[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7F5X0FJRH7KFVX79DCEJ98R":3,"skill-same:01M7F5X0FJRH7KFVX79DCEJ98R":66},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":34,"commitSha":35,"skillDirectory":36,"forksCount":37,"repositoryHtmlUrl":38,"repositoryHomepageUrl":6,"ownerHtmlUrl":39,"ownerBlogUrl":6,"ownerDescription":40,"updatedAt":41,"install":42,"resources":55},"01M7F5X0FJRH7KFVX79DCEJ98R","tasks",null,"Use when an approved plan needs slicing into an ordered task list before any code — the SDD `tasks` phase. Each row carries a literal done-check, dependencies, a parallel-safe marker and a spec trace, appended to the plan artifact. NOT writing the plan (that is `plan`), NOT the consistency gate (that is `analyze`), NOT executing the tasks (that is `implement`).","lza6\u002FArmor-piercing-system-prompts","lza6","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F128880206?v=4","01M7EB3K7WPX5T28EJ2NVHE6EG","Chinese-tingfeng","01M7EB3K97E1KQ16QWKJMFNGX8",11,"MIT","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F05-技能与代理开发\u002Ftasks\u002FSKILL.md",[],[],"2026-11-16 10:05:34",false,"---\nname: tasks\ndescription: \"Use when an approved plan needs slicing into an ordered task list before any code — the SDD `tasks` phase. Each row carries a literal done-check, dependencies, a parallel-safe marker and a spec trace, appended to the plan artifact. NOT writing the plan (that is `plan`), NOT the consistency gate (that is `analyze`), NOT executing the tasks (that is `implement`).\"\ntags: [sdd, tasks, breakdown]\nrecommends: [analyze, implement]\nprofiles: [core, full]\norigin: risco\n---\n\n# Tasks — turn an approved plan into a verifiable work list\n\nThe `tasks` phase is the hinge between *thinking* and *doing*. The plan already\ndecided the architecture, the interfaces, and the testing strategy. This phase\nslices that plan into the **smallest units a coding agent can finish, prove, and\nhand off** — each one ordered, each one carrying a *done-check* that a machine or\na reviewer can run without trusting anyone's word.\n\nA task list is not a to-do list. A to-do list says \"build the auth endpoint\". A\ntask list says \"T004: implement `POST \u002Flogin`; done when `pytest\ntests\u002Fauth\u002Ftest_login.py` is green AND a 401 is returned for a bad password —\ndepends on T002, T003; parallel-safe with T005.\" The difference is that the\nsecond one can be **handed to a subagent, executed, and verified** with no\nfurther questions. This phase writes zero runtime code; it produces one artifact,\nappended to the plan it was built from.\n\n## Model tier — `balanced` (opt-in routing)\n\nThis phase's default tier is **`balanced`**: decomposing an approved plan is\nstructured work, not architecture. Routing is off unless `models.enabled: true`\nin `02-DOCS\u002Fwiki\u002Fsdd\u002Fconfig.yaml`; when on, resolve the tier and apply it per\n`..\u002Fsdd\u002Freferences\u002Fmodel-routing.md`, which owns the resolution order and the\nannouncement rules. Routing off or no profile → honor the session model silently.\n\n## Read the harness profile first\n\nRead `02-DOCS\u002Fwiki\u002Fharness\u002Fuser-profile.md` before producing anything and match\nthe accompaniment dial: **L0** emit the task table and nothing else; **L1** add\none line of *why this slicing* above it; **L2** justify the ordering and the\nparallel markers as you go, flagging the risky tasks; **L3** walk each dependency\nand confirm the done-checks make sense to the user before writing. No profile →\nassume non-technical and let `init` gauge the level; never invent one.\n\n## Inputs (refuse to start without them)\n\nYou break down a **plan**, not an intent. Before slicing, confirm all three exist:\n\n1. **The plan** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md`, written by the `plan`\n   phase. It must contain the architecture, interfaces, and testing strategy.\n2. **The spec** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fspecs\u002F\u003Cslug>.md`, so every task traces to a\n   requirement. A task with no spec line behind it is scope you are inventing.\n3. **The constitution** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fconstitution.md`, the non-negotiable\n   bars (stack canon, quality gates) every task must respect.\n4. **The SDD config** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fconfig.yaml`, if present. Use its\n   `review_budget`, `delivery_strategy` and `testing.commands` when writing\n   done-checks and review forecasts. If it is missing on non-trivial work,\n   recommend `sdd-init`.\n\nIf the plan is missing, **stop and route to `plan`**. If the plan exists but is\nvague on interfaces or testing, route back to `plan` to tighten it — do not\npaper over a thin plan with guesswork in the task list. That guesswork is the\nexact thing `analyze` will catch next, and you will have wasted the round trip.\n\n## What makes a task well-formed\n\nEvery task is one row. A row is well-formed only when all six fields hold:\n\n| Field | Rule |\n| --- | --- |\n| **ID** | `T001`, `T002`, … — sequential, zero-padded, no hyphen (GitHub Spec Kit canon). Stable, never renumbered once written (downstream phases cite it). |\n| **[P]** | A **separate field after the ID**: `[P]` if the task is parallel-safe, blank otherwise. Never fold the marker into the ID. |\n| **Title** | One imperative verb + object. \"Implement\", \"Add\", \"Wire\", \"Migrate\" — not \"Handle\" or \"Support\". |\n| **Done-check** | The **literal** command or observation that proves done. Runnable, not a feeling. See below. |\n| **Depends-on** | The task IDs that must finish first, or `—` for none. Drives the ordering and the parallel markers. |\n| **Trace** | The spec section \u002F acceptance criterion this task satisfies. Every task traces to one; if it can't, it's out of scope. |\n\nPut `[P]` (parallel-safe) in its own field only when the task shares **no files\nand no state** with another unblocked task. When in doubt, leave it blank — a\nfalse `[P]` causes the merge collisions `parallel` exists to prevent.\n\n## The done-check is the whole point\n\nA done-check has to be **executable evidence**, owned by the stack, not a\nself-graded claim. Delegate the *form* of the check to the relevant stack skill\nand quote the actual command:\n\n- Backend (Python\u002FFastAPI) → a `pytest` invocation; route to `..\u002Ffastapi\u002FSKILL.md`.\n- Go services → a `go test .\u002F...` target; route to `..\u002Fgo\u002FSKILL.md`.\n- Frontend (Next.js) → a component\u002Fe2e test or a typed build; route to `..\u002Fnextjs\u002FSKILL.md`.\n- Flutter → a `flutter test` target; route to `..\u002Fflutter\u002FSKILL.md`.\n- Schema\u002Fmigrations → a migration applies + a query returns expected rows; route to `..\u002Fpostgresdb\u002FSKILL.md`.\n\n```text\nWEAK   done-check: \"login works\"\nSTRONG done-check: pytest tests\u002Fauth\u002Ftest_login.py::test_bad_password_returns_401  → green\n                   AND manual: POST \u002Flogin {wrong pw} returns 401, no token in body\n```\n\nA done-check you cannot run is not a done-check. If the plan's testing strategy\ndoesn't support a runnable check for a slice, that is a **plan gap** — name it\nand send it back, don't write a vague check to keep moving.\n\n## TDD-shaped slicing (default)\n\nThe `implement` phase runs red → green → refactor. Slice so it can. For each\nbehavioral unit, the natural shape is a test-first pair the implementer executes\nin order:\n\n1. **Txxx (test):** write the failing test for the behavior. *Done when the test\n   exists and fails for the right reason.*\n2. **Txxx+1 (impl):** implement until that test is green. *Done when the test\n   passes and nothing else regresses.*\n\nYou don't have to split every task in two, but the done-check of an\nimplementation task should always be *a test that was red and is now green*. This\nis where `tasks` and `implement` shake hands: the list you write is the list TDD\nwill execute.\n\n## The slicing procedure\n\n```text\n1. WALK the plan top to bottom. List every concrete deliverable\n   (endpoint, model, migration, component, job, config).\n2. For each deliverable, ask: smallest slice that produces a runnable\n   done-check? Split until each task is verifiable on its own.\n3. ORDER by dependency. Foundations first: schema → data layer →\n   service → API → UI. A task never precedes what it depends on.\n4. MARK [P] only on tasks that share no files\u002Fstate with another\n   unblocked task. Default to sequential.\n5. TRACE each task back to a spec line. No trace → cut it (scope creep)\n   or send the gap to `clarify`\u002F`specify`.\n6. ADD the cross-cutting closers: a final \"all done-checks pass\" task\n   and a \"verify.sh green\" task that hands off to `verify`.\n7. ADD a review workload + delivery forecast before implementation.\n8. APPEND the table to the plan artifact (see below). Do not start a\n   new file; the list lives with the plan it came from.\n```\n\n## Where the artifact lives\n\nThe task list is **not** a new document. Append it to the existing plan under a\n`## Tasks` heading:\n\n```text\n02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md\n  ├─ ## Architecture        (from plan)\n  ├─ ## Interfaces          (from plan)\n  ├─ ## Testing strategy    (from plan)\n  └─ ## Tasks               ← you append this\n       T001 … T0NN as the table below, + a one-line \"generated by tasks on \u003Cdate>\"\n```\n\nThen ensure the plan is indexed in `02-DOCS\u002Fwiki\u002Findex.md` (the Knowledge map;\nroot `CLAUDE.md` keeps only a short pointer) under the `sdd\u002F` topic — the `plan`\nphase usually added the row, so confirm it points at\n`02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md` and add it if missing. Additive only: never\ndelete a user's map entry.\n\n### The table format\n\n```markdown\n## Tasks\n\u003C!-- generated by tasks on 2026-06-01; IDs are stable, do not renumber -->\n\n| ID | [P] | Task | Done-check | Depends-on | Trace |\n| --- | --- | --- | --- | --- | --- |\n| T001 |  | Add `users` table migration | `alembic upgrade head` applies clean; `\\d users` shows email UNIQUE | — | spec §3 Data |\n| T002 | [P] | Add `sessions` table migration | migration applies; FK to users present | T001 | spec §3 Data |\n| T003 |  | Write failing test for `POST \u002Flogin` | `pytest tests\u002Fauth\u002Ftest_login.py` fails: no route | T001 | spec §4 Auth |\n| T004 |  | Implement `POST \u002Flogin` | T003 test green; bad pw → 401 | T003 | spec §4 Auth |\n| ... | ... | ... | ... | ... | ... |\n| T0NN |  | All done-checks pass + `verify.sh` green | every row above checked; `scripts\u002Fverify.sh` exits 0 | all | spec §Acceptance |\n```\n\n### Per-task Interfaces (for context-isolated implementers)\n\n`implement` and `parallel` dispatch tasks to **context-isolated** workers (the `developer`\nsubagent) that see *only their own task* — not the whole plan. Such a worker can't infer a\nneighbor's function signature, payload shape, or column name from a one-line row. For any task\nwhose correctness depends on a contract it doesn't own, attach an **Interfaces block** right under\nits row. Trivial, self-contained tasks don't need one — don't add ceremony where there's no\ncross-task contract.\n\n```markdown\n**T004 — Interfaces**\n- Consumes: `auth.verifyPassword(plain: str, hash: str) -> bool` (from T003); `users.email UNIQUE`\n- Produces: `POST \u002Flogin` → `200 {token}` | `401 {error}`; sets `Set-Cookie: sid=…; HttpOnly`\n```\n\nQuote **exact** signatures and shapes, not descriptions — the isolated worker copies them, it can't\ngo look them up. \"Returns the user\" is invisible; `-> {id, email}` is usable. `Consumes` names what\nthe task reads from a neighbor or the environment; `Produces` names the contract later tasks (and\nthe per-task reviewer) will hold it to. The plan's **§0 Global Constraints** are inherited by every\ntask implicitly — do **not** repeat them per task. Interfaces carry the *task-local* contract;\nGlobal Constraints carry the *project-wide* one. Together they are everything a blind implementer\nneeds.\n\n## Review workload + delivery strategy forecast\n\nAfter the task table, append a short forecast. This protects the human reviewer\nbefore a giant diff exists.\n\n```markdown\n## Review Workload Forecast\n\n| Dimension | Forecast | Why |\n| --- | --- | --- |\n| Estimated changed lines | \u003Cnumber or range> | based on task count + touched areas |\n| Files \u002F areas | \u003Ccount and names> | modules, migrations, UI, tests, docs |\n| Review risk | low \u002F medium \u002F high | complexity, cross-stack scope, security\u002Fdata changes |\n| Suggested delivery | single-pr \u002F ask-on-risk \u002F autochain \u002F exception | matched to config.review_budget |\n```\n\nPick the delivery strategy from the forecast: tiny change → `single-pr`; estimated\nlines over `config.sdd.review_budget.line_budget` (default 400) → recommend\nsplitting or `ask-on-risk`; many areas or cross-stack contracts touched →\n`autochain` or a feature-track branch. A deadline or emergency that justifies a\nlarge review is marked `exception` and requires explicit user approval later.\n\n## Anti-patterns → STOP\n\n| Tempting move | Why it's wrong \u002F Fix |\n| --- | --- |\n| \"I'll write tasks straight from the user's intent\" | You're skipping `plan`. Tasks slice a plan, not a wish. Route to `plan`. |\n| \"Done-check: 'feature works'\" | Not runnable, not evidence. Quote the literal test\u002Fcommand, or it's not a done-check. |\n| \"One giant task: 'build the backend'\" | Unverifiable, un-handoffable. Split until each slice has its own runnable check. |\n| \"Mark everything `[P]` so it goes faster\" | Shared files = merge collisions. `[P]` only when scope is truly disjoint. |\n| \"This task has no spec line but we obviously need it\" | That's scope creep wearing a hat. Trace it or cut it; if it's real, send it to `clarify`. |\n| \"The plan is thin here, I'll guess the slice\" | `analyze` will reject the guess next phase. Tighten the plan first. |\n| \"I'll start coding the easy task while I list the rest\" | This phase writes zero runtime code. Implementation is `implement`. |\n| \"Renumber the IDs so they're contiguous\" | Downstream phases cite IDs. They're stable once written. Append, don't renumber. |\n\n## Result envelope\n\nEnd with:\n\n```json result-envelope\n{\n  \"status\": \"complete\",\n  \"executive_summary\": \"Task list and review workload forecast appended to the plan.\",\n  \"artifact\": \"02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md\",\n  \"next_recommended\": \"analyze\",\n  \"risk\": \"low|medium|high\",\n  \"skill_resolution\": {\n    \"used\": [\"tasks\"],\n    \"missing\": [],\n    \"fallback\": [],\n    \"compact_rules\": [\"Every task needs a runnable done-check.\", \"Forecast review load before implementation.\"]\n  },\n  \"evidence\": [\"task table appended\", \"review workload forecast appended\"]\n}\n```\n\n## Next in the SDD chain\n\nThe task list is now the contract for the rest of the build. Hand off to\n**`analyze`** — the consistency gate that cross-checks constitution ↔ spec ↔ plan\n↔ **tasks** and surfaces orphan tasks and untraceable scope — and then to\n**`implement`**, which executes the list TDD-style, using `parallel` for the `[P]`\ntasks and `worktrees` for isolation.\n\nDo not jump straight to `implement`. The whole point of writing a checkable list\nwas to let `analyze` audit it cheaply *before* code exists.\n\n## Orientación (siempre)\n\nCierra cada turno con el **bloque-brújula** (📍 dónde estás · ✅ qué hiciste · 🧭 por qué · ➡️ siguiente, terminando en pregunta), calibrado al dial de `02-DOCS\u002Fwiki\u002Fharness\u002Fuser-profile.md`. **Nunca termines en seco.** Protocolo completo: skill `orient` → `skills\u002Forient\u002Freferences\u002Forientation-contract.md`. (Defiere a `suggest` el \"¿instalo la skill que falta?\".)\n","# Tasks — turn an approved plan into a verifiable work list\n\nThe `tasks` phase is the hinge between *thinking* and *doing*. The plan already\ndecided the architecture, the interfaces, and the testing strategy. This phase\nslices that plan into the **smallest units a coding agent can finish, prove, and\nhand off** — each one ordered, each one carrying a *done-check* that a machine or\na reviewer can run without trusting anyone's word.\n\nA task list is not a to-do list. A to-do list says \"build the auth endpoint\". A\ntask list says \"T004: implement `POST \u002Flogin`; done when `pytest\ntests\u002Fauth\u002Ftest_login.py` is green AND a 401 is returned for a bad password —\ndepends on T002, T003; parallel-safe with T005.\" The difference is that the\nsecond one can be **handed to a subagent, executed, and verified** with no\nfurther questions. This phase writes zero runtime code; it produces one artifact,\nappended to the plan it was built from.\n\n## Model tier — `balanced` (opt-in routing)\n\nThis phase's default tier is **`balanced`**: decomposing an approved plan is\nstructured work, not architecture. Routing is off unless `models.enabled: true`\nin `02-DOCS\u002Fwiki\u002Fsdd\u002Fconfig.yaml`; when on, resolve the tier and apply it per\n`..\u002Fsdd\u002Freferences\u002Fmodel-routing.md`, which owns the resolution order and the\nannouncement rules. Routing off or no profile → honor the session model silently.\n\n## Read the harness profile first\n\nRead `02-DOCS\u002Fwiki\u002Fharness\u002Fuser-profile.md` before producing anything and match\nthe accompaniment dial: **L0** emit the task table and nothing else; **L1** add\none line of *why this slicing* above it; **L2** justify the ordering and the\nparallel markers as you go, flagging the risky tasks; **L3** walk each dependency\nand confirm the done-checks make sense to the user before writing. No profile →\nassume non-technical and let `init` gauge the level; never invent one.\n\n## Inputs (refuse to start without them)\n\nYou break down a **plan**, not an intent. Before slicing, confirm all three exist:\n\n1. **The plan** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md`, written by the `plan`\n   phase. It must contain the architecture, interfaces, and testing strategy.\n2. **The spec** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fspecs\u002F\u003Cslug>.md`, so every task traces to a\n   requirement. A task with no spec line behind it is scope you are inventing.\n3. **The constitution** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fconstitution.md`, the non-negotiable\n   bars (stack canon, quality gates) every task must respect.\n4. **The SDD config** — `02-DOCS\u002Fwiki\u002Fsdd\u002Fconfig.yaml`, if present. Use its\n   `review_budget`, `delivery_strategy` and `testing.commands` when writing\n   done-checks and review forecasts. If it is missing on non-trivial work,\n   recommend `sdd-init`.\n\nIf the plan is missing, **stop and route to `plan`**. If the plan exists but is\nvague on interfaces or testing, route back to `plan` to tighten it — do not\npaper over a thin plan with guesswork in the task list. That guesswork is the\nexact thing `analyze` will catch next, and you will have wasted the round trip.\n\n## What makes a task well-formed\n\nEvery task is one row. A row is well-formed only when all six fields hold:\n\n| Field | Rule |\n| --- | --- |\n| **ID** | `T001`, `T002`, … — sequential, zero-padded, no hyphen (GitHub Spec Kit canon). Stable, never renumbered once written (downstream phases cite it). |\n| **[P]** | A **separate field after the ID**: `[P]` if the task is parallel-safe, blank otherwise. Never fold the marker into the ID. |\n| **Title** | One imperative verb + object. \"Implement\", \"Add\", \"Wire\", \"Migrate\" — not \"Handle\" or \"Support\". |\n| **Done-check** | The **literal** command or observation that proves done. Runnable, not a feeling. See below. |\n| **Depends-on** | The task IDs that must finish first, or `—` for none. Drives the ordering and the parallel markers. |\n| **Trace** | The spec section \u002F acceptance criterion this task satisfies. Every task traces to one; if it can't, it's out of scope. |\n\nPut `[P]` (parallel-safe) in its own field only when the task shares **no files\nand no state** with another unblocked task. When in doubt, leave it blank — a\nfalse `[P]` causes the merge collisions `parallel` exists to prevent.\n\n## The done-check is the whole point\n\nA done-check has to be **executable evidence**, owned by the stack, not a\nself-graded claim. Delegate the *form* of the check to the relevant stack skill\nand quote the actual command:\n\n- Backend (Python\u002FFastAPI) → a `pytest` invocation; route to `..\u002Ffastapi\u002FSKILL.md`.\n- Go services → a `go test .\u002F...` target; route to `..\u002Fgo\u002FSKILL.md`.\n- Frontend (Next.js) → a component\u002Fe2e test or a typed build; route to `..\u002Fnextjs\u002FSKILL.md`.\n- Flutter → a `flutter test` target; route to `..\u002Fflutter\u002FSKILL.md`.\n- Schema\u002Fmigrations → a migration applies + a query returns expected rows; route to `..\u002Fpostgresdb\u002FSKILL.md`.\n\n```text\nWEAK   done-check: \"login works\"\nSTRONG done-check: pytest tests\u002Fauth\u002Ftest_login.py::test_bad_password_returns_401  → green\n                   AND manual: POST \u002Flogin {wrong pw} returns 401, no token in body\n```\n\nA done-check you cannot run is not a done-check. If the plan's testing strategy\ndoesn't support a runnable check for a slice, that is a **plan gap** — name it\nand send it back, don't write a vague check to keep moving.\n\n## TDD-shaped slicing (default)\n\nThe `implement` phase runs red → green → refactor. Slice so it can. For each\nbehavioral unit, the natural shape is a test-first pair the implementer executes\nin order:\n\n1. **Txxx (test):** write the failing test for the behavior. *Done when the test\n   exists and fails for the right reason.*\n2. **Txxx+1 (impl):** implement until that test is green. *Done when the test\n   passes and nothing else regresses.*\n\nYou don't have to split every task in two, but the done-check of an\nimplementation task should always be *a test that was red and is now green*. This\nis where `tasks` and `implement` shake hands: the list you write is the list TDD\nwill execute.\n\n## The slicing procedure\n\n```text\n1. WALK the plan top to bottom. List every concrete deliverable\n   (endpoint, model, migration, component, job, config).\n2. For each deliverable, ask: smallest slice that produces a runnable\n   done-check? Split until each task is verifiable on its own.\n3. ORDER by dependency. Foundations first: schema → data layer →\n   service → API → UI. A task never precedes what it depends on.\n4. MARK [P] only on tasks that share no files\u002Fstate with another\n   unblocked task. Default to sequential.\n5. TRACE each task back to a spec line. No trace → cut it (scope creep)\n   or send the gap to `clarify`\u002F`specify`.\n6. ADD the cross-cutting closers: a final \"all done-checks pass\" task\n   and a \"verify.sh green\" task that hands off to `verify`.\n7. ADD a review workload + delivery forecast before implementation.\n8. APPEND the table to the plan artifact (see below). Do not start a\n   new file; the list lives with the plan it came from.\n```\n\n## Where the artifact lives\n\nThe task list is **not** a new document. Append it to the existing plan under a\n`## Tasks` heading:\n\n```text\n02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md\n  ├─ ## Architecture        (from plan)\n  ├─ ## Interfaces          (from plan)\n  ├─ ## Testing strategy    (from plan)\n  └─ ## Tasks               ← you append this\n       T001 … T0NN as the table below, + a one-line \"generated by tasks on \u003Cdate>\"\n```\n\nThen ensure the plan is indexed in `02-DOCS\u002Fwiki\u002Findex.md` (the Knowledge map;\nroot `CLAUDE.md` keeps only a short pointer) under the `sdd\u002F` topic — the `plan`\nphase usually added the row, so confirm it points at\n`02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md` and add it if missing. Additive only: never\ndelete a user's map entry.\n\n### The table format\n\n```markdown\n## Tasks\n\u003C!-- generated by tasks on 2026-06-01; IDs are stable, do not renumber -->\n\n| ID | [P] | Task | Done-check | Depends-on | Trace |\n| --- | --- | --- | --- | --- | --- |\n| T001 |  | Add `users` table migration | `alembic upgrade head` applies clean; `\\d users` shows email UNIQUE | — | spec §3 Data |\n| T002 | [P] | Add `sessions` table migration | migration applies; FK to users present | T001 | spec §3 Data |\n| T003 |  | Write failing test for `POST \u002Flogin` | `pytest tests\u002Fauth\u002Ftest_login.py` fails: no route | T001 | spec §4 Auth |\n| T004 |  | Implement `POST \u002Flogin` | T003 test green; bad pw → 401 | T003 | spec §4 Auth |\n| ... | ... | ... | ... | ... | ... |\n| T0NN |  | All done-checks pass + `verify.sh` green | every row above checked; `scripts\u002Fverify.sh` exits 0 | all | spec §Acceptance |\n```\n\n### Per-task Interfaces (for context-isolated implementers)\n\n`implement` and `parallel` dispatch tasks to **context-isolated** workers (the `developer`\nsubagent) that see *only their own task* — not the whole plan. Such a worker can't infer a\nneighbor's function signature, payload shape, or column name from a one-line row. For any task\nwhose correctness depends on a contract it doesn't own, attach an **Interfaces block** right under\nits row. Trivial, self-contained tasks don't need one — don't add ceremony where there's no\ncross-task contract.\n\n```markdown\n**T004 — Interfaces**\n- Consumes: `auth.verifyPassword(plain: str, hash: str) -> bool` (from T003); `users.email UNIQUE`\n- Produces: `POST \u002Flogin` → `200 {token}` | `401 {error}`; sets `Set-Cookie: sid=…; HttpOnly`\n```\n\nQuote **exact** signatures and shapes, not descriptions — the isolated worker copies them, it can't\ngo look them up. \"Returns the user\" is invisible; `-> {id, email}` is usable. `Consumes` names what\nthe task reads from a neighbor or the environment; `Produces` names the contract later tasks (and\nthe per-task reviewer) will hold it to. The plan's **§0 Global Constraints** are inherited by every\ntask implicitly — do **not** repeat them per task. Interfaces carry the *task-local* contract;\nGlobal Constraints carry the *project-wide* one. Together they are everything a blind implementer\nneeds.\n\n## Review workload + delivery strategy forecast\n\nAfter the task table, append a short forecast. This protects the human reviewer\nbefore a giant diff exists.\n\n```markdown\n## Review Workload Forecast\n\n| Dimension | Forecast | Why |\n| --- | --- | --- |\n| Estimated changed lines | \u003Cnumber or range> | based on task count + touched areas |\n| Files \u002F areas | \u003Ccount and names> | modules, migrations, UI, tests, docs |\n| Review risk | low \u002F medium \u002F high | complexity, cross-stack scope, security\u002Fdata changes |\n| Suggested delivery | single-pr \u002F ask-on-risk \u002F autochain \u002F exception | matched to config.review_budget |\n```\n\nPick the delivery strategy from the forecast: tiny change → `single-pr`; estimated\nlines over `config.sdd.review_budget.line_budget` (default 400) → recommend\nsplitting or `ask-on-risk`; many areas or cross-stack contracts touched →\n`autochain` or a feature-track branch. A deadline or emergency that justifies a\nlarge review is marked `exception` and requires explicit user approval later.\n\n## Anti-patterns → STOP\n\n| Tempting move | Why it's wrong \u002F Fix |\n| --- | --- |\n| \"I'll write tasks straight from the user's intent\" | You're skipping `plan`. Tasks slice a plan, not a wish. Route to `plan`. |\n| \"Done-check: 'feature works'\" | Not runnable, not evidence. Quote the literal test\u002Fcommand, or it's not a done-check. |\n| \"One giant task: 'build the backend'\" | Unverifiable, un-handoffable. Split until each slice has its own runnable check. |\n| \"Mark everything `[P]` so it goes faster\" | Shared files = merge collisions. `[P]` only when scope is truly disjoint. |\n| \"This task has no spec line but we obviously need it\" | That's scope creep wearing a hat. Trace it or cut it; if it's real, send it to `clarify`. |\n| \"The plan is thin here, I'll guess the slice\" | `analyze` will reject the guess next phase. Tighten the plan first. |\n| \"I'll start coding the easy task while I list the rest\" | This phase writes zero runtime code. Implementation is `implement`. |\n| \"Renumber the IDs so they're contiguous\" | Downstream phases cite IDs. They're stable once written. Append, don't renumber. |\n\n## Result envelope\n\nEnd with:\n\n```json result-envelope\n{\n  \"status\": \"complete\",\n  \"executive_summary\": \"Task list and review workload forecast appended to the plan.\",\n  \"artifact\": \"02-DOCS\u002Fwiki\u002Fsdd\u002Fplans\u002F\u003Cslug>.md\",\n  \"next_recommended\": \"analyze\",\n  \"risk\": \"low|medium|high\",\n  \"skill_resolution\": {\n    \"used\": [\"tasks\"],\n    \"missing\": [],\n    \"fallback\": [],\n    \"compact_rules\": [\"Every task needs a runnable done-check.\", \"Forecast review load before implementation.\"]\n  },\n  \"evidence\": [\"task table appended\", \"review workload forecast appended\"]\n}\n```\n\n## Next in the SDD chain\n\nThe task list is now the contract for the rest of the build. Hand off to\n**`analyze`** — the consistency gate that cross-checks constitution ↔ spec ↔ plan\n↔ **tasks** and surfaces orphan tasks and untraceable scope — and then to\n**`implement`**, which executes the list TDD-style, using `parallel` for the `[P]`\ntasks and `worktrees` for isolation.\n\nDo not jump straight to `implement`. The whole point of writing a checkable list\nwas to let `analyze` audit it cheaply *before* code exists.\n\n## Orientación (siempre)\n\nCierra cada turno con el **bloque-brújula** (📍 dónde estás · ✅ qué hiciste · 🧭 por qué · ➡️ siguiente, terminando en pregunta), calibrado al dial de `02-DOCS\u002Fwiki\u002Fharness\u002Fuser-profile.md`. **Nunca termines en seco.** Protocolo completo: skill `orient` → `skills\u002Forient\u002Freferences\u002Forientation-contract.md`. (Defiere a `suggest` el \"¿instalo la skill que falta?\".)\n",{"name":5,"tags":24,"origin":27,"profiles":28,"recommends":31,"description":7},[25,5,26],"sdd","breakdown","risco",[29,30],"core","full",[32,33],"analyze","implement","main","d69096e239f2a89aabc2cd078c47664a9d3b1e11","05-技能与代理开发\u002Ftasks",3,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts","https:\u002F\u002Fgithub.com\u002Flza6","个人vx：Tf00798\r\n个人✈：@TFGY999","2026-10-08 14:29:22",{"source":8,"skillName":5,"commands":43,"prompt":53,"zipAvailable":54},[44,47,50],{"tool":45,"command":46},"npx","npx skills add lza6\u002FArmor-piercing-system-prompts --skill tasks",{"tool":48,"command":49},"bunx","bunx skills add lza6\u002FArmor-piercing-system-prompts --skill tasks",{"tool":51,"command":52},"pnpm","pnpm dlx skills add lza6\u002FArmor-piercing-system-prompts --skill tasks","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F05-技能与代理开发\u002Ftasks\u002FSKILL.md",true,[56,62],{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"evals\u002Fcases.yaml","REFERENCE","参考文档",6553,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F05-技能与代理开发\u002Ftasks\u002Fevals\u002Fcases.yaml",{"path":63,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":64,"sourceHtmlUrl":65},"evals\u002FREADME.md",3764,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F05-技能与代理开发\u002Ftasks\u002Fevals\u002FREADME.md",[67,74,79,86,93,100],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M7FFG485KY83TD6HSC4YPCMC","obsidian-clipper-template-creator","Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator\u002FSKILL.md",[],[],{"skillNo":75,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M7FFG0T6AZ0STEJPSB3CX84K","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator-antigravity-awesome-skills-main\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M7FFFY0QEM14SKWPDYGMXM2P","python-api","Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP\u002FmAP semantics, and source-style import diagnostics.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Fpython-api\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":19},"01M7FFFVZQZ2GJ0C0TGP7JHD20","file-evaluation","Folder-based PASCAL VOC AP\u002FmAP evaluation from ground-truth and detection text files.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Ffile-evaluation\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":19},"01M7FFFSFNYSE77SJ6PE1CK2DA","object-detection-metrics","Guides legacy Object-Detection-Metrics PASCAL VOC AP\u002FmAP evaluation from detection text folders and direct Python API classes.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":19},"01M7FFFP0GCK1RVXBCPQ8APBMT","nx-workspace-patterns","Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fnx-workspace-patterns\u002FSKILL.md",[],[]]