Built-In Orchestration Tools

Built-in tool inventory for Locus, with categorized commands and practical workflows that extend raw provider CLIs.

Locus adds an operational toolchain on top of provider CLIs (Claude/Codex): planning, GitHub-native execution, review loops, status visibility, logging, sandbox lifecycle management, and package extensibility.

Why Built-In Tools (vs Raw Provider CLIs)

Raw provider CLIs can generate code, but they do not natively manage your GitHub workflow end to end.

Locus built-in tools add:

• GitHub task lifecycle control (issue, sprint, plan, run)

• Delivery automation (autoLabel, autoPR, run --resume)

• Structured quality loops (review, iterate)

• Operational visibility (status, logs, artifacts)

Tool Inventory

1) Setup and Configuration

Tool	Purpose	Example
locus init	Initialize .locus/, verify gh, and seed label schema	locus init
locus config	Read/write project settings (model, automation, execution behavior)	locus config set ai.model gpt-5.3-codex
locus upgrade	Check and install newer CLI versions	locus upgrade --check

2) Work Modeling in GitHub

Tool	Purpose	Example
locus issue	Create/manage GitHub issues as execution units	locus issue create "Add webhook signature validation" --sprint "Sprint 8"
locus sprint	Create/activate/manage sprint milestones	locus sprint active "Sprint 8"
locus plan	Generate plan files, review, and approve into GitHub issues	locus plan "Build billing retries" --sprint "Sprint 8" then locus plan approve <id> "Sprint 8"

3) Execution, Review, and Iteration

Tool	Purpose	Example
locus run	Execute active sprint or standalone issue set	locus run --resume
locus review	AI review for open agent:managed PRs	locus review
locus iterate	Apply PR feedback in loop	locus iterate --pr 42
locus status	Operational dashboard for sprint/issues/PRs	locus status
locus logs	Tail and inspect run logs for failures/debugging	locus logs --follow

4) Interactive and Analysis Tools

Tool	Purpose	Example
locus exec	Interactive or one-shot coding session	locus exec "Add retry guard to webhook handler"
locus discuss	Architecture discussion and planning prompts	locus discuss "Should we split webhook processing into a queue worker?"
locus memory	List, search, and manage structured project memory	locus memory list --category architecture
locus artifacts	Inspect and convert AI-generated artifacts in .locus/artifacts/	locus artifacts plan webhook-reliability-prd

5) Package Ecosystem Tools

Tool	Purpose	Example
locus install	Install community package from npm	locus install telegram
locus uninstall	Remove installed package	locus uninstall telegram
locus packages	List installed packages / check outdated	locus packages outdated
locus pkg	Run package-provided commands	locus pkg telegram start

6) Sandbox Management Tools

Tool	Purpose	Example
locus sandbox	Create/manage provider sandboxes, auth sessions, package installs, shell access, and logs	locus sandbox install bun --provider codex

Practical Workflows Where Built-In Tools Win

Workflow A: Issue-to-PR Delivery with Recovery

locus sprint active "Sprint 8"
locus run
locus review
locus iterate --sprint
locus run --resume

Why this beats raw provider CLI usage:

• Keeps task state in GitHub labels/issues/PRs

• Recovers from interruption without re-running completed tasks

• Preserves one team workflow across Claude/Codex model switching

Workflow B: Plan Approval Before Execution

locus plan "Add SSO login and role-based access"
locus plan list
locus plan show <id>
locus plan approve <id> "Sprint 9"
locus run

Why this beats raw provider CLI usage:

• Separates planning from execution with an approval gate

• Converts approved plan into ordered GitHub work automatically

• Keeps execution order explicit via order:N labels

Related Docs

• Auto-Approval Mode

• Execution Model (Technical)

• GitHub-Native Workflows

• Sandboxing Setup (Docker-First)

• Unified Interface Across AI Clients