Slash commands - full reference
Navin ships built-in slash commands. Type / in any chat composer to open the command palette. Commands fall into three lifecycles:
- Agent workflow - rewrites your message into a mission brief (with the right skills preloaded) and runs a normal agent turn.
- Side channel - answers immediately without consuming an agent turn.
- Turn control - manages the current chat/turn itself.
Most security / quality workflows are also available from the Code module Actions menu. Composer turn modes (Plan / Agent / Review / Security / Debug) are documented in Modes.
Agent workflows
Build & quality
| Command | Title | Arguments | What it does |
|---|---|---|---|
/blueprint | Plan mode | [task] | Designs an implementation plan and mission Task Ledger before writing any code: constraints, facts, missing info, acceptance criteria, ordered steps with validation. No code changes. Sets composer mode to Plan. See Plan Mode. |
/forge | Build mode | [task] | Full autonomous build: Progress Ledger loop (claim → work → validate with evidence → next), local replan on stall. Sets composer mode to Agent. |
/cruise | Autopilot build | [task] | Plan, execute, test, detect stalls/loops, replan locally until done or paused - no Build wait. Sets composer mode to Agent. |
/mission | Long mission | [goal] | Durable multi-turn mission with ledger, checkpoints, resume, sub-agents, and a final report. Sets composer mode to Agent. |
/mobile | Run Mobile | [android|ios|web|metro|doctor] | Detect Expo / React Native / Flutter, doctor the toolchain, start the packager (and optional Android emulator), then use the Mobile preview tab to see and interact with the device. |
/atlas | Project atlas | [init|refresh|query <question>] | Builds or refreshes .metadata/ (role, kind, and dependencies of every file) powering the Dev workbench Graph tab, and answers "where is X?" questions from the index instead of grepping. |
/inspect | Code review | [path|diff|scope] | Expert review (Review mode): correctness, SQL/data, API, frontend, security smells, tests, perf, maintainability. Tools: code_review (scope/filter/report). Each finding needs file:line, impact, fix, and a real example. Closes with review-report-*.html (auto File Preview) and a numbered remediation plan - asks which # to start. Optional pr_comments (PR + gh). Read-only unless asked to fix. |
/debug | Debug mode | [signal|path|scope] | Reproduce the failure, prove root cause with real evidence (stack/logs/test/DebugMCP). Tools: debug_repair (mcp_status/start_branch/report). Closes with debug-report-*.html (auto File Preview) and numbered fix choices. Asks which # to start. Applies code only if asked. |
/turbo | Performance audit | [path|scope] | Profiles and optimizes: hot paths, N+1 queries, blocking I/O, missing caches, bundle size, memory. Measures before recommending. |
/pulse | Metrics & analytics | [focus] | Project health scoreboard: code size/complexity, dependency freshness, lint findings, coverage, TODO debt, KPIs. |
/board | Task board | [task <id>|loop|status|plan <goal>] | Works the shared project board: plan, take tasks, dispatch, and report. |
/goal | Long-running goal | <goal> | Registers the request as a sustained goal the agent keeps pursuing across turns until completed or stopped. |
Security audit (defensive)
| Command | Title | Arguments | What it does |
|---|---|---|---|
/fortify | Security review | [path|scope] | Expert AppSec pass (Security mode): surface map, security_scan, host scanners when present, injection/data, frontend, authz, network, supply chain, privacy, LLM/agent. Each finding needs severity, proof, real PoC, minimal fix. Closes with security-report-*.html (auto File Preview) and numbered hardening choices - asks which # to start. Optional pr_comments. Read-only unless asked to fix. |
/probe | Vulnerability scan | [path|scope] | Hunts exploitable weaknesses: OWASP Top 10, hardcoded secrets, vulnerable dependency versions, SSRF/path traversal, prompt injection. |
/unmask | Secrets scan | [path|scope] | Leaked keys, tokens and passwords in code, config and git history. |
/lineage | Supply-chain audit | [path|scope] | Dependencies: CVEs, outdated packages, lockfile integrity, licenses. |
/xray | Deep static analysis | [path|scope] | SAST: traces user input to dangerous sinks across the code. |
/gatekeeper | Access control audit | [path|scope] | Authentication and authorization: sessions, tokens, roles, IDOR. |
/perimeter | API & web surface | [path|scope] | Exposed endpoints: authz, CORS, CSRF, SSRF, headers, rate limits. |
/bastion | Infra & IaC audit | [path|scope] | Docker, Kubernetes, Terraform, CI/CD and cloud configuration. |
/vault | Data & privacy audit | [path|scope] | PII handling, encryption, logging leaks, retention, GDPR gaps. |
Offensive & compliance
| Command | Title | Arguments | What it does |
|---|---|---|---|
/recon | Reconnaissance | [path|target] | Maps the attack surface: routes, services, subdomains, tech fingerprint. |
/threatmap | Threat model | [path|scope] | Attack surface, trust boundaries and STRIDE threats with mitigations. |
/dast | Dynamic testing | [target|scope] | Runs the app and probes live: injection, auth, IDOR, SSRF, XSS/CSRF. |
/redteam | Attack simulation | [path|scope] | Chains weaknesses into realistic exploit paths with safe proof-of-concept. |
/pentest | Full autonomous pentest | [target|scope] | End-to-end cycle: recon, threat model, scan, exploit with PoC, validated report. |
/comply | Compliance mapping | [framework|scope] | Gap analysis vs OWASP ASVS, CIS, SOC 2, ISO 27001, PCI-DSS. |
/report | Pentest report | [format|scope] | Compiles validated findings into a shareable, compliance-ready report. |
Studios
| Command | Title | Arguments | What it does |
|---|---|---|---|
/studio | Document studio | [format + brief] | Creates polished PowerPoint, Word, PDF, or Excel documents. See navin_contenant. |
/campaign | Marketing studio | [brief] | Generates campaigns end to end: ad copy, social posts, product images, video scripts. See navin_marketing. |
/seo | SEO studio | [url|topic] | Full SEO: technical audits, keyword research, optimized content, competitor analysis. See navin_seo. |
/leads | Leads & sales studio | [icp|company|brief] | Desk hunt (open data first), BANT-F, Start loop, heartbeat watch. See navin_leads, loop, desktop. |
| All workflow commands accept an optional focus/target. Without one, the agent infers the most useful scope itself. |
Utilities (side channel)
| Command | Arguments | What it does |
|---|---|---|
/checkpoint | [save [note]|list|restore <name> [all|chat|code]|delete <name>] | Checkpoints are auto-saved before each prompt. Rewind the conversation, the code, or both. They complement git, not replace it. |
/pilot | [task] | Manual model-per-task switch. Tasks: search, plan, review, security, dev, fast, deep, docs. Assign each role a named preset in Settings → Models → Task routing (modelRoutes). Workflows (/forge, /blueprint, audits, studios…) apply those routes automatically; /pilot review also switches the session preset. Without argument, shows the current mapping. |
/pack | [list|install <git-url|path>|enable <name>|disable <name>|remove <name>] | Manages plugin packs (skills + MCP servers). See Plugins. |
/model | [preset] | Shows or switches the active model preset. |
/status | - | Runtime, provider, and channel status (including web-search quota when available). |
/skill | - | Lists all enabled skills available to the agent. |
/help | - | Lists available slash commands. |
/history | [n] | Prints the last N persisted conversation messages. |
/trigger | <name> | Creates a named CLI trigger bound to this chat session. |
/pairing | [list|approve <code>|deny <code>|revoke <user_id>] | Manages DM pairing requests per channel. |
/restart | - | Restarts the Navin process. |
Memory (Dream)
| Command | Arguments | What it does |
|---|---|---|
/dream | - | Manually triggers the two-phase memory consolidation. |
/dream-log | - | Shows what the last Dream consolidation changed. |
/dream-restore | - | Reverts memory to a previous Dream snapshot. |
/dream-prompt | [init] | Customizes how Dream organizes this workspace's memory. |
/evaluator-prompt | [init] | Customizes the heartbeat notification gate prompt. |
Turn control
| Command | What it does |
|---|---|
/new | Resets this chat and starts a fresh conversation (finalizes the active turn first). |
/stop | Cancels the active agent turn for this chat. |
Tips
- Composer modes: pick Plan / Agent / Review / Security / Debug in the Mode menu - free text is prefixed automatically. See Modes.
- Combine with the file target: after opening a file in the editor, the Actions menu can scope any command to that file (e.g.
/inspect src/App.tsx). - Chain workflows:
/blueprintfirst, review the plan, then/forge. Use/cruisefor autopilot and/missionfor long durable work (see Plan Mode). - After Review / Security / Debug: open the HTML report in File Preview, reply
Start with #1(or another number), then Agent //forge. - Pentest cycle:
/recon→/threatmap→/probeor/dast→/redteam→/comply→/report(or/pentestto run the whole cycle). - Task routing is automatic for workflows (
/forge→dev,/blueprint→plan,/inspect→review,/fortify→security,/debug→deep, …). Use/pilot <task>only for free-form chat when you want to stay on a role. /checkpoint save before-refactorbefore risky operations,/checkpoint restore before-refactor codeto roll back files only.- Full menu reference: Actions.