handled on a platform: polite (⛔ message) or silent (no reply)

nuvira config gateway send-authority

Manage who may command the agent to send to OTHER people (gateway_send). Outbound-only gate — separate from allow (who may trigger).

nuvira config gateway notify

Manage status recipients — contacts/groups that ALWAYS get pipeline completion summaries

nuvira config gateway ask-user-wait

Ask-and-wait for clarifying questions on messaging channels: when a turn asks the sender a question, hold it for their reply instead of assuming option 1

nuvira cache

Manage inference cache

nuvira cache stats

Show cache statistics

nuvira cache clear

Clear all cached responses

nuvira models

List available models from inference providers

  • flags: --all, --json, --provider <provider>, --search <search>, --verify

nuvira models refresh

Probe providers and spot-check models, updating the Model Availability Registry

  • flags: --json, --no-spot-check

nuvira models status

Show the Model Availability Registry (verified / unavailable / quota-parked models)

  • flags: --json, --verbose

nuvira models unblock

Manually release a registry-blocked provider (escape hatch) and re-probe it against the live API

  • flags: --json, --no-spot-check

nuvira models excluded

Show which providers routing is currently skipping, and why (failure cooldowns, registry blocks, governance policy)

  • flags: --json

nuvira models staleness

Show model staleness: last probe time, days since verification, and removal risk

  • flags: --json

nuvira models watch

Run the model-registry maintenance daemon: probe + spot-check on a schedule

  • flags: --interval <interval>, --no-spot-check

nuvira execute

Run a multi-agent pipeline to accomplish a goal

  • flags: --auto-branch, --auto-route, --checkpoint, --checkpoint-list, --context-limit <context_limit>, --context-prune <context_prune>, --dry-run, --engine <engine>, --gatherer-model <gatherer_model>, --json-events, --max-repairs <max_repairs>, --memory, --memory-clear, --memory-stats, --model <model>, --no-tool-calling, --plan-mode <plan_mode>, --planner-model <planner_model>, --provider <provider>, --repair-fallback-models <repair_fallback_models>, --repair-mode <repair_mode>, --resume [resume], --review, --reviewer-model <reviewer_model>, --sandbox, --skip-tests, --tool-calling, --verbose, --writer-model <writer_model>

nuvira run

Execute a shell command and show output (lightweight alternative to the full pipeline)

  • flags: --timeout <timeout>, --verbose

nuvira workflow

Run, manage, and share workflow templates

nuvira workflow list

Show available workflow templates (built-in + installed)

nuvira workflow run

Run a workflow template

  • flags: --dry-run, --model <model>, --provider <provider>, --verbose

Search the GitHub workflow template registry

  • flags: --refresh

nuvira workflow install

Install a workflow template from the GitHub registry

nuvira workflow publish

Prepare a local workflow template for publishing to the registry

nuvira workflow info

Show detailed information about a registry template

nuvira workflow upgrade

Check for and apply template upgrades from the registry

nuvira whatsapp

WhatsApp bridge (Baileys) — pair a personal number via QR and check status (I8)

nuvira whatsapp pair

Pair WhatsApp (personal number — no Meta Business account, no paid API)

  • flags: --phone <phone>, --timeout <timeout>

nuvira whatsapp status

Show bridge session path + pairing state

nuvira whatsapp contacts

List resolvable contacts (mapping file + learned) — connects briefly to learn names from live traffic

  • flags: --timeout <timeout>

nuvira whatsapp contact

Manage the WhatsApp contact-name mapping (send by name: nuvira gateway send whatsapp: "message")

nuvira whatsapp contact add

Map a display name to a number (E.164, no +): nuvira whatsapp contact add Name 919876543210

nuvira whatsapp contact remove

Remove a mapped contact name

nuvira plugins

Manage provider plugins, agent plugins, and workflow templates

nuvira plugins list

List all discovered plugins and workflows

nuvira plugins scan

Force re-scan all plugin directories

nuvira learn

Self-improvement system — agent stats, patterns, and optimization

nuvira learn stats

Show per-agent performance statistics

nuvira learn patterns

Show extracted coding patterns

  • flags: --extract, --model <model>, --provider <provider>

nuvira learn lessons

Show failure lessons — what past runs learned from mistakes

  • flags: --extract, --model <model>, --provider <provider>

nuvira learn optimize

Generate optimized model-to-agent routing recommendations

nuvira learn status

Show overall self-improvement status

nuvira learn clear

Reset all learning data (stats, patterns, memory)

  • flags: --force

nuvira learn compare

Compare benchmark results between two models

  • flags: --all, --last

nuvira learn feedback

Rate a trajectory or view feedback statistics

  • flags: --comment <comment>, --rating <rating>, --stats, --trajectory <trajectory>

nuvira learn quality

Show pattern quality and decay metrics

  • flags: --details

nuvira learn gc

Garbage-collect low-quality patterns

  • flags: --dry-run

nuvira init

Scaffold a new project from a template

  • flags: --list, --model <model>, --provider <provider>, --template <template>, --template-dir <template_dir>

nuvira stats

View usage statistics and cost tracking

nuvira stats cost

Show API cost tracking details

  • flags: --clear

nuvira stats history

Show conversation history statistics

nuvira history

Browse and search conversation history

nuvira history list

Show all saved conversations

  • flags: --limit <limit>

Search conversations by keyword or semantic similarity

  • flags: --limit <limit>, --semantic

nuvira history show

Show a specific conversation

nuvira history clear

Clear all conversation history

nuvira history prune

Remove conversations older than the retention period

  • flags: --days <days>

nuvira history reindex

Rebuild the semantic search index for all past conversations (embeds each session for vector search)

nuvira skill

Manage and run compiled skills — reusable execution plans derived from past trajectories

nuvira skill list

List all compiled skills

  • flags: --quality <quality>

nuvira skill show

Show detailed skill definition

nuvira skill run

Run a skill — resolves parameters and invokes the multi-agent pipeline

  • flags: --auto-route, --dry-run, --memory, --model <model>, --params <params>, --provider <provider>, --verbose

nuvira skill compile

Force skill compilation from stored trajectories

  • flags: --model <model>, --provider <provider>

Search skills by name, tag, or description

nuvira skill gc

Garbage-collect low-quality skills

  • flags: --dry-run

nuvira skill quality

Show skill quality and decay metrics

  • flags: --details

nuvira skill clear

Remove all compiled skills

  • flags: --force

nuvira skills

Skills hub — search, install, update, and audit community skills (J3)

Search all configured skill registries (I7 P1 multi-source)

  • flags: --refresh, --source <source>

nuvira skills install

Install a skill from the configured registries into /.agents/skills/ (sandboxed, checksummed)

  • flags: --project <project>, --source <source>

nuvira skills uninstall

Uninstall a skill: removes .agents/skills/ and its provenance record

  • flags: --project <project>

nuvira skills update

Update installed skills to newer registry versions (checksum-verified)

  • flags: --project <project>

nuvira skills bundle

P6b — skill bundles: group skills under one id and load them together (Hermes parity)

  • flags: --create, --delete, --description <description>, --name <name>, --skills <skills>

nuvira skills list

List installed skills with provenance (origin: registry vs local)

  • flags: --origin <origin>, --project <project>

nuvira gateway

Multi-channel gateway — talk to the agent from Telegram/Discord/Slack/WhatsApp/Email/Signal (J1)

nuvira gateway status

Show configured adapters and reachable channels

nuvira gateway send

Send a message to a channel alias or platform:channelId

nuvira gateway send-media

Send a media file (image/video/audio/document) — type from the file extension; WhatsApp/Telegram/Discord

  • flags: --caption <caption>

nuvira gateway alias

Manage channel aliases

nuvira gateway alias add

Register an alias: nuvira gateway alias add ops slack C0123

nuvira gateway alias remove

Remove an alias

nuvira gateway contact

Manage contacts for outbound messaging (name → platform:id resolution)

nuvira gateway contact list

List all contacts with status, platform, and ID

  • flags: --pending, --platform <platform>

nuvira gateway contact approve

Approve a contact for outbound messaging

  • flags: --platform <platform>

nuvira gateway contact reject

Reject a contact (block outbound messages)

  • flags: --platform <platform>

nuvira gateway contact delete

Delete a contact permanently

  • flags: --platform <platform>

nuvira gateway contact add

Manually add a contact (e.g. nuvira gateway contact add Anuj telegram 616825477)

  • flags: --phone <phone>

nuvira gateway delivery

Show the delivery ledger (failed sends awaiting retry) and optionally drain it

  • flags: --flush

nuvira gateway logs

Show the structured gateway log (send failures, refused senders, pipeline outcomes)

  • flags: --event <event>, --limit <limit>, --path

nuvira gateway history

Manage per-contact conversation history (gateway chat memory)

nuvira gateway history list

List all stored conversations with last message preview

  • flags: --platform <platform>

nuvira gateway history show

Show conversation history for a contact (e.g. nuvira gateway history show whatsapp:918800663237)

  • flags: --limit <limit>

nuvira gateway history clear

Clear conversation history for a contact

nuvira gateway history prune

Remove conversations older than 7 days

nuvira gateway start

Run all configured adapters in the foreground (Ctrl-C to stop; --supervise auto-restarts on crash)

  • flags: --host <host>, --no-events, --port <port>, --supervise

nuvira gateway stop

Stop a running gateway gracefully (SIGTERM — from any terminal)

  • flags: --port <port>

nuvira gateway setup

Interactive setup wizard for a messaging platform (e.g. nuvira gateway setup telegram)

nuvira model

Manage inference providers and models — switch, list, inspect, and recommend

nuvira model list (aliases: ls)

List all providers and their configuration status

  • flags: --all, --json

nuvira model switch

Switch active provider/model (interactive or via argument). Use auto for smart routing

  • flags: --model <model>, --provider <provider>

nuvira model info

Show current active provider and model configuration

  • flags: --verbose

nuvira model recommend

Show model routing recommendations

nuvira model explain

Explain Auto model routing — why a provider/model would be picked for a task

  • flags: --agent <agent>, --json, --since <since>

nuvira model health

Quick health check for the currently active provider

  • flags: --provider <provider>, --verbose

nuvira model bandit

Show learning-router bandit state (Thompson-sampling priors per provider × complexity bucket). Action: reset

  • flags: --json

nuvira model ml

Show the ML task-similarity router state (learned outcomes per provider, kNN over task features). Action: reset

  • flags: --json

nuvira model quota

Show the central quota ledger (tokens/requests per provider × model, reset windows, parked state). Actions: reset | set | clear

  • flags: --cost-usd <cost_usd>, --json, --requests <requests>, --tokens <tokens>, --window-ms <window_ms>

nuvira benchmark

Run standardized model benchmarks against coding tasks

nuvira benchmark run

Run the benchmark suite

  • flags: --budget <budget>, --format <format>, --model <model>, --provider <provider>, --routing, --tasks <tasks>

nuvira benchmark list

List available benchmark tasks

nuvira benchmark results

Show previous benchmark results

  • flags: --compare, --format <format>, --last

nuvira benchmark clear

Clear all benchmark data

nuvira eval

Run the Agent-Nuvira evaluation framework — measures if the agent is actually improving

nuvira eval run

Run the evaluation suite

  • flags: --budget <budget>, --engine <engine>, --format <format>, --keep-workspaces, --model <model>, --pace, --provider <provider>, --routing, --suite <suite>, --tasks <tasks>

nuvira eval list

List available eval tasks

nuvira eval results

Show previous eval runs

  • flags: --compare, --format <format>, --last

nuvira eval score

Show the evaluation scoring rules

nuvira eval clear

Clear all eval data

nuvira sandbox

Manage Docker sandbox isolation for code execution

nuvira sandbox status

Check Docker availability and sandbox status

nuvira sandbox config

Show or update sandbox configuration

  • flags: --cpu <cpu>, --disable, --disk <disk>, --enable, --image <image>, --memory <memory>, --network, --timeout <timeout>

nuvira sandbox images

List available pre-defined sandbox images

nuvira sandbox run

Run a command inside a new sandbox container

  • flags: --cpu <cpu>, --image <image>, --memory <memory>, --network, --project <project>, --timeout <timeout>

nuvira sandbox cleanup

Destroy all active sandbox containers

nuvira doctor

Run diagnostic checks on all provider configurations and system health

  • flags: --enterprise, --fix, --nuvira, --provider <provider>, --verbose, --watch

nuvira memory

Manage agent memory store — compression, pruning, and optimization

nuvira memory stats

Show memory store statistics

nuvira memory optimize

Run automatic memory compression and pruning

  • flags: --aggressive, --dry-run

nuvira memory prune

Prune old or low-quality trajectories

  • flags: --max-age <max_age>, --max-count <max_count>, --min-score <min_score>, --verbose

nuvira memory summarize

Summarize old trajectories by merging similar ones

  • flags: --retention <retention>, --verbose

nuvira memory info

Show detailed compression analysis

nuvira memory backend

Show the active vector-search backend and why it was chosen

  • flags: --check

nuvira memory list

List all memory entries (facts, preferences, lessons, observations)

  • flags: --limit <limit>, --type <type>

Search memory entries by content

  • flags: --limit <limit>, --type <type>

nuvira memory add

Add a memory entry

  • flags: --tags <tags>, --type <type>

nuvira memory delete

Delete a memory entry by ID

nuvira memory export

Export all memories to a JSON file

  • flags: --output <output>

nuvira memory import

Import memories from a JSON file

  • flags: --merge

nuvira memory facts

Fact & preference memory (project-scoped, cross-session)

nuvira memory facts list

List facts, optionally for a project (git slug or cwd:)

  • flags: --project <project>

nuvira memory facts add

Manually store a fact for the current project

  • flags: --project <project>, --tags <tags>

nuvira memory facts stats

Show fact-store statistics

nuvira memory clear

Clear all stored trajectories and reset memory

  • flags: --force

nuvira dashboard

Launch the web-based dashboard for visualizing agent execution and system status

  • flags: --build, --cwd <cwd>, --force, --gateway-port <gateway_port>, --host <host>, --keep-gateway, --no-gateway, --no-open, --port <port>

nuvira dashboard stop

Stop a running dashboard gracefully (SIGTERM — from any terminal)

  • flags: --port <port>

nuvira agent

Scaffold and manage custom agent-baba-d agents

nuvira agent create

Create a new custom agent project from a template

  • flags: --agent-name <agent_name>, --description <description>, --dir <dir>

nuvira agent list

List all discovered custom agent plugins

nuvira agent info

Show details about a discovered agent plugin

nuvira federation

Connect to and manage remote agent instances

nuvira federation status

Show federation connection status and configuration

nuvira federation start

Start the federation server (listens for incoming connections)

  • flags: --auth <auth>, --daemon, --host <host>, --oidc-public-key <oidc_public_key>, --port <port>, --secret <secret>

nuvira federation connect

Connect to a remote federation server

  • flags: --port <port>, --secret <secret>

nuvira federation disconnect

Disconnect from the remote federation server

nuvira federation run

Run a task on the remote federation server

  • flags: --agent <agent>, --model <model>, --no-stream, --provider <provider>, --timeout <timeout>

nuvira federation health

Check the health of the remote federation server

nuvira federation config

Show or update federation configuration

  • flags: --set-port <set_port>, --set-secret <set_secret>, --show

nuvira federation a2a

A2A (Agent-to-Agent) protocol — discover and connect to external A2A-compliant agents

nuvira federation a2a discover

Discover an A2A-compliant agent and fetch its AgentCard

nuvira federation a2a start

Start the A2A server (listens for incoming A2A connections)

  • flags: --host <host>, --port <port>

nuvira federation a2a status

Check the health and status of a remote A2A agent

nuvira federation a2a run

Delegate a task to a remote A2A-compliant agent

  • flags: --agent <agent>, --skill <skill>, --timeout <timeout>

nuvira team

Team collaboration — shared config, git-synced memory, and review workflow

nuvira team init

Initialize team configuration in the working directory

  • flags: --branch <branch>, --repo <repo>

nuvira team join

Clone and join an existing team repository

nuvira team sync

Sync team memory with remote (pull latest + push local changes)

nuvira team status

Show team configuration and memory status

nuvira team share

Share local trajectories and patterns with the team

nuvira team review

Manage review bundles — agent PR → review → merge workflow

nuvira team review list

List all review bundles

  • flags: --limit <limit>

nuvira team review show

Show a specific review bundle with full details

nuvira team review approve

Approve a review bundle (sets status to approved)

  • flags: --message <message>

nuvira team review request-changes

Request changes on a review bundle

nuvira team review reject

Reject a review bundle

nuvira team review merge

Merge an approved review into the working directory

nuvira team review create

Create a review bundle from specified files

  • flags: --files <files>, --model <model>, --provider <provider>

nuvira sdk

Create and manage custom agents with the Agent-Nuvira SDK

nuvira sdk scaffold

Generate a new custom agent project

  • flags: --agent-type <agent_type>, --template <template>

nuvira sdk templates

List available scaffold templates

nuvira sdk info

Show SDK package info and version

nuvira sdk register

Register a custom agent with the orchestrator

  • flags: --icon <icon>, --orchestrator-path <orchestrator_path>

nuvira sdk unregister

Remove a custom agent from the orchestrator

  • flags: --orchestrator-path <orchestrator_path>

nuvira provider

List and check health of all inference providers

nuvira provider list

Show all providers with color-coded status table

  • flags: --all

nuvira provider health

Show detailed health checks for one or all providers

  • flags: --verbose, --watch

nuvira security

Scan code, prompts, or files for security issues

nuvira security scan

Scan for PII, injection attempts, or dangerous code patterns

  • flags: --code, --file <file>, --generated, --json, --pii, --prompt, --stdin, --strict

nuvira audit

Verify and export the tamper-evident (hash-chained) audit trail (P6 M6.3)

nuvira audit verify

Verify hash-chain integrity of audit stores (tamper detection)

  • flags: --file <file>, --json

nuvira audit export

Export an audit store as SIEM-friendly CEF lines

  • flags: --file <file>, --out <out>

nuvira sbom

Generate and verify the CycloneDX software bill of materials (P6 M6.6)

  • flags: --json, --licenses, --out <out>, --reproducible, --sbom <sbom>, --verify

nuvira feedback

Record, view, and manage user feedback on agent outputs

nuvira feedback record

Record feedback for a trajectory (interactive)

  • flags: --comment <comment>, --negative, --neutral, --positive

nuvira feedback list

Show recent feedback entries

  • flags: --limit <limit>, --trajectory <trajectory>

nuvira feedback stats

Show aggregated feedback statistics

nuvira feedback clear

Clear all feedback data

nuvira nlu

NLU request understanding — debug intent/entity/action resolution

nuvira nlu debug

Debug how a request is understood (rule path + optional LLM verify)

  • flags: --llm

nuvira nlu learnings

Show the routing corrections learned from CONFIRMED misreadings (the agent improving itself)

  • flags: --forget <forget>, --json

nuvira intent

Plain-English → CLI routing — resolve an ask into the exact nuvira command(s) to run

nuvira intent resolve

Resolve a plain-English ask (e.g. "stop the dashboard", "add Rahul to whatsapp") into CLI commands

  • flags: --json, --semantic

nuvira intent eval

Score deterministic vs semantic matching over a labeled ask corpus (novel phrasings included)

  • flags: --json

nuvira code-map

Project symbol map — functions, classes, methods with line numbers (AST engine)

  • flags: --json

nuvira tools

Inspect the agent tool registry (pipeline + experience tools)

nuvira tools list

List every registered tool (the H1 tool-calling surface)

nuvira tools show

Show a tool's description and input schema

nuvira tools toolsets

List toolset groups (capability gating) with enabled state; enable/disable a group

  • flags: --disable <disable>, --enable <enable>

nuvira session

Project-scoped session continuity (debug surface — continue is the primary path)

nuvira session list

List recent sessions, optionally filtered by project + time range

  • flags: --limit <limit>, --project <project>, --since <since>

nuvira session summarize

Show summary + metadata for one session (transcript: nuvira history show)

nuvira session resume

Run D1 auto-recall for a project and print the recall card

  • flags: --project <project>, --since <since>

nuvira marketplace

Browse and install community plugins and workflow templates

nuvira marketplace browse

Browse all available marketplace items

  • flags: --plugins, --refresh, --workflows

Search across plugins and workflow templates

nuvira marketplace install

Install a workflow template from the registry

nuvira marketplace info

Show detailed information about a marketplace item

nuvira mcp

Manage MCP (Model Context Protocol) server connections

nuvira mcp list

List all discovered MCP servers and their tools

nuvira mcp connect

Connect to an MCP server

  • flags: --all

nuvira mcp call

Call a tool on an MCP server

  • flags: --args <args>, --server <server>

nuvira mcp info

Show detailed information for an MCP server

nuvira mcp refresh

Re-discover and reconnect to all MCP servers

nuvira mcp catalog

List the curated MCP server catalog (vetted, exact-version pins)

  • flags: --search <search>

nuvira mcp install

Install a vetted MCP server from the catalog (writes ~/.nuvira/mcp/.json)

  • flags: --env <env>

nuvira mcp uninstall

Remove an installed MCP server config

nuvira mcp serve

Expose the agent's H1 tools as an MCP server over stdio (MCP clients / IDEs / other agents can connect)

  • flags: --with <with>

nuvira ci

Headless CI/CD mode — structured JSON output and exit codes for pipelines

nuvira ci execute

Execute a goal and emit JSON result (exit 0 = success, 1 = failure)

  • flags: --context-limit <context_limit>, --context-prune <context_prune>, --github-annotations, --memory, --model <model>, --planner-model <planner_model>, --provider <provider>, --reviewer-model <reviewer_model>, --sandbox, --timeout <timeout>, --writer-model <writer_model>

nuvira ci check

Run a gate check (exit 0 = pass, 1 = fail) — minimal output, ideal for workflow gates

  • flags: --model <model>, --provider <provider>, --verbose

nuvira ci review

Review one or more files and emit JSON findings

  • flags: --context <context>, --format <format>, --model <model>, --provider <provider>

nuvira publish

Autonomous publish workflow — version, build, publish to npm & GitHub

  • flags: --dry-run, --major, --minor, --model <model>, --patch, --provider <provider>, --skip-tests, --verbose

nuvira credentials

Store GitHub/npm release credentials so the agent can release without re-entering them

nuvira credentials status

Show which release credentials are available, and where each came from

nuvira credentials set

Store a credential in the nuvira credential store

  • flags: --stdin, --value <value>

nuvira credentials forget

Delete a stored credential

nuvira credentials verify

Check the stored GitHub and npm tokens against the live services

nuvira bedrock

AWS Bedrock setup and management (dedicated onboarding for Bedrock)

nuvira bedrock setup

Interactive wizard to configure AWS Bedrock (credentials, region, model access)

nuvira bedrock status

Show current Bedrock configuration and connectivity status

nuvira bedrock test

Probe Bedrock models and test inference

  • flags: --region <region>

nuvira phase

Phase-wise project scope execution — multi-goal pipelines

nuvira phase create

Create a new phase scope with ordered goals

  • flags: --output <output>

nuvira phase execute

Execute a phase scope from start to finish

  • flags: --dry-run, --model <model>, --non-interactive, --provider <provider>, --skip-tests, --verbose

nuvira phase resume

Resume a saved phase scope from where it left off

  • flags: --model <model>, --provider <provider>, --skip-tests, --verbose

nuvira phase status

Show progress of a phase scope

nuvira phase delete

Delete a saved phase scope

nuvira phase list

List all saved phase scopes

nuvira retrieval

Vector retrieval — token-efficient context via local embeddings

  • flags: --verbose

nuvira retrieval stats

Show token-savings transparency (how many tokens retrieval saved)

nuvira retrieval index

Index a repo/file into the retrieval store (chunks embedded locally)

nuvira retrieval query

Semantic search over the indexed repo (top-k chunks)

nuvira retrieval clear

Clear the retrieval index and token-savings stats

nuvira trace

Inspect and replay per-step reasoning traces (every LLM call in a pipeline)

nuvira trace list

Show recent traces

  • flags: --limit <limit>

nuvira trace show

Show a single trace (goal, timing, step summary)

nuvira trace replay

Step-by-step replay of a trace — every LLM call with prompt digest, model, tokens, latency, and routing

  • flags: --full

nuvira trace clear

Delete all stored traces

429 commands (incl. subcommands) · generated from the live CLI — this file is the drift-guarded surface.