j-mauro-r/SAT-Dengue-V2 / .agents/skills/deepworkplan
deepworkplan
VerifiedDeepWorkPlan — turn any repo AI-first and run Deep Work Plans. Routes to create, execute, refine, resume, status, verify, upgrade, and repo-onboarding sub-skills based on intent. Use when the developer wants to plan, execute, manage, or verify structured multi-task work, or make a repository AI-agent-ready.
spec/LITE_PLANS.md
View on GitHub# Lite and Full Plan Lifecycle
> **Version scope:** This is a retained v5.0.0 base document. The current
> v6 standard also requires the applicable `V6_*.md` extensions indexed in
> [README.md](README.md). Existing v5 plans keep this document’s recorded rules.
| Field | Value |
| --- | --- |
| Version | 5.0.0 |
| Status | Stable |
## Representation and lifecycle
Each plan has one folder and immutable creation manifest. `lite` stores compact,
executable task records in the README; `full` stores detailed task files. Lite is
not a partial Full plan: it has stable task IDs, anchors, touched surface,
acceptance criteria, gates, completion evidence (skills disposition and
documentation decision included — `DWP_SPECIFICATION.md` §6.2, §6.6) and a
Final Review (security pass, final-state validation, skills and documentation
reconciliation — §6.1).
Format, materialization, approval and execution are independent. Format is
`lite` or `full`; materialization is `materializing`, `ready` or `promoting`;
approval is `pending`, `approved` or `pre_approved`; execution is `pending`,
`in_progress`, `blocked` or `completed`. Guided create writes a reviewable
pending proposal. Trust materializes a ready pre-approved format and returns
control. Create and promotion never execute product work. An explicit execute or
resume request approves the ready plan's current scope; record that approval
before starting work. Without that request, a pending proposal is not executable.
Unresolved promotions always require recovery before product work.
## Creation and selection
`/dwp-create` serves planning intent at every scale. Small bounded work uses
Lite; a direct edit, explanation, status, resume or explicit no-plan request
keeps its intended route. `lite`/`full` are format preferences; `trust`/`auto`
are interaction options. Boundary options can appear at either end in any order:
```text
/dwp-create trust fix the label
/dwp-create lite trust fix the label
/dwp-create fix the label trust lite
/dwp-create fix the migration full trust
```
No preference requests an explainable recommendation. An explicit Full wins.
An explicit Lite is honored when compact task records carry all requirements and
gates; otherwise DWP records why Full is required. Identical options repeat
idempotently; conflicting formats are an error. `--` ends option parsing.
Selection records observed scope, dependencies, required instruction detail,
unknowns and rationale; it is an auditable judgment, not a cross-model guarantee.
Lite carries the parallelization decision the same way Full does
(`DWP_SPECIFICATION.md` §9; `guide/team-agents.md` §14.2): the
`Execution: sequential — {rationale}` line, or the Team Agents Configuration
section — with per-task Team Agents Metadata attached to the anchored task
records instead of task files. The decision is never silent in Lite either.
## Promotion and compatibility
Promotion is representation-only: it writes intent, destination tasks, validates
coverage, switches authoritative README/state, then clears the marker. Execute
and resume stop while a marker remains. Unchanged task IDs and evidence persist;
new scope follows refine and invalidates affected evidence. Full-to-Lite is not
automatic. v1 Full and legacy plans retain their recorded shape. v1 schemas stay
closed; Lite uses v2 schema URLs with typed inline/file locators. Creation format
lives in the immutable v2 manifest. Mutable format, materialization, approval and
promotion live in `state.json`, which is the machine-readable authority; the
README carries the same facts in human-readable form and wins on task-completion
desync. `approval` is optional in the v2 state schema so that a plan written
before it was recorded still validates: when it is absent, treat the README's
`Approval` row as the value, and `pending` when neither is present. The refined
draft, the `.dwp/drafts/` directory and the `refined-draft` /
`from-refined-draft` / `from` parameters were removed in 2.4.0; `create` writes
the plan folder and nothing else.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |