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.
examples/V6_PLAN_EXAMPLE.md
View on GitHub# A v6 Plan — Minimal Worked Example
A v6 plan is **two authored artifacts + one guarded sequence**. Allocate the
folder with `shared/plan_paths.py` before authoring; this example uses the
first ID in a new repository. This
example shows the smallest honest shape; the normative rules live in
[`../spec/V6_CONTRACT.md`](../spec/V6_CONTRACT.md) (contract + journal)
and [`../spec/V6_LIFECYCLE.md`](../spec/V6_LIFECYCLE.md) (detection,
materialization, activation, amendments, coexistence).
```
.dwp/plans/PLAN_001_ship_feature_x/
├── README.md # the human plan (what the approval digests)
├── 1.task_implement.md
├── 2.task_final_review.md
├── manifest.json # WRITTEN BY materialize — contract pointer
├── contract.json # WRITTEN BY materialize — stamped, content-addressed
├── journal.ndjson # first event: the approval
└── ... # state.json / views/ / gates/ are generated later
```
## 1. The markdown (authored)
`README.md` carries the goal, the task list and the execution rules — the
recorded create flow's shape, unchanged. Task files declare their
objective, `Touched Surface`, `Read Before Starting`, `Acceptance Criteria`
with `AC-*` ids, and `Validation` gates. The last task is the Final Review.
## 2. The contract draft (authored, outside the plan folder)
```json
{
"schema": "https://deepworkplan.com/schema/plan-contract/v6.json",
"spec_version": "6.0.0",
"plan": "PLAN_001_ship_feature_x",
"revision": 1,
"created_at": "2026-09-26T12:00:00Z",
"title": "Ship feature X under the v6 contract",
"outcome": {
"statement": "Feature X ships with verified behavior.",
"success_definition": "Both acceptance criteria hold on observed evidence.",
"out_of_scope": ["publicity", "release tagging"]
},
"acceptance": {
"criteria": [
{ "id": "AC-tests-pass", "statement": "The suite passes.",
"observable_check": "the repo's test command exits 0",
"accepted_evidence": ["observed"] },
{ "id": "AC-human-review", "statement": "A human reviewed the diff.",
"observable_check": "review sign-off recorded",
"accepted_evidence": ["asserted"] }
]
},
"tasks": [
{ "id": "T-implement", "title": "Implement feature X",
"prerequisites": [], "touched_surface": ["src/feature_x.py"],
"gate_intent": [
{ "criterion": "AC-tests-pass", "check": "pytest -q" }] },
{ "id": "T-final-review", "title": "Final Review",
"prerequisites": ["T-implement"], "touched_surface": ["src/", "tests/"],
"gate_intent": [
{ "criterion": "AC-human-review",
"check": "a human reviews the diff (recorded by ledger.py signoff, not executed)" }] }
],
"invariants": [
{ "id": "INV-no-publication",
"statement": "No pushes, tags or publication during this plan." }],
"scope": {
"allowed_paths": ["src/", "tests/"],
"allowed_command_classes": ["python3", "pytest"],
"forbidden_operations": ["publish", "force-push"] },
"resource_envelope": {
"limits": [
{ "id": "spend_usd", "unit": "USD", "limit": 40,
"enforcement": "enforced",
"metering_source": "host adapter: usage.total_cost_usd" },
{ "id": "wall_clock_h", "unit": "hours", "limit": 6,
"enforcement": "advisory" } ] },
"scheduling": {
"max_retries_per_gate": 1, "max_adaptations_per_task": 2,
"starvation_threshold_events": 24,
"handoff": { "fresh_context": "context pressure at 60%",
"cross_host_resume": "custodian seal broken" } },
"permissions": {
"granted": ["gate_command_exec", "fs_write_repo_scope", "git_operations"],
"not_granted": ["network_access", "agent_delegation"] },
"authorization": {
"authority": "the developer who asked for the plan",
"mechanism": "plan_authorship",
"boundaries": "This contract governs feature X only.",
"consent_checkpoints": ["GO before the release commit"],
"timestamp": "2026-09-26T12:00:00Z" },
"dependencies": [
{ "kind": "pinned_input", "name": "python3", "detail": ">= 3.9" } ]
}
```
Validate it before proposing:
`python3 ../shared/contract_v6.py validate-contract draft.json`.
## 3. Materialization (guarded — the helper writes, never your hands)
```bash
python3 ../shared/ledger.py --plan .dwp/plans/PLAN_001_ship_feature_x \
materialize --contract draft.json --authority sergio \
--mechanism plan_authorship
# OK: materialized PLAN_001_ship_feature_x contract <cid> (manifest, contract,
# approval seq 1)
```
`AC-human-review` accepts only `asserted`, so no gate closes it: once
`T-final-review` has started and the human has reviewed the diff and
written their note, record their sign-off (the intent's `check` describes
the human check; it is never executed):
```bash
python3 ../shared/ledger.py --plan .dwp/plans/PLAN_001_ship_feature_x \
signoff --criterion AC-human-review \
--evidence-path analysis_results/REVIEW_NOTE.md --authority sergio
# OK: signoff AC-human-review by sergio at seq <n> (asserted: a human claim,
# never executed)
```
From here the v6 execute loop runs the plan:
`ready` selects, `start` opens the attempt and records the fingerprint,
`gate` mints observed evidence, `complete` is refused until every
criterion holds, `project` + `views.py render` regenerate the
projections, `receipt` recomputes the outcome from the records.
**What never happens:** the contract edited in place after approval (a
change is a revision under `contracts/` with fresh authority — refine),
a manifest pointer rewritten, a v5 plan migrated silently, or a task
closed on narrative.
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 |