feat(goals): characterize Goal instance bindings

[Duang777]

Goal And Delivered Outcome

• Goal/source and gap: Implements the M1 characterization from docs(rfc): define Goal instance identity and orphan recovery #4914. Goal action projection had no typed Goal lifetime comparison and the repository had no checked inventory of first-party binding and effect owners.
• Observable before -> after, with the validation row that proves it: goal-actions JSON now includes an observation-only goal_binding result with all seven RFC variants. A reused Effect runtime classifies retained instance A against current instance B as goal_instance_mismatch while preserving the existing action and argv.
• Issue/task and intended base: Closes [Task]: Characterize Goal instance bindings in the typed lifecycle #4927. Related design: docs(rfc): define Goal instance identity and orphan recovery #4914. Base: main.

Scope And Continuation

• Completed scope and remaining work: Adds the pure TypeScript classifier, passes raw source and binding facts through the existing action projection transaction, keeps catalog schema v1 with one additive field, and checks an 11-family owner inventory. This PR does not mint or backfill IDs, change registry or session schemas, enforce identity, mutate orphaned state, or claim ABA protection.
• Slice boundary / successor: M2 owns the local Goal lifetime transaction and commit fence. M3 owns supported effect-owner qualification. This M1 slice is observation-only and independently testable.

Validation

• Tested revision: cbbdd837f
• Run state: finished
• Input classes: synthetic, public_fixture

Check kind	Result	Public-safe evidence / limitation
unit	passed	node --experimental-strip-types --test tests/control_plane_ts/goal_instance_identity.test.ts: 4 tests cover all seven variants, malformed identity precedence, legacy/stamped mismatch, and request-local warm-process behavior.
integration	passed	Focused project registry, action projection, and inventory suite: 88 tests passed.
integration	passed	tests/control_plane_ts/authority_store.test.ts: 274 tests passed in isolation with the supported Python environment, including the four file-provider cases that failed during an earlier resource-contended full-suite run.
real_entrypoint	passed	tests/control_plane/test_goal_operator_actions.py exercises the real goal-actions CLI and executes its projected lifecycle argv verbatim.
regression_parity	passed	Legacy object and strict envelope v1 bytes remain unchanged; Markdown is byte-for-byte unchanged; removing goal_binding yields the complete prior catalog payload; mismatch does not change actions.
static	passed	npm run typecheck:control-plane, Ruff on changed Python files, git diff --check, repository hygiene, and a public-boundary scan of all 9 changed files passed.
static	passed	Standard loopx canary premerge --from-git-diff --git-diff-base origin/main: 5 direct checks and 14 selected checks passed with zero warnings.
static	failed	python examples/docs-governance-smoke.py still fails on the unmodified automatic-execution-admission-v0.md semantic-mirror marker. This PR has no documentation diff and does not change that baseline.

• Coverage and gaps: The checks cover classifier semantics, the Python-to-TypeScript production boundary, direct and global registry routes, stale and orphaned sources, warm runtime reuse, compatibility bytes, and inventory anchors. M2/M3 mutation and enforcement behavior is intentionally absent.

See validation disclosure guidance.

Frontend / Visual Evidence

• UI impact: none
• Before: N/A
• After: N/A
• States and viewports shown: N/A
• Source data: none
• Attention review: N/A; the Markdown renderer and frontend behavior are unchanged.

Type of Change

• Bug fix
• New feature
• Breaking change
• Refactoring (no functional changes)
• Documentation update
• Test update

LoopX Area

• Control plane (goals, todos, quota, scheduler, registry, runtime)
• Benchmark boundary (adapters, runners, verifiers, scoring, evidence)
• Capability or extension (providers, adapters, skills)
• Public docs or presentation surface (README, protocols, dashboard)
• Build, packaging, installer, or CI
• Host or runtime integration

Technical Direction

• Direction / acceptance reference, when applicable: Core control-plane hardening; Goal instance identity M1 from docs(rfc): define Goal instance identity and orphan recovery #4914.

Shared-authority RFC fixture impact

• Production-scale fixture schema: No production-scale fixture schema change. The existing Goal action catalog remains v1 and gains one additive observation field.
• Semantic dimensions changed, or reviewed no-impact rationale: Adds Goal instance binding classification only. Action selection, argv, lifecycle transitions, authorization, and persistence are unchanged.
• Provider conformance arms run: Direct source, fresh and stale global projection, legacy object, strict envelope v1, absent source, unavailable source, and reused Effect runtime.
• Read-only legacy/file/PostgreSQL three-arm rehearsal (required for promotion, runtime-routing, or compatibility-projection changes): Not applicable. This PR does not change authority providers, promotion, runtime routing, or provider compatibility projections.

Boundary Checklist

• Neither the diff nor this PR body/comments/attachments disclose private state, credentials, raw traces or verifier output, internal links, or local machine paths (including .loopx/, .codex/goals/, and live ACTIVE_GOAL_STATE.md).
• I did not duplicate maintainer-owned benchmark work unless a maintainer split out a public issue for it.
• I kept the change scoped to the linked issue/task.
• I completed the visual evidence section for UI changes, or marked UI impact none.
• Every commit includes a DCO Signed-off-by trailer (git commit -s).

[huangruiteng]

评审 exact head：cbbdd837f65c8ba28161115cc4ce39093bfa2951。结论：APPROVE（限 #4927 的 M1 观察与盘点，不宣称已解决同名 Goal 的 ABA）。

动机

#4914 RFC 的 M1 需要一个被真实生命周期入口调用的 typed Goal lifetime 比较，以及 first-party binding/effect owner 盘点；此前 goal-actions 只有 alias 和生命周期动作，没有实例匹配结果。此 PR 让源记录 B 与保留的全局投影 A 在同一长驻 Effect runtime 中显示 goal_instance_mismatch，同时保留原动作。M2 的事务/提交 fence、M3 的 effect-owner 资格仍是明确未交付项；M1 是可独立验收的前置切片，不是 ABA 修复完成声明。

改动思路

项目源 registry 是身份权威，global registry 只是投影/绑定；Python 沿用 _source_and_target 与现有 codec 读取事实，经 goal.operator_actions.project 的一个粗粒度请求交给 TypeScript 纯分类器。分类结果只进入现有 catalog 的附加 JSON goal_binding，不参与 lifecycleAction、授权或写入。对 malformed、缺源、缺目标和 stale 投影分别给 typed 结果，避免从路径、hash 或 lease 推断 lifetime。未引入新 capability、持久化状态或平行 Python 决策器。

具体改动

• goal_instance_identity.ts 实现七种 RFC BindingMatch 结果与输入校验；operator_actions.py/.ts 传递 source/global 事实并把结果加入 catalog，内部请求升到 v3，外部 catalog 仍为 v1。
• goal_instance_binding_inventory_v1.json 枚举 11 类一方绑定/效果 owner；架构测试核验路径与符号，registry I/O manifest 更新读点；TS config 纳入新模块与测试。
• Python/TS 测试覆盖真实 CLI、legacy/strict-envelope-v1、缺源、stale global、长驻 runtime、旧动作/Markdown 兼容。

关键代码讲解

• projectGoalBindingMatch（line 159）：在 TS 内以 discriminated unions 做精确实例比较，invalid 优先于后续状态匹配；它是只读分类，不是 admission。
• _identity_observation（line 47）和 build_goal_action_catalog（line 79）：将既有 route 的源/投影状态送入同一 Effect 请求，不 mint ID、不写 registry。
• projectGoalOperatorActions（line 88）：附加 goal_binding，仍无条件使用原 lifecycleAction 分支与 argv；即使 mismatch 也不伪装成已执行 fence。

对主干的风险

最重要的回归是新增观察暗中改变非激活 Goal 的动作，或把 stale global A 与 source B 错报为 current。我在同一路径、同一合成 registry 上对 base/head 实跑 goal-actions：删去唯一新增的 goal_binding 后，整个旧 JSON（含 fingerprint、动作 argv）完全相同。长驻 runtime 的 A→B、缺失/不可读源、legacy 与 strict envelope、无效 id 分支由 focused 测试覆盖：Python 17/17、TS 4/4、TS typecheck 通过；风险 canary 14/14 通过、无 warning。初次 inventory 测试仅因新 worktree 未装 Node parser 依赖失败，npm ci --ignore-scripts 后重跑全过。按 Goal wait_for_ci=false 未查询远端 CI。

语义与 CI 对齐

新增的是 RFC 已定义的 GoalRef/BindingMatch 观察语义。resolution_in_progress 目前只有 typed 分类与测试，没有 M2 journal 生产者；11 类 owner 盘点也不等于这些 owner 已被 fence。observe_only 与现存仍可执行的 legacy action 一起暴露，调用方不可把此字段当授权回执。

我的整体评价

9 文件、约 1.3k 行看起来较大，但行为代码约 315 行，其余主要是可审计 owner 盘点与正/负/兼容测试；相对 #4927 明确的 M1 验收是相称的。未来小范围重构应继续收敛在这个 TS Goal 边界，避免 M2 再造 Python 匹配规则；此 PR 本身无需扩为未准备好的 activation。未发现本切片的阻断性问题，批准此 exact head。合并与后续真实 ABA/并发/恢复验收仍需独立审查。

English verdict: APPROVE - At cbbdd837f65c8ba28161115cc4ce39093bfa2951, typed M1 observation and the owner inventory match #4927; real CLI base/head parity, 21 focused tests, typecheck and 14 risk canaries pass. No enforcement or ABA protection is claimed.