Commands Reference

SNIPER provides slash commands that drive the project lifecycle. Each command is invoked in Claude Code and orchestrates agent teams, manages state, and produces artifacts.

Lifecycle Commands

These commands run the core project phases in order.

Command	Phase	Description
/sniper-init	Setup	Initialize SNIPER in a new or existing project
/sniper-discover	1 -- Discover	Run discovery and analysis with a 3-agent parallel team
/sniper-plan	2 -- Plan	Run planning and architecture with a 4-agent team
/sniper-solve	3 -- Solve	Shard epics into self-contained stories (single agent)
/sniper-sprint	4 -- Sprint	Run an implementation sprint with a dynamic team
/sniper-review	Gate	Evaluate phase artifacts against quality checklists

Utility Commands

These commands support the lifecycle without advancing phases.

Command	Description
/sniper-compose	Compose a spawn prompt from persona layers
/sniper-status	Show lifecycle status, artifact state, and team health
/sniper-doc	Generate or update project documentation with a 3-agent team
/sniper-memory	View and manage conventions, anti-patterns, and decisions

Extended Commands

These commands handle specialized workflows beyond the standard lifecycle.

Command	Description
/sniper-feature	Run a scoped feature lifecycle (5 phases) for adding a feature to an existing project
/sniper-ingest	Reverse-engineer SNIPER artifacts from an existing codebase using a 3-agent team
/sniper-debug	Investigate and fix bugs with a 3-phase structured approach
/sniper-audit	Run targeted audits (refactor, review, tests, security, performance)

Workspace Commands

These commands orchestrate SNIPER across multiple repositories.

Command	Description
/sniper-workspace init	Initialize a workspace with dependency detection
/sniper-workspace feature	Plan and implement a feature spanning multiple repos
/sniper-workspace status	View workspace state, repo status, and contract versions
/sniper-workspace validate	Validate implementations against interface contracts

Command Patterns

All phase commands follow a consistent pattern:

1. State check -- verify SNIPER is initialized and determine the current phase
2. Team loading -- read the team YAML to determine agents, tasks, and dependencies
3. Spawn prompt composition -- compose prompts from persona layers for each agent
4. Delegate mode -- the team lead coordinates agents without producing artifacts directly
5. Artifact output -- agents produce documents and code in the docs/ directory
6. Review gate -- quality evaluation before the lifecycle advances

TIP

Run /sniper-status at any time to see where you are in the lifecycle and what command to run next.

Related

• Full Lifecycle -- how phases connect end-to-end
• Teams -- team compositions used by each command
• Review Gates -- quality gates between phases