planner-task-packet

Drafts CCB workflow task packets, readiness recommendations, and clarification questions without mutating state.

3.5k|342|Updated Oct 25, 2025
One-click install
npx skills add https://github.com/bfly123/claude_code_bridge --skill planner-task-packet
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: planner-task-packet
Source: https://github.com/bfly123/claude_code_bridge/tree/main/docs/plantree/plans/agentic-loop-workflow/drafts/agentroles.ccb_planner/skills/planner-task-packet
Command: npx skills add https://github.com/bfly123/claude_code_bridge --skill planner-task-packet

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Converting a macro user request or frontdesk intake into a structured, reviewable plan artifact is error-prone: planners may overstep authority, mutate task state, or emit inconsistent output formats. This Skill constrains the planning step to produce exact, controller-validated artifacts without executing anything.

Core Features & Use Cases

  • Structured Task Packets: Generates task-packet.md with mandatory sections (Goal, Acceptance Criteria, Interface Contracts, Constraints, Decomposition Inputs) plus readiness.json with strict readiness and route values.
  • Task-Set Decomposition: Emits task-set.json only for genuinely independent deliverables or distinct routes, keeping cohesive deliverables in one envelope for the orchestrator to slice.
  • Bounded Authority Handling: Validates a controller-issued terminal_status_constraint (schema version, digests, revisions) and fails closed with a blocker recommendation when fields are missing or stale.
  • Use Case: A frontdesk intake arrives requesting a new feature. The planner produces a fenced task packet with route direct_execution, non-empty allowed_paths, executable verification commands, and a matching readiness JSON for the controller to validate and settle.

Quick Start

Ask the planner to convert this macro request into a task packet with readiness recommendation and candidate clarification questions, without running any commands.

Frequently Asked Questions about planner-task-packet

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert a user request into a structured task packet for an agentic workflow?

Provide the macro request, scope, non-goals, and relevant references as intake, then emit a fenced task-packet.md with Goal, Acceptance Criteria, Interface Contracts, Constraints, and Decomposition Inputs sections, plus a matching readiness.json. Do not run shell commands or file reads before replying.

When should I use task-set.json instead of a single task packet?

Use task-set.json only for explicit independent deliverables, distinct routes, or route-mix intake. Keep one cohesive product deliverable in a single task envelope even when it spans multiple files, since the orchestrator owns execution slicing and parallelism.

What readiness values are allowed in the planner readiness.json output?

Readiness must be exactly one of ready, needs_clarification, blocked, or not_ready. The route field uses direct_execution, needs_detail, macro_adjustment_request, blocked, or partial_completion, and each route has specific rules for allowed_paths and blockers.

Can the planner authorize execution after producing detail artifacts?

Only when a valid controller-issued terminal_status_constraint is present with matching schema version, digests, revisions, and reason. Otherwise the planner recommends readiness and route, but the controller alone owns task-status authority and settles the recommendation.

Why must verification commands avoid git diff in some planning replies?

When the intake declares git_repository=false or not_guaranteed, Git-only scope checks are unreliable, so verification must use repo-independent file existence, content, test, or manifest checks. This is a per-activation capability constraint, not a global Git ban.