with one click
hive-project-setup
// Set up and configure Hive workspaces, projects, tasks, and evaluator policy. Use this skill when bootstrapping, creating projects, managing tasks, or configuring PROGRAM.md.
// Set up and configure Hive workspaces, projects, tasks, and evaluator policy. Use this skill when bootstrapping, creating projects, managing tasks, or configuring PROGRAM.md.
Discipline for developing Hive itself — PR sizing, review handling, merge discipline, release workflow, delegation, and cleanup hygiene. Use this skill when working on Hive repo changes that span multiple PRs or review cycles.
Coordinate work across multiple agents and projects in Hive. Covers task claims, blockers, handoffs, campaigns, portfolio management, briefs, and shared memory.
Hive mental model and orientation. Read this first before using any other Hive skill. Covers the entity hierarchy, observe-and-steer pattern, drivers, sandboxes, console vs CLI, and workspace conventions.
Use the Hive MCP server for the thin v2 search and execute tool surface. Use this skill when configuring MCP access to Hive or when an agent needs workspace search and bounded local execution.
The core agent work cycle in Hive — from finding a task through claiming, launching a run, handling approvals, finishing, and promoting. Use this skill for task-first project work, governed runs, and clean handoff.
| name | hive-project-setup |
| description | Set up and configure Hive workspaces, projects, tasks, and evaluator policy. Use this skill when bootstrapping, creating projects, managing tasks, or configuring PROGRAM.md. |
This skill covers workspace bootstrap, project creation, task management, and evaluator configuration.
hive onboard <project-slug> --json
This creates:
.hive/ substrate directory with cache and task storageprojects/<slug>/AGENCY.md — narrative project documentprojects/<slug>/PROGRAM.md — execution policy with forgiving defaultslocal-smoke evaluator templateDefault PROGRAM.md uses forgiving settings so the first hive finish succeeds cleanly:
auto_close_task: trueallow_accept_without_changes: trueTighten these for production use.
hive init --json
Creates the .hive/ layout only. No starter project or tasks.
hive adopt <project-slug> --json
Wraps existing v1 state and enables v2 features incrementally.
hive migrate v1-to-v2 --dry-run --json # preview
hive migrate v1-to-v2 --project <id> --json # execute
hive project create <slug> --title "Project title" --json
hive project list --json
hive project show <project-id> --json
hive project sync --json
Each project gets:
projects/<slug>/AGENCY.md — write narrative context here (objective, notes, recommended flow)projects/<slug>/PROGRAM.md — define execution policy herehive task create --project-id <id> --title "Task title" --json
hive task create --project-id <id> --title "..." --status ready --json
hive task list --project-id <id> --json
hive task list --status ready --status review --json
hive task show <task-id> --json
hive task update <task-id> --status in_progress --priority 1 --json
hive task update <task-id> --title "New title" --label "bug" --json
proposed → ready → claimed → in_progress → review → done
Side states: blocked (manual), archived (deprecated).
0 = critical, 1 = high, 2 = medium, 3 = low. hive next returns highest priority first.
hive task link <blocked-id> blocked_by <blocker-id> --json
hive task link <parent-id> parent_of <child-id> --json
hive task link <id-a> relates_to <id-b> --json
hive deps --json # project-level blocker summary
Edge types: blocks, parent_of, relates_to, duplicates, supersedes.
hive task claim <task-id> --owner <name> --ttl-minutes 30 --json
hive task release <task-id> --json
Claims are time-limited work leases. Use them, not hand-edited owner fields in AGENCY.md.
hive program doctor <project-id> --json
This shows the current PROGRAM.md state, available evaluator templates, and any policy issues.
hive program add-evaluator <project-id> local-smoke --json
Available templates:
local-smoke — minimal validation stub, proves the evaluator loop worksCustom evaluators can be defined directly in PROGRAM.md.
program_version: 1
mode: workflow
default_executor: local
budgets:
max_wall_clock_minutes: 30
max_steps: 25
max_tokens: 20000
max_cost_usd: 2.0
paths:
allow:
- README.md
- docs/**
- projects/<project>/**
deny:
- secrets/**
- infra/prod/**
commands:
allow: []
deny:
- rm -rf /
evaluators: []
promotion:
allow_unsafe_without_evaluators: false
auto_close_task: true
requires_all: []
escalation:
when_paths_match: []
when_commands_match: []
Read PROGRAM.md before:
If work needs new guardrails, update PROGRAM.md rather than inventing ad hoc rules.
hive workspace checkpoint --message "before refactor" --json
Creates a git commit to freeze workspace state. hive work does this automatically before launching runs.
hive cache rebuild --json
hive sync projections --json
Run this after task, run, or memory changes so GLOBAL.md, AGENCY.md task rollups, and AGENTS.md stay current.
hive doctor --json # full workspace health
hive doctor workspace --json # workspace structure only
hive doctor program <project-id> --json # PROGRAM.md policy check
.hive/tasks/*.md. AGENCY.md is narrative, not machine state..hive/.auto_close_task, allow_accept_without_changes) and tighten for production.owner field as the primary locking mechanism — use task claims.