@event4u/agent-config 8.12.0 → 9.0.0

package/docs/benchmark.md

@@ -453,7 +453,7 @@ covers only 14/89 rules (operator hand-labelling), a separate gap.
 - Runner: `src/scripts/bench_quality_run.ts` (`--dry-run` free / live is
   API-gated); gate: `src/scripts/check_quality_regression.ts` (inert until a
   `quality-run.json` — gitignored — exists locally).
-- Roadmap: `agents/roadmaps/road-to-token-saving.md` Phase 0.
+- Roadmap: `agents/roadmaps/later/road-to-token-saving.md` Phase 0.
 
 ## Length-neutral judge RERUN (2026-07-12) — SECOND INCONCLUSIVE, gate CLOSED-BY-DIAGNOSIS
 
@@ -492,4 +492,69 @@ anchor-scoring against `must_include`/`must_not`, not pairwise LLM judging).
 
 - Runner: `src/scripts/bench_quality_rerun.ts` (design-pinned constants:
   band ±15%, κ floor 0.60, ρ flag 0.3, effect floor 10pp; `--max-usd` guard).
-- Roadmap: `agents/roadmaps/road-to-opt-measurement-unblock.md` Phase 1.
+- Roadmap: `agents/roadmaps/archive/road-to-opt-measurement-unblock.md` Phase 1.
+
+## Cross-model parity count (2026-07-12) — PASS, `finding_floor` armed
+
+The most-deferred portfolio item (cross-model parity eval, deferred across 4+
+roadmaps) landed in its re-scoped form: council-transport execution instead of
+an in-host subagent harness (`docs/design/cross-model-parity-eval.md`). Each
+orchestration-corpus task is rendered self-contained (fixture files inlined)
+and dispatched identically to two vendors; the output contract forces a
+numbered findings list (or `NO FINDINGS`), counted with the **same**
+`_count_findings` the `finding_floor` eval gate uses.
+
+| Task | sonnet median | gpt-4o median | Calibrated floor |
+|---|---|---|---|
+| orch-01 multi-file analysis | 11 | 5 | 5 |
+| orch-02 ordered refactor | 5 | 3 | 3 |
+| orch-03 competitive impl | 3 | 0 | 1 (clamp) |
+| pv-01 hollow detection | 2 | 2 | 2 |
+| pv-02 negative control | 0 | 0 | — (control, excluded) |
+
+Signal, honestly stated:
+
+- **Real cross-vendor gap** — sonnet surfaces ~2× gpt-4o's findings on the
+  multi-file analysis task. A floor calibrated on sonnet alone would
+  systematically fail gpt-4o; the cross-host lower envelope
+  (`max(1, min over hosts of median)`) is the correct floor shape.
+- **Planted defect is vendor-stable** — both vendors cite the hollow
+  `charge.ts` (2 findings each, zero variance across repeats).
+- **Negative control perfect** — 0 findings from both vendors on clean code
+  across all repeats; the counting contract does not reward spurious findings.
+
+Cost: $0.16 actual (30 calls; ceiling $8 authorized). `finding_floor` is now an
+**enforcing** gate (comment flipped in `run_skill_evals.ts`); calibration data
+in `internal/bench/reports/parity-count.json`.
+
+- Runner: `src/scripts/bench_parity_count.ts` (pre-registered: 3 repeats,
+  2 vendors, envelope floor rule, controls excluded, `--max-usd` abort).
+- Roadmap: `agents/roadmaps/archive/road-to-opt-measurement-unblock.md` Phase 3.
+
+## Non-Claude lift replication, second vendor host (2026-07-12) — HONEST NULL
+
+The `discipline_profile: auto` flip gate required a replicated lift on a
+non-Claude host. Executed per the parked design (paired vanilla vs
+`rules-kernel-dc` = the essential cut, 30 tasks × 3 seeds) on the codex CLI
+host with `gpt-5.5` (ChatGPT account).
+
+- **Capability:** 92% → 89% (McNemar p = 1.0) — no effect.
+- **Discipline:** 1.000 → 0.892 (Δ = −0.108, Wilcoxon p = 0.0225 on 7
+  non-zero pairs) — no lift; directionally **negative** on this host.
+- **Verdict:** NO replicated lift. `discipline_profile` keeps its
+  vendor-granular default (`unknown_defaults`, no `auto` flip) — consistent
+  with the 2026-07-10 two-host precedent (claude strong-host ceilings) and
+  the earlier gpt-5-mini replication failure on the same injection surface.
+
+Limitations, published: the ChatGPT-account usage limit rejected 100 of 180
+runs mid-corpus (plus 2 timeouts), leaving n = 37 valid pairs — the
+truncation is **balanced across arms** (50/52), so it weakens power but does
+not bias the comparison. An earlier same-day fire misfired entirely (model
+pin omitted → all runs API-rejected, $0) and is excluded as a harness error,
+not evidence. Raw report is local-only by design
+(`internal/bench/reports/ab-v2/` is gitignored); this section + the roadmap
+notes are the durable record. Re-open path: a full uncapped 180-run
+replication on a per-token-billed key.
+
+- Runner: `src/scripts/bench_ab_v2_run.ts --host codex` (checkpoint-resumable).
+- Roadmap: `agents/roadmaps/archive/road-to-opt-measurement-unblock.md` Phase 2.

package/docs/catalog.md

@@ -1,6 +1,6 @@
 # agent-config — Public Catalog
 
-Consumer-facing catalog of all **642 public artefacts** shipped by
+Consumer-facing catalog of all **648 public artefacts** shipped by
 this package. Internal package-maintenance rules and deprecation shims
 are excluded.
 
@@ -179,7 +179,7 @@ are excluded.
 | skill | [`messaging-architecture`](../dist/agent-src/skills/messaging-architecture/SKILL.md) |  | Use when shaping the primary message, supporting proofs, and audience-by-message matrix from a locked positioning frame — before any copy or launch beat. Triggers on 'tighten the message stack'. |
 | skill | [`migration-architect`](../dist/agent-src/skills/migration-architect/SKILL.md) |  | Use when shaping a non-trivial migration — rollout phases, dual-write windows, cutover sequencing, deprecation cycles — hands off to the framework-specific migration skill for DDL once locked. |
 | skill | [`mobile-e2e-strategy`](../dist/agent-src/skills/mobile-e2e-strategy/SKILL.md) |  | Use when picking a mobile E2E framework — Detox / Appium / Maestro / XCUITest / Espresso — or planning iOS Simulator / Android Emulator coverage in CI for RN, Expo, or native apps. |
-| skill | [`module-detect-on-the-fly`](../dist/agent-src/skills/module-detect-on-the-fly/SKILL.md) |  | Use when editing a file under a module-shaped path (`Modules/*`, `packages/*`, `apps/*`, `internal/*`) while `modules.enabled` is `false` — asks the user once whether to enable the config. |
+| skill | [`module-detect-on-the-fly`](../dist/agent-src/skills/module-detect-on-the-fly/SKILL.md) |  | Use when editing a module-shaped path (`Modules/*`, `packages/*`, `apps/*`) while `modules.enabled` is false — asks once to enable it; also the project/stack + task-runner detection reference. |
 | skill | [`module-management`](../dist/agent-src/skills/module-management/SKILL.md) |  | Use when working within any module under `modules.root_paths` from `.agent-project-settings.yml` — Laravel HMVC, Symfony DDD-lite, Node monorepo, Python src/, Go internal/, or a custom path. |
 | skill | [`motion-choreographer`](../dist/agent-src/skills/motion-choreographer/SKILL.md) |  | Use when turning a locked still + blueprint into a provider-tuned motion prompt — camera, primary + secondary motion, physics, native-audio sync. Triggers 'motion prompt for Veo/Kling/Sora'. |
 | skill | [`multi-tenancy`](../dist/agent-src/skills/multi-tenancy/SKILL.md) |  | Use when working with the multi-tenant architecture — customer DB switching, FQDN routing, tenant isolation, or cross-tenant operations. |
@@ -407,7 +407,7 @@ are excluded.
 | rule | [`user-interrupt-priority`](../dist/agent-src/rules/user-interrupt-priority.md) | auto | New user instruction mid-flight — STOP the current task, run the new one in full, ASK before resuming |
 | rule | [`verify-before-complete`](../dist/agent-src/rules/verify-before-complete.md) | always | Verify before completion — run tests and quality tools before claiming done |
 
-## Commands (178)
+## Commands (183)
 
 | kind | name | cluster | description |
 |---|---|---|---|
@@ -569,6 +569,11 @@ are excluded.
 | command | [`team-knowledge`](../dist/agent-src/commands/team-knowledge.md) | cluster: team-knowledge | Team-knowledge orchestrator — routes to consolidate and bootstrap |
 | command | [`team-knowledge-bootstrap`](../dist/agent-src/commands/team-knowledge/bootstrap.md) | cluster: team-knowledge | One-shot deterministic seed for a fresh project's knowledge layer — stages template pages from real config/directory detection, never LLM-invented claims. Review-then-commit. |
 | command | [`team-knowledge-consolidate`](../dist/agent-src/commands/team-knowledge/consolidate.md) | cluster: team-knowledge | Review pending typed knowledge-observation events and file them into agents/knowledge/ pages as a human-reviewed batch — never writes without approval. |
+| command | [`team`](../dist/agent-src/commands/team.md) | cluster: team | Team orchestrator — governed cross-model access layer (a second strong model reviews the real diff; read-only multi-host fallback); routes to review, adversarial, delegate, status |
+| command | [`team-adversarial`](../dist/agent-src/commands/team/adversarial.md) | cluster: team | Thin wrapper — adversarial cross-model review on a named focus via the official plugin (/codex:adversarial-review). Escalation rung above the single-model adversarial-review skill. |
+| command | [`team-delegate`](../dist/agent-src/commands/team/delegate.md) | cluster: team | Thin wrapper — hand a task to the second model as a native worker via the official plugin (/codex:rescue). The only write-access wrapper; double-gated behind ai_team.allow_delegate. |
+| command | [`team-review`](../dist/agent-src/commands/team/review.md) | cluster: team | Thin wrapper — cross-model review of the current diff via the official plugin (/codex:review). Gated on ai_team.enabled; fails closed when the plugin is absent. |
+| command | [`team-status`](../dist/agent-src/commands/team/status.md) | cluster: team | Thin wrapper — plugin job status via /codex:status plus a quota block showing the shared openai counter against both ceilings (team + council). Gated on ai_team.enabled; fails closed when the plugin… |
 | command | [`tests`](../dist/agent-src/commands/tests.md) | cluster: tests | Tests orchestrator — routes to create, execute, e2e-plan, e2e-heal |
 | command | [`tests-create`](../dist/agent-src/commands/tests/create.md) | cluster: tests | Write meaningful tests for the current branch — stack-adaptive (pest / phpunit / vitest / jest / pytest / …) |
 | command | [`tests-e2e-heal`](../dist/agent-src/commands/tests/e2e-heal.md) | cluster: tests | Find, debug, and fix failing Playwright E2E tests |
@@ -590,7 +595,7 @@ are excluded.
 | command | [`worktree-status`](../dist/agent-src/commands/worktree/status.md) | cluster: worktree | List active worktrees — ownership (scope lock), dirty state, ahead/behind, merge-readiness incl. verification evidence |
 | command | [`worktree-verify`](../dist/agent-src/commands/worktree/verify.md) | cluster: worktree | Run the scoped verification for a worktree's declared change — narrow probes matched to the diff, never the full CI pipeline |
 
-## Guidelines (90)
+## Guidelines (91)
 
 | kind | name | category | description |
 |---|---|---|---|
@@ -629,6 +634,7 @@ are excluded.
 | guideline | [`roadmap-progress-mechanics`](../docs/guidelines/agent-infra/roadmap-progress-mechanics.md) | agent-infra |  |
 | guideline | [`role-contracts`](../docs/guidelines/agent-infra/role-contracts.md) | agent-infra |  |
 | guideline | [`role-mode-router`](../docs/guidelines/agent-infra/role-mode-router.md) | agent-infra |  |
+| guideline | [`rule-body-migration-inventory`](../docs/guidelines/agent-infra/rule-body-migration-inventory.md) | agent-infra |  |
 | guideline | [`rule-type-governance`](../docs/guidelines/agent-infra/rule-type-governance.md) | agent-infra |  |
 | guideline | [`runtime-layer`](../docs/guidelines/agent-infra/runtime-layer.md) | agent-infra |  |
 | guideline | [`scqa-framework`](../docs/guidelines/agent-infra/scqa-framework.md) | agent-infra |  |

package/docs/command-flows.md

@@ -7,7 +7,7 @@
 
 > Every command is classified in `src/flows/surface-map.yaml` (coverage enforced by `scripts/lint_command_flow_coverage.py`). The four user-work flows carry rich definitions in `src/flows/<flow>.yaml`.
 
-**178 commands** across **10 surfaces** (4 user-work flows + 6 platform/domain surfaces).
+**183 commands** across **10 surfaces** (4 user-work flows + 6 platform/domain surfaces).
 
 ## User-work flows
 
@@ -135,6 +135,11 @@ _Deliberately-started features — not daily flow members, never buried as skill
 - `roadmap-process-full` — Autonomously process every open step across every phase of a roadmap until the file is fully closed. Larges…
 - `roadmap-process-phase` — Autonomously process every open step in the next or current phase of a roadmap, then stop. Default executio…
 - `roadmap-process-step` — Autonomously process the single next open step of a roadmap and stop. Smallest execution scope of the /road…
+- `team` — Team orchestrator — cross-model depth review (one strong model builds, a second reviews the real diff); rou…
+- `team-adversarial` — Thin wrapper — adversarial cross-model review on a named focus via the official plugin (/codex:adversarial-…
+- `team-delegate` — Thin wrapper — hand a task to the second model as a native worker via the official plugin (/codex:rescue).…
+- `team-review` — Thin wrapper — cross-model review of the current diff via the official plugin (/codex:review). Gated on ai_…
+- `team-status` — Thin wrapper — plugin job status via /codex:status plus a ledger line with today's cli_call_budget openai c…
 
 ### Platform surface — `agent-admin`
 

package/docs/contracts/ai-council-config.md

@@ -787,11 +787,18 @@ error) when any of these hold:
 
 ## Migration footprint (Phase 0)
 
+> Historical record of the Step-2 consolidation. **The project-local file
+> described below is no longer read** — ADR-104 (superseding ADR-093) moved
+> the council config to the user-global location documented at the top of
+> this contract. A `.ai-council.yml` under any project `agents/` tree is
+> ignored. Read this section for provenance only, not for placement.
+
 - `.agent-settings.yml` → 14-key inventory under `ai_council:` removed
   after a one-line breadcrumb comment is in place pointing at this
   contract.
 - `.agent-settings.template.yml` → same removal.
-- New file `agents/settings/.ai-council.yml` checked in with two enabled
+- New file `agents/settings/.ai-council.yml` checked in with two enabled <!-- council-config-allowed -->
   providers (anthropic + openai) and three disabled
   (gemini + xai + perplexity). Models pre-filled from the Phase 0
-  default set; comments mirror this contract.
+  default set; comments mirror this contract. *(This project-tracked file
+  was later removed; per ADR-104 the config is user-global only.)*

package/docs/contracts/ai-team-config.md

@@ -0,0 +1,224 @@
+---
+stability: beta
+keep-beta-until: 2026-10-10
+---
+
+# AI-Team Config (`ai_team` block)
+
+**Purpose.** Lock the schema, validation, and quota rules for the `/team`
+cross-model review family — one strong external model (via the codex CLI,
+subscription-authed) reviewing inside the repo. Team mode is the **depth**
+complement to the council's **breadth**; this contract is the boundary that
+prevents drift across the settings template, the loader, the `/team`
+wrappers, and the quota counter.
+
+**Audience.** Authors of `src/scripts/ai_team/config.ts`, the `/team`
+command family, the doctor `team` section, and the setup wizard. Also
+reviewers checking that new team features keep the schema intact.
+
+**Status.** Internal-locked. Changes require a contract version bump and a
+revision entry in the consuming roadmap.
+
+## File location
+
+Unlike the council (ALWAYS user-global `.ai-council.yml` — see
+[ai-council-config.md](ai-council-config.md)), the `ai_team` block lives in
+the **project settings cascade**:
+
+```
+<project>/.agent-settings.yml   →   ai_team: { … }
+```
+
+read via `load_agent_settings` (`src/scripts/_lib/agent_settings.ts`) and
+validated by `src/scripts/ai_team/config.ts:load_ai_team_config`. Team mode
+is a per-project posture (default-off), not a per-developer credential
+store — the codex CLI carries the subscription auth itself, so there is
+nothing secret to centralize. The shipped defaults live in
+`src/config/agent-settings.template.yml` and are mirrored by the wizard
+schema (`src/server/schemas/settings.ts`, enforced by the schema↔template
+parity test).
+
+## Schema
+
+```yaml
+ai_team:
+  enabled: <bool>                 # master switch, default false
+  model: <string>                 # default 'auto'
+  allow_delegate: <bool>          # second opt-in for the write path, default false
+  max_calls_per_day: <int >= 0>   # default 50; counts into the shared openai bucket
+  suppress_setup_hint: <bool>     # default false; cosmetic wizard-hint suppression
+  review_gate:                    # managed governance of the plugin's Stop-hook Review Gate
+    managed: <bool>               # default false = byte-identical pre-Phase-4 behavior
+    max_consecutive_blocks: <int >= 1>  # default 3; circuit-breaker loop bound
+```
+
+| Key | Type | Default | Semantics |
+|---|---|---|---|
+| `enabled` | bool | `false` | Master switch. `false` = `/team` commands are never suggested and every invocation refuses with an enable pointer — byte-identical to pre-feature behavior. `true` = the family is live; per-command gates below still apply. |
+| `model` | string | `'auto'` | Model handed to the codex CLI. `'auto'` = pass **no** `--model` flag; the codex CLI's own default applies. Any other value passes through **verbatim** as `--model <value>`. |
+| `allow_delegate` | bool | `false` | Second opt-in for the **only** wrapper that delegates write access (`/team:delegate`). Refuses with an enable pointer until BOTH `enabled` and `allow_delegate` are `true`. |
+| `max_calls_per_day` | int ≥ 0 | `50` | Per-day ceiling on team calls, read against the **shared** `cli_call_budget` openai bucket (see § Quota). `0` blocks all team calls. |
+| `suppress_setup_hint` | bool | `false` | Suppress the one-line wizard/init recommendation to install the codex plugin on Claude-Code hosts. Cosmetic only. |
+| `review_gate.managed` | bool | `false` | Managed governance of the codex plugin's Stop-hook Review Gate (Phase 4). `false` = no counting, no circuit breaker — the Stop path is byte-identical to pre-Phase-4 dispatch. `true` = consecutive BLOCK verdicts are counted per session (first-line `ALLOW:`/`BLOCK:` contract of the gate transcript; anything else is honestly `UNKNOWN` and never counted) and the circuit breaker trips at the bound. |
+| `review_gate.max_consecutive_blocks` | int ≥ 1 | `3` | Circuit-breaker bound. At this many CONSECUTIVE BLOCKs in one session a visible notice is injected **exactly once** and the managed layer stops re-blocking — the user decides, never an infinite Claude↔Codex loop. An ALLOW verdict resets the counter. Unknown keys inside `review_gate` are rejected fail-closed, same as the parent block. |
+
+### Why `model: 'auto'` instead of a pinned ID
+
+A hardcoded model ID rots silently (the council's own CLI default is a
+deliberate, documented *pin* — `DEFAULT_OPENAI_CLI_MODEL` in
+`src/scripts/ai_council/clients.ts`, see
+[ai-council-config.md](ai-council-config.md)). Team mode inverts that:
+under a subscription the *strongest* available model is the rational
+default, and only the codex CLI knows what that currently is. `'auto'`
+therefore delegates the choice to the CLI; a set value is a user pin and
+passes through verbatim — the loader never rewrites or "corrects" it.
+
+**`review_gate` (Phase 4, shipped):** the loop-bound governance block is
+live — module `src/scripts/ai_team/review_gate.ts` (counter state under
+`agents/runtime/state/team-review-gate.json`, ledger under
+`agents/runtime/team/events.log`, one `team.gate: BLOCK 2/3` /
+`ALLOW reset` line per gate verdict). The gate itself stays upstream's;
+upstream's own warning applies while it is enabled unmanaged: "The
+review gate can create a long-running Claude/Codex loop and may drain
+usage limits quickly." Any other key stays rejected fail-closed.
+
+## Role semantics — the council-verdict design constraint
+
+Verbatim from the consuming roadmap's council verdict, binding on every
+team-mode phase:
+
+> **Council verdict on team-mode frontmatter (claude-sonnet-4-5 + gpt-4o,
+> 2026-07-12, unanimous REJECT):** skills/commands get NO `team_mode` /
+> role frontmatter. Role semantics belong in the `ai_team` config block +
+> prompt library (reusing subagent-orchestration's implementer/judge frame
+> and status envelope), not smeared across 271 skill files while the
+> feature's role model does not operationally exist. Revisit-if: Phase 2
+> execution turns out to be blocked SPECIFICALLY on artefact-level role
+> metadata — then re-open via decision-revisit-gate, not by silently
+> adding the key.
+
+Operationally: ALL role semantics (builder/reviewer/verifier selection,
+eligibility) live HERE — in this `ai_team` block — plus the prompt
+library, reusing `subagent-orchestration`'s implementer/judge frame and
+its `subagent-status.json` envelope. **No `team_mode`/role frontmatter key
+is added to skill/command/rule schemas.**
+
+## Quota — one subscription, one counter
+
+```
+TEAM CALLS COUNT INTO THE EXISTING cli_call_budget OPENAI BUCKET.
+ONE SUBSCRIPTION, ONE COUNTER. NEVER A PARALLEL COUNTER.
+```
+
+The daily CLI-call counter is already **generic per-provider** — no
+team-specific counting system exists or may be built:
+
+- **State file:** `~/.event4u/agent-config/cli-calls.json`
+  (`CLI_CALLS_FILENAME`, `src/scripts/ai_council/clients.ts:716`), shape
+  `{ date: "YYYY-MM-DD", counts: { <provider>: <int> } }`, resets on UTC
+  date rollover.
+- **Read:** `load_cli_call_counts(path?)` —
+  `src/scripts/ai_council/clients.ts:738`. Returns today's per-provider
+  counts; empty on rollover.
+- **Increment:** `record_cli_call(provider, path?)` —
+  `src/scripts/ai_council/clients.ts:774`. Returns the new total.
+- **Gate + increment on the transport itself:** `CliClient.ask()` checks
+  the counter *before* spawning (`clients.ts` § quota gate, returns
+  `error: 'cli_quota_exhausted'` + a `block_quota` events-log line when
+  exhausted) and records the call *after* — even failed calls count, so a
+  broken CLI cannot burn the budget in a tight loop.
+- **The openai bucket:** `OpenAICliClient` (`clients.ts:1275`) has
+  `name = 'openai'`; council CLI calls and team calls land on the **same**
+  `counts.openai` entry.
+
+A `/team` invocation therefore reuses the existing seams, in preference
+order:
+
+1. **Through the transport** — construct `OpenAICliClient` with
+   `max_calls_per_day: <ai_team.max_calls_per_day>`; the gate, the
+   exhausted-error, the events-log line, and the increment all come for
+   free.
+2. **Direct** (for `/team:status` rendering or a pre-flight check) —
+   `load_cli_call_counts()['openai'] ?? 0` to read,
+   `record_cli_call('openai')` to count. No new helper exists because
+   none is needed; adding a team-side duplicate of this counter is a
+   contract violation.
+
+`ai_team.max_calls_per_day` and the council's
+`cli_call_budget.max_calls_per_day.openai` are two *ceilings* over the
+same *counter*: whichever path a call takes, it increments the one shared
+bucket, and each surface enforces its own configured cap against that
+shared count.
+
+**Worked example — two ceilings, one counter.** Council ceiling
+`cli_call_budget.max_calls_per_day.openai: 100`, team ceiling
+`ai_team.max_calls_per_day: 50`, shared counter at 45 openai calls today:
+both paths are open. Five team calls later the counter reads 50 — team
+calls are now blocked (50 >= 50) while council CLI calls continue until
+the counter reaches 100. Whichever path fires a call, the same counter
+moves; each surface only decides where *its* ceiling sits on that shared
+count. `/team status` renders exactly this state: the live counter, both
+ceilings, and which path (if any) is currently blocked.
+
+### Team-review envelope
+
+Every `/team:review` return — plugin path and multi-host fallback alike — is emitted in the
+team-review status envelope (`src/skills/subagent-orchestration/schemas/team-review-status.json`),
+which extends the `subagent-status.json` frame: `status` (`DONE | DONE_WITH_CONCERNS |
+NEEDS_CONTEXT | BLOCKED`), `findings[]` (`severity`/`evidence`/`suggested_fix`/`location?`),
+`reviewed_ref` (HEAD sha of the bundle), `model` (`'auto'` = codex CLI default), and
+`quota` (`{used, ceiling}` on the shared `cli_call_budget` openai bucket); unparseable model
+output is preserved verbatim in `raw` with status `DONE_WITH_CONCERNS` — never silently dropped.
+Plugin text is summarized INTO the envelope and preserved verbatim beneath — never rewritten.
+<!-- PLACEHOLDER: the team-review-envelope subsection (review-gate ledger
+     line format `team.gate: BLOCK n/N` read by /team status) is inserted
+     here by the orchestrator in a later step. Do not remove this marker. -->
+
+## Validation rules (fail-closed)
+
+`load_ai_team_config` / `build_ai_team_config`
+(`src/scripts/ai_team/config.ts`) raise `TeamConfigError` when:
+
+1. `ai_team` is present but not a mapping,
+2. any **unknown key** appears under `ai_team` (a typo must never
+   silently disable a gate),
+3. `enabled` / `allow_delegate` / `suppress_setup_hint` is not a boolean,
+4. `model` is not a non-empty string,
+5. `max_calls_per_day` is not a non-negative integer (booleans rejected).
+
+An **absent** `ai_team` block is not an error — it yields the defaults
+(`AI_TEAM_DEFAULTS`) and the feature stays off.
+
+## Default-off posture
+
+Every team-mode capability is default-off. With the block absent or
+`enabled: false`, all paths are byte-identical to pre-feature behavior:
+no command suggestions, no wrapper execution, no quota consumption, no
+wizard nagging beyond the single suppressible setup hint.
+`/team:delegate` is doubly gated (`enabled` AND `allow_delegate`).
+
+## No-claims constraint
+
+No public claim about cross-model review quality ("a second model finds
+errors the first never sees", "two strongest models catch each other's
+errors", or any review-lift assertion) may exist anywhere in the tree
+until the pre-registered three-arm defect-finding benchmark has produced
+a verdict bound in CLAIMS.md. Workflow-value prose is allowed; lift
+claims are not. An honest null keeps the feature documented as workflow
+value only.
+
+## Boundary vs. the council
+
+Team mode and the council are complements, not variants — the council's
+neutrality contract is untouched by team mode:
+
+| Axis | Council ([ai-council-config.md](ai-council-config.md)) | Team (`ai_team`) |
+|---|---|---|
+| Shape | N members, breadth | One strong reviewer, depth |
+| Repo access | NEVER (artefact text + neutral preamble only) | Full (reviewer runs inside the working tree) |
+| Config location | User-global `.ai-council.yml` (ADR-104) | Project `.agent-settings.yml` cascade |
+| Default models | Mid-tier, cost-conscious | Subscription's strongest (`model: 'auto'`) |
+| Quota | `cli_call_budget` per provider | Same counter — shared openai bucket |
+
+This link is deliberately one-directional (team → council); the council
+contract does not reference team mode.

package/docs/contracts/command-clusters.md

@@ -45,6 +45,7 @@ column 1 of this table.
 | `council` | 3 | `default` · `pr` · `design` · `optimize` · `analysis` | `council` (legacy default lens) · `council-pr` · `council-design` · `council-optimize`; `analysis` added 2026-05-14 — wrapper for local analysis outputs with a Top-N consensus tail block consumed by `/roadmap create` |
 | `challenge-me` | — | `vision` · `with-docs` | new — Pocock-inspired one-question-at-a-time interview; `vision` is the standard 95%-confidence variant, `with-docs` adds doc/glossary awareness with a session-scoped glossary and load-bearing claim-vs-code verification |
 | `research` | 2 | `deep` · `report` | preliminary-research scaffolder ported from `Weizhena/Deep-Research-skills` (cluster head emits `outline.yaml` + `fields.yaml` against the `research-schema` contract). `:deep` populates per-item JSON in batches with native web-search + JSON-Schema self-validation (no Python runtime); `:report` renders `report.md` directly + optionally emits a `jq` template for deterministic regeneration. `add-items` / `add-fields` intentionally **not** ported — re-run `/research <topic>` to extend the field framework. |
+| `team` | 2 | `review` · `adversarial` · `delegate` · `status` | new 2026-07-12 (road-to-team-mode Phase 2) — cross-model depth review: thin fail-closed delegations to the official codex plugin on Claude-Code hosts under `ai_team` governance ([`ai-team-config`](ai-team-config.md)); default-off (`ai_team.enabled: false`), `delegate` behind the second opt-in `allow_delegate`; internal visibility until the ADR promotion gate |
 | `orchestrate` | — | _(none yet — cluster head only)_ | new — runtime executor for YAML pipelines under `.agent-config/orchestrations/` per the [`orchestration-dsl-v1`](orchestration-dsl-v1.md) contract; chains personas / skills / commands / sub-agents deterministically. Single cluster head; sub-commands deferred until a second verb is needed. |
 | `sync` | 4 | `agent-settings` · `gitignore` · `gitignore-fix` | new cluster 2026-07-08 — absorbs the former single-sub `sync-gitignore` cluster (slug-neutral nests: `sync-agent-settings`, `sync-gitignore`, `sync-gitignore-fix` keep their slugs; the gitignore node keeps its own `fix` sub-dir, mirroring the `agents/user` deep-nest precedent). Bare `/sync` shows the menu. |
 | `ghostwriter` | — | `fetch` · `write` · `list` · `show` · `delete` | new cluster 2026-05-15 — third voice primitive for AI-assisted writing in the public voice of a public figure. Hybrid storage: real-person profiles live consumer-side under `agents/reference/ghostwriter/<slug>.md` (gitignored by default); package source ships only `fictional: true` fixtures. Zero network code in the package — `:fetch` delegates web-fetch / web-search to the host agent. Mandatory disclosure footer on every `:write` output (no opt-out). Schema: [`ghostwriter-schema`](ghostwriter-schema.md). |

package/docs/contracts/prelaunch-diagnostics.md

@@ -0,0 +1,151 @@
+# Pre-Launch Diagnostics — finding IDs, epistemics, regression gate
+
+> Contract for the consumer-facing launch diagnostic
+> (road-to-ecosystem-harvest-prelaunch-diagnostics, Source L — a
+> production-readiness diagnostic skill; provenance in the harvest index).
+> The consumer-side sibling of the suite's own claims-ledger: a "ready"
+> verdict cannot be reached by assertion.
+>
+> Schema: [`prelaunch-report.schema.json`](../../src/scripts/schemas/prelaunch-report.schema.json) ·
+> Area vocabulary: [`prelaunch-areas.yml`](../../src/config/prelaunch-areas.yml) ·
+> Tooling: [`prelaunch_diagnostics.ts`](../../src/scripts/prelaunch_diagnostics.ts) ·
+> Skill surface: [`launch-readiness`](../../src/skills/launch-readiness/SKILL.md).
+
+## 1 — Finding-ID grammar (immutable, diff-stable)
+
+```
+AC-<AREA>-NNN        e.g. AC-AUTH-003, AC-RBK-001
+```
+
+- `<AREA>` — the uppercase code of a coverage area from
+  [`prelaunch-areas.yml`](../../src/config/prelaunch-areas.yml). The vocabulary
+  is **append-only**: codes are never renamed or re-used.
+- `NNN` — zero-padded, monotonically assigned per area, **never re-assigned**.
+  A finding that is retitled, re-scoped, or resolved keeps its ID forever; a
+  genuinely new problem gets the next number. Resolved findings stay in the
+  report with a flipped `status` — the entry never disappears, so baselines
+  diff cleanly across months.
+- The **ID is the diff key**. Titles, evidence, and severity are mutable
+  metadata; two runs are compared by ID set + per-ID state, never by prose.
+
+## 2 — Epistemics (Pass needs evidence; Unknown ≠ Pass)
+
+Every coverage area carries exactly one state:
+
+| State | Meaning | Requirement |
+|---|---|---|
+| `pass` | Diagnosed clean | **Cited evidence required** — file:line, command output, run URL. A pass without evidence fails validation: that is assertion, not diagnosis. |
+| `finding` | At least one open finding | The finding IDs live in `findings[]`. |
+| `unknown` | Not diagnosed | The honest default. **Absence of findings is `unknown`, never an automatic `pass`.** An area missing from the report entirely is `unknown` by definition. |
+| `not-applicable` | Does not exist on this surface | **Reason required** ("no DB → no migrations"). |
+
+Verdict rule (the launch gate):
+
+```
+ANY OPEN P0 FINDING → NOT READY, REGARDLESS OF EVERY OTHER AREA.
+ANY OPEN P1, OR A launch_gate AREA NOT AT pass/not-applicable → NOT READY.
+READY IS THE RESIDUAL STATE, NEVER THE DEFAULT.
+```
+
+`launch_gate: true` areas (auth, migrations, secrets, rollback) block the
+verdict while `unknown`; non-gate areas (observability, agent-governance)
+surface as unknowns + questions but do not block alone.
+
+## 3 — Coverage backbone
+
+The area vocabulary is deliberately pruned to the stacks this suite targets:
+**auth · migrations · secrets · observability · rollback · agent-governance**.
+It is not a generic compliance matrix — new areas enter by appending to
+`prelaunch-areas.yml` with a maintainer-reviewed `launch_gate` decision.
+
+Every report may carry a `questions` list — **questions that would change the
+diagnosis** ("is the admin panel exposed on the same origin?"). Questions are
+the actionable rendering of unknowns; the next run should convert them into
+evidence or findings.
+
+## 4 — Regression gate (findings diff)
+
+`prelaunch_diagnostics.ts diff <baseline.json> <current.json>` compares two
+reports **by ID**:
+
+- **new** — ID in current, not in baseline.
+- **resolved** — open in baseline, `fixed` in current.
+- **regressed** — a `launch_gate` area that was `pass` in the baseline and is
+  `finding`/`unknown` in current (Pass→Finding flip).
+- **retitled findings do not appear in the diff** — the ID survives the
+  rename (acceptance-tested).
+
+`--ci` mode exits non-zero when the diff contains a **new open P0/P1** or a
+**Pass→Finding flip in a launch-gate area** — the consumer-side sibling of the
+suite's claims-ledger CI. Everything else (new P2/P3, resolved findings,
+question churn) reports informationally and exits 0.
+
+## 5 — Suppression (evidence-backed, never silent)
+
+A consumer may suppress a finding via the project-local suppression file
+`agents/settings/prelaunch-suppressions.yml`:
+
+```yaml
+schema_version: prelaunch-suppressions-v1
+suppressions:
+  - id: AC-OBS-002
+    reason: "Dashboards land with the Q3 observability epic; alerting for the launch path exists."
+    evidence: "agents/evidence/prelaunch/obs-alerting.png"
+```
+
+- A finding with `status: suppressed-with-evidence` MUST have a matching
+  suppression entry carrying **both** `reason` and `evidence` — validation
+  fails otherwise.
+- Suppressed findings are excluded from the `--ci` triggers but **render in a
+  collapsible "suppressed" section** of the report output — the same
+  transparency contract as the review cluster's dropped-false-positives list.
+  Suppression hides nothing; it moves the item behind a fold with its receipt.
+- Suppressing a P0 does not restore a ready verdict — P0 caps the verdict
+  regardless (§ 2).
+
+## 6 — Consumer recipe (opt-in CI gating)
+
+1. Run the diagnostic (via `launch-readiness` § epistemics) and save the
+   report to `agents/evidence/prelaunch/baseline.json`. Commit it.
+2. Before a launch, produce a fresh `current.json` and run
+   `prelaunch_diagnostics.ts diff baseline.json current.json --ci` in the
+   pipeline — red on a new P0/P1 or a launch-gate Pass→Finding flip.
+3. After the launch settles, promote `current.json` to the new committed
+   baseline (a reviewed diff, like any lockfile bump).
+
+## 7 — Fix-loop discipline (Phase 3)
+
+- Rank open findings by severity, then by launch-gate membership.
+- Propose the **safest first approval batch** — explicitly the first *safe*
+  batch (reversible, small blast radius), not the full backlog. Approval is
+  the user's; the diagnostic never self-authorizes fixes
+  ([`autonomous-execution`](../../src/rules/autonomous-execution.md) and the
+  reversibility floor govern, unrestated).
+- Per-finding status vocabulary (schema-enforced): `open · fixed ·
+  accepted-risk · deferred-with-reason · suppressed-with-evidence ·
+  not-applicable` — everything except open/fixed carries a `reason`.
+- **Rescore whenever evidence changes** — a fix, a new probe, or an answered
+  question re-runs the affected area, never the memory of it.
+
+## 8 — Read-only guarantee
+
+The diagnostic **never mutates a consumer project**: the tooling reads
+reports and writes nothing outside its own output. Fixes flow through the
+normal approval-gated engines, not through this contract.
+
+## 9 — Score: council-locked (do not add)
+
+The pass-1 council **dropped the 0–100 readiness score** (gameable false
+precision). A severity-capped, never-published internal-ordering variant is
+**surfaced for council re-evaluation in the roadmap's revisit note — not
+adopted**. Until that revisit resolves: no numeric score field in the schema,
+no score in any rendered output, no score in CI messages. The ID scheme,
+epistemics, diff gate, and suppression above stand on their own.
+
+## See also
+
+- [`launch-readiness`](../../src/skills/launch-readiness/SKILL.md) — the skill
+  surface that produces the report.
+- [`docs/proof.md`](../proof.md) — the suite-side claims-ledger this mirrors.
+- `road-to-ecosystem-harvest-reliability-measurement` (archived) — measures the
+  *suite itself*; this contract measures the *consumer's launch surface*.

package/docs/contracts/release-pr-gating.md

@@ -81,6 +81,7 @@ the feature-PR floor by adding:
 | `release-validation.yml` (Phase B) | `changelog-entry` | CHANGELOG carries an entry matching the head-branch version |
 | `release-validation.yml` (Phase B) | `version-consistency` | `package.json` / `marketplace.json` / pack manifests agree on the version |
 | `consumer-matrix.yml` | `consumer-matrix` · `publish-dry-run` · `mcp-worker-dry-run` · `plugin-bootstrap` | pack-based consumer E2E + pre-tag dry-runs of the release-adjacent workflows — see the exemption note below |
+| (maintainer-local) | `task smoke-host-loadability REQUIRE=1` | real-host loadability — `claude plugin validate` + temp-home plugin install + metadata cross-consistency (marketplace ↔ plugin dirs ↔ docs). Optional in CI (runners lack the claude CLI, the step self-skips); **required before a release is cut** — `REQUIRE=1` turns a missing CLI into a failure |
 
 ## Consumer-matrix exemption — the tarball window
 

package/docs/decisions/ADR-049-configuration-trust-boundary.md

@@ -44,7 +44,9 @@ filtered** to a small curated set (`MERGEABLE_KEYS` = `name`, `ide`,
 ignored. `ai_council` is not on the whitelist, and council configuration lives
 project-local in `agents/settings/.ai-council.yml` (per
 [`ai-council-config`](../contracts/ai-council-config.md), the Phase-0 single
-source of truth).
+source of truth). <!-- Council placement superseded by ADR-104: the config is
+now user-global only (`~/.event4u/agent-config/settings/.ai-council.yml`),
+never project-local. The trust-boundary reasoning below is unaffected. -->
 
 The maintainer questioned whether that project-local placement is a bug. The AI
 council was asked to decide and **converged** that it is **not** a bug — the
@@ -72,6 +74,12 @@ personas, rule tier) **belongs in version control**, and onboarding stays clean
    for council config is `agents/settings/.ai-council.yml`
    ([`ai-council-config`](../contracts/ai-council-config.md)); personas live in
    the project surface. Neither is a user-global concern.
+   <!-- Superseded for the COUNCIL part by ADR-104: council config is now
+   user-global only, never project-local (personas are unaffected). -->
+   **Update (ADR-104):** the council half of this decision was reversed — the
+   council config now lives user-global only
+   (`~/.event4u/agent-config/settings/.ai-council.yml`) and the project tree is
+   never read. Personas stay project-local as stated.
 
 3. **Only secrets and the curated identity whitelist are global.** Provider API
    keys already live global in `~/.event4u/agent-config/<provider>.key` (0600);

package/docs/featured-skills.md

@@ -1,6 +1,6 @@
 # Featured Skills
 
-A curated subset of the 271 skills. Three tiers, one per primary audience. Full catalog lives in [`dist/agent-src/skills/`](../dist/agent-src/skills/); see also [`docs/catalog.md`](catalog.md) for the complete index and [`docs/featured-commands.md`](featured-commands.md) for command-level highlights.
+A curated subset of the 274 skills. Three tiers, one per primary audience. Full catalog lives in [`dist/agent-src/skills/`](../dist/agent-src/skills/); see also [`docs/catalog.md`](catalog.md) for the complete index and [`docs/featured-commands.md`](featured-commands.md) for command-level highlights.
 
 > **Eval-gated messaging note.** Until `task bench --corpus non-dev` reports `selection_accuracy >= 0.60` (see `road-to-product-adoption.md` Phase 1), this page is documentation, not marketing. The entries below are the candidates the corpus tests against; their description quality is what the eval validates.
 
@@ -92,4 +92,4 @@ task lint-featured-skills
 
 CI runs this in `taskfiles/ci-fast.yml`. Stale entries (renamed / removed skill or command) fail the build. See [`scripts/lint_featured_skills.py`](../src/scripts/lint_featured_skills.py) for the matcher.
 
-→ Browse all 271 skills: [`dist/agent-src/skills/`](../dist/agent-src/skills/) · all 178 commands: [`dist/agent-src/commands/`](../dist/agent-src/commands/)
+→ Browse all 274 skills: [`dist/agent-src/skills/`](../dist/agent-src/skills/) · all 183 commands: [`dist/agent-src/commands/`](../dist/agent-src/commands/)

package/docs/getting-started-by-role.md

@@ -2,7 +2,7 @@
 
 > Pick the entry that matches what you do day-to-day. Each section names the three skills you will reach for first and shows whether MCP (no terminal) or CLI (terminal) is the simpler install path for that role.
 
-`agent-config` ships 271 skills, 104 governed rules, and 178 commands. You do not need all of them. Each role below filters to the slice that pays back in week one; the rest stays available and shows up on demand when a task references it.
+`agent-config` ships 274 skills, 104 governed rules, and 183 commands. You do not need all of them. Each role below filters to the slice that pays back in week one; the rest stays available and shows up on demand when a task references it.
 
 > **Taglines at a glance →** [`docs/role-experiences.md`](role-experiences.md) — the one-line pitch per role experience, in one table.
 
@@ -73,6 +73,8 @@ Cinematic-blueprint approach: the agent expands your script into a 12-block scen
 - [`nextjs-patterns`](../dist/agent-src/skills/nextjs-patterns/SKILL.md) — App Router, Server Components, Server Actions, caching.
 - [`quality-tools`](../dist/agent-src/skills/quality-tools/SKILL.md) — PHPStan, Rector, ECS error triage and fix loop.
 
+Cross-model review of the real diff (**team mode**, default off) is documented in [Getting Started § Team mode](getting-started.md#team-mode--cross-model-review-default-off).
+
 **Install path:** **CLI.** Run `npx @event4u/agent-config init --tools=claude-code,cursor` in the project root. MCP works too but loses git / file-system tooling that the IDE-integrated path gives you.
 
 ---