Cursor Task Orchestration
Cursor-style research, implement, verify via delegate_task.
Skill metadata
| Source | Bundled (installed by default) |
| Path | skills/software-development/cursor-task-orchestration |
| Version | 1.0.0 |
| Author | VibeOS |
| License | MIT |
| Platforms | linux, macos, windows |
| Tags | cursor, delegation, subagent, review, orchestration |
| Related skills | plan, requesting-code-review, vibeos-local-verify-deploy, cursor-capabilities-bridge, project-kanban-worktrees |
Reference: full SKILL.md
信息
The following is the complete skill definition that VibeOS loads when this skill is triggered. This is what the agent sees as instructions when the skill is active.
Cursor Task Orchestration
Map Cursor’s Task/subagent pattern onto VibeOS delegate_task without new
core tools. Short parallel work stays in-process; work that must survive
restart goes to kanban (kanban-automations skill).
When to Use
- Multi-step coding: research → implement → verify
- User asks for “как Task в Cursor”, parallel workers, or fresh-context review
- After a plan from the
planskill is ready to execute
Skip when: single small edit; user forbids subagents; task must outlive this process (use kanban instead).
Prerequisites
- Delegation enabled (
delegation.orchestrator_enabled, default true) - Prefer leaf workers with narrow
toolsets(usuallyfile+terminal) - Optional deeper recipe: install
optional-skills/software-development/subagent-driven-developmentviavibeos skills install official/software-development/subagent-driven-development
How to Run
- Write or load a short plan (or use
planskill). - Dispatch workers with
delegate_task— one goal per leaf, full context inlined. - Run an independent verify pass (
requesting-code-reviewor a review leaf). - Parent integrates results; do not trust a worker to grade itself.
Example batch (conceptual):
delegate_task(tasks=[
{goal: "Research: locate X and summarize touch points",
toolsets: ["file", "terminal"], role: "leaf"},
{goal: "Implement: apply the agreed minimal patch for X",
toolsets: ["file", "terminal"], role: "leaf"},
])
# after implement returns:
delegate_task(goal="Verify: run targeted tests + security skim",
toolsets=["file", "terminal"], role="leaf")
Quick Reference
| Need | Use |
|---|---|
| Plan only | plan skill |
| Fresh implementer per task + 2-stage review | optional subagent-driven-development |
| Quick 2–3 parallel leaves | delegate_task batch (this skill) |
| Survive restart / multi-profile | kanban-automations |
| Pre-commit gate | requesting-code-review |
| Timeweb after code lands | vibeos-local-verify-deploy |
Config knobs (~/.vibeos/config.yaml or coding profile):
| Key | Typical |
|---|---|
delegation.max_concurrent_children | 3 |
delegation.child_timeout_seconds | 600 |
delegation.max_spawn_depth | 2 |
delegation.inherit_mcp_toolsets | true on coding profile |
Procedure
Research leaf
- Read-only bias:
search_files/read_file/ narrowterminalgreps - Return: files to touch, risks, open questions — no large unrelated edits
Implement leaf
- Minimal diff; follow TDD when tests exist (
test-driven-development) - No
git push/gh pr(blocked byvibeos-policyunless opted in)
Verify leaf
- Run
scripts/run_tests.sh <touched>(never bare pytest for suite parity) - Security skim via
requesting-code-reviewsteps orsecurity-guidancewarnings - Report pass/fail + residual risks
Pitfalls
- Background
delegate_taskis process-local — restart loses it → kanban/cron - Do not give leaf workers
delegate_taskunless orchestrator role is intended - Do not swap toolsets mid-parent-conversation just to “look like Cursor”
- Inlining the full task text beats “read the plan file yourself”
Verification
- Parent receives three distinct summaries (research / implement / verify)
- Targeted tests green via
scripts/run_tests.sh - No GitHub mutation attempted without explicit user approval