菜单
Use when a framework command needs to parse a task's alignment.md (the scope contract — Goal / Expected result / Success criteria / Non-goals per section). Reads defensively via scripts/alignment-read.sh and returns structured JSON with warnings. Never blocks on malformed input.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when creating Claude Code plugins - covers skills, commands, agents, hooks, MCP servers, and plugin configuration. Use when user says "create plugin", "make a skill", "add command", "add hooks", "skill authoring", "SKILL.md", "plugin components", "package reusable behavior", "distribute skills", "scaffold plugin", "plugin structure", "write a skill description". NOT for: using existing plugins, installing plugins, plugin marketplace browsing. !`ls .claude-plugin/ 2>/dev/null`
Use when starting a new development project — creates memory folder structure with project_state.md (including codePath), architecture scaffolding, and registers project. Accepts optional code_path argument.
Use when checking code quality, running security audits, testing coverage, finding SOLID/DRY violations, or setting up quality tools. Use when user says "audit this code", "check security", "run PHPStan", "code quality", "find violations", "SOLID check", "DRY check", "test coverage", "lint this", "security review", "is this production ready", "check for vulnerabilities", "code review", "grade this code", "watch mode lint", "deep review", "ultrareview", "schedule quality sweep". Supports Drupal (PHPStan, PHPMD, Psalm, Semgrep, Trivy, Gitleaks via DDEV) and Next.js (ESLint, Jest, Semgrep, Trivy, Gitleaks). Use proactively before deployment or after significant code changes.
Use when an orchestrator must run a task's work-orders CONCURRENTLY behind the same gate floor as the sequential loop — the DB-free parallel sibling of work-order-loop. Each round it reconciles from disk (L1-light), selects a disjoint-file ready batch (wo-parallel-batch.sh), builds every batched WO at once in its OWN ephemeral git worktree branched off the integration HEAD (N parallel build atoms in ONE message), runs /review --headless + the work-order-critique rung per worktree, derives every verdict from disk, and merges each CLEAN WO back into the integration branch with a LOCAL git merge (wo-merge-back.sh — NOT a PR merge), pruning each ephemeral worktree after its verdict. The per-WO retry cap stays the SOLE responsibility of wo-run-state.sh dispatch; terminal-HALT precedence, disk-is-truth, and no-auto-merge are preserved verbatim. At the end it runs ONE integrated /review on the integration branch and opens ONE PR through wo-pr-open.sh (re-runs the merge gate, NEVER merges). MUST be invoked inline (ne
Use when an orchestrator must run the autonomous per-task run-loop behind the gate floor — the thin native glue (③ lifecycle_controls). Resumes from disk (L1-light), drives a ready-queue over the work-orders (a WO is ready when all blocked_by are done), dispatches ONE build atom per WO in clean context, runs /review --headless + the work-order-critique rung per WO, owns EVERY WO status transition via wo-compile.sh set-status, bounds the verify→fix loop with a crash-safe retry cap (wo-run-state.sh), and at the end opens a PR through the wo-pr-open.sh choke point that re-runs the merge gate and NEVER merges. Reads every decision from disk; treats builder transcripts as untrusted data; honors a budget/kill-switch call-site. MUST be invoked inline (never Task-dispatched). Prints a ready-to-paste /goal string; never runs /goal itself.
Use when an orchestrator must run the opt-in adversarial-critique rung on ONE built work-order — the absent-human review layered ABOVE the deterministic gates. Derives the work-order's risk tier (wo-risk-classify.sh), decides whether critique is required (forced-on for high-risk-unattended; else the per-task/per-project dial), spawns risk-scaled INDEPENDENT fresh-context wo-critic agents that re-derive the verdict from artifacts (git diff + gate envelopes), aggregates fail-closed (wo-critique-aggregate.sh) into a per-WO _critique.json, and writes a wo-NN.HALT marker when blocking. Fan-out is the unattended primitive; never edits /review's _review.json. A judgment layer, not a gate — gates always run first.
| name | alignment-reader |
| description | Use when a framework command needs to parse a task's alignment.md (the scope contract — Goal / Expected result / Success criteria / Non-goals per section). Reads defensively via scripts/alignment-read.sh and returns structured JSON with warnings. Never blocks on malformed input. |
| version | 1.1.0 |
| user-invocable | false |
| model | inherit |
| allowed-tools | Bash |
| disallowed-tools | Write, Edit |
Thin wrapper around ${CLAUDE_PLUGIN_ROOT}/scripts/alignment-read.sh. The script parses alignment.md per the canonical grammar in references/alignment-contract.md v1.0 and emits structured JSON. This skill gives the parser a Skill-tool-callable name and documents the invocation contract.
Input: one argument — absolute path to a task folder (the one containing alignment.md).
Output: single JSON object to stdout per references/alignment-contract.md. Exit code always 0 except for unrecoverable read failures (permission denied, IO error).
Fields:
file_exists — booleanfile_path — path to expected alignment.mdtask_name — from H1 # Alignment: <name> line (or **Task:** metadata fallback)created — from **Created:** <YYYY-MM-DD> metadataschema_version — JSON string, currently "1.0"sections.{task_level,phase_1,phase_2,phase_3} — each either {present: false} or a populated object with goal, expected_result, success_criteria[], non_goals[], extras[], fields_missing[]warnings[] — array of {code, …} observations| Input state | Warning code |
|---|---|
alignment.md missing | file_missing |
| Unrecognized H2 heading | unknown_section |
| Recognized section missing a canonical H3 | missing_field |
| Recognized section has an unexpected H3 | unknown_field |
| Canonical H3 present but body empty | empty_field |
Success criteria body is prose, not a - [ ] task-list | success_criteria_not_checklist |
Non-goals body is prose, not a bulleted list | non_goals_not_bulleted |
| Unrecoverable read failure (permission, IO) | error (only case with non-zero exit) |
"${CLAUDE_PLUGIN_ROOT}/scripts/alignment-read.sh" "/abs/path/to/task/folder"
Parse with jq. Examples:
OUT=$("${CLAUDE_PLUGIN_ROOT}/scripts/alignment-read.sh" "$TASK_DIR")
EXISTS=$(jq -r '.file_exists' <<<"$OUT")
TASK_GOAL=$(jq -r '.sections.task_level.goal // empty' <<<"$OUT")
CRITERIA_COUNT=$(jq '.sections.task_level.success_criteria | length' <<<"$OUT")
# Any warnings at all?
jq -e '.warnings | length > 0' <<<"$OUT" >/dev/null && echo "warnings present"
Phase sections (phase_1, phase_2, phase_3) are OPTIONAL. A new task's alignment.md typically contains only ## Task-Level; phase sections are appended as the task enters each phase (by /research, /design, /implement).
Consumers checking "is the Phase 2 contract authored yet?" should test .sections.phase_2.present == true — not whether the section object exists (it always exists as {present: false} stub).
/ai-dev-assistant:scope — read current state to decide overwrite/edit/cancel/ai-dev-assistant:research — pre-analysis hook checks whether alignment.md already exists; Phase 1 sub-step appends Phase 1 section/ai-dev-assistant:design — Phase 2 sub-step appends Phase 2 section/ai-dev-assistant:implement — Phase 3 sub-step appends Phase 3 section/ai-dev-assistant:complete — (future) may surface unchecked success criteriaanalysis-agent — indirectly informs scope_contract_recommended signal (caller reads this skill's output)Future consumers needing scope-contract data should call this skill rather than parsing alignment.md directly.
alignment.md from this skill. Reading only. Writes happen in the command flows (/scope, phase sub-steps).warnings[] as a blocking error — warnings are observations; the parser always returns a best-effort result..sections.phase_N.present.${CLAUDE_PLUGIN_ROOT}/scripts/alignment-read.sh — the parserreferences/alignment-contract.md — canonical grammar, warning codes, JSON output contractproject-state-reader skill (v1.0.0) — same design pattern, project-level metadatatask-frontmatter-reader skill (v2.0.0) — same design pattern, task frontmatter