@event4u/agent-config 9.8.0 → 9.10.0

package/src/agent-src/templates/scripts/telemetry/engagement.ts

@@ -21,6 +21,10 @@
  *              (same PII-exclusion-by-construction shape as consulted).
  *       "outcomes":  ["blocked", "verification_failed", ...]  # optional
  *       "tokens_estimate": {"consulted_load": <int>}   # optional
+ *       "cross_source": [{"id": "d1", "type": "text-image", "asked": true}]  # optional
+ *              (road-to-feedback-9.2.0-followups 1.4): per-task ask-rate facet
+ *              for the cross-source-consistency rule. Structural only — id +
+ *              a closed type enum + a bool, no free-form fields.
  *     }
  */
 import * as fs from 'node:fs';
@@ -55,6 +59,18 @@ export const ALLOWED_OUTCOMES = [
 ] as const;
 export const MAX_OUTCOMES_PER_EVENT = ALLOWED_OUTCOMES.length;
 
+// cross_source facet (road-to-feedback-9.2.0-followups 1.4) — the
+// cross-source-consistency rule's discrepancy taxonomy (a)-(d) from
+// src/rules/cross-source-consistency.md.
+export const ALLOWED_CROSS_SOURCE_TYPES = [
+    'text-image',
+    'silent-needed',
+    'spec-code',
+    'intra-ticket',
+] as const;
+export const MAX_CROSS_SOURCE_PER_EVENT = 32;
+const _CROSS_SOURCE_KEYS = ['asked', 'id', 'type'] as const;
+
 // Phase 5 redaction validator — keep id fields from leaking paths,
 // free-text, or filenames.
 const _FORBIDDEN_ID_CHARS = ['/', '\\', '\n', '\r', '\t'] as const;
@@ -125,6 +141,13 @@ export function check_id_redaction(label: string, value: unknown): void {
     }
 }
 
+/** One surfaced cross-source discrepancy — structural only, no free text. */
+export interface CrossSourceEntry {
+    id: string;
+    type: string;
+    asked: boolean;
+}
+
 export interface EngagementEventInit {
     ts: string;
     task_id: string;
@@ -134,6 +157,7 @@ export interface EngagementEventInit {
     loaded?: Record<string, string[]> | null;
     outcomes?: string[] | null;
     tokens_estimate?: Record<string, number> | null;
+    cross_source?: CrossSourceEntry[] | null;
     schema_version?: number;
 }
 
@@ -146,6 +170,7 @@ export class EngagementEvent {
     loaded: Record<string, string[]> | null;
     outcomes: string[] | null;
     tokens_estimate: Record<string, number> | null;
+    cross_source: CrossSourceEntry[] | null;
     schema_version: number;
 
     constructor(init: EngagementEventInit) {
@@ -157,6 +182,7 @@ export class EngagementEvent {
         this.loaded = init.loaded ?? null;
         this.outcomes = init.outcomes ?? null;
         this.tokens_estimate = init.tokens_estimate ?? null;
+        this.cross_source = init.cross_source ?? null;
         this.schema_version = init.schema_version ?? SCHEMA_VERSION;
     }
 
@@ -195,6 +221,9 @@ export class EngagementEvent {
         if (this.outcomes !== null) {
             _validate_outcomes(this.outcomes);
         }
+        if (this.cross_source !== null) {
+            _validate_cross_source(this.cross_source);
+        }
         if (this.tokens_estimate !== null) {
             if (!_isPlainObject(this.tokens_estimate)) {
                 throw new EngagementSchemaError(
@@ -236,6 +265,13 @@ export class EngagementEvent {
         if (this.outcomes && this.outcomes.length > 0) {
             out['outcomes'] = [...this.outcomes];
         }
+        if (this.cross_source && this.cross_source.length > 0) {
+            out['cross_source'] = this.cross_source.map((e) => ({
+                id: e.id,
+                type: e.type,
+                asked: e.asked,
+            }));
+        }
         if (this.tokens_estimate && Object.keys(this.tokens_estimate).length > 0) {
             out['tokens_estimate'] = { ...this.tokens_estimate };
         }
@@ -317,6 +353,46 @@ function _validate_outcomes(payload: unknown): void {
     }
 }
 
+function _validate_cross_source(payload: unknown): void {
+    if (!Array.isArray(payload)) {
+        throw new EngagementSchemaError('cross_source must be a list of dict or None');
+    }
+    if (payload.length > MAX_CROSS_SOURCE_PER_EVENT) {
+        throw new EngagementSchemaError(
+            `cross_source exceeds ${MAX_CROSS_SOURCE_PER_EVENT} entries`,
+        );
+    }
+    for (const entry of payload) {
+        if (!_isPlainObject(entry)) {
+            throw new EngagementSchemaError('cross_source entries must be dict[str, ...]');
+        }
+        const keys = Object.keys(entry);
+        const extra = keys.filter((k) => !(_CROSS_SOURCE_KEYS as readonly string[]).includes(k));
+        if (extra.length > 0) {
+            throw new EngagementSchemaError(
+                `cross_source entry has unexpected key(s): ${extra.sort().join(', ')}; `
+                + `allowed keys: ${_CROSS_SOURCE_KEYS.join(', ')}`,
+            );
+        }
+        for (const required of _CROSS_SOURCE_KEYS) {
+            if (!(required in entry)) {
+                throw new EngagementSchemaError(`cross_source entry missing required key '${required}'`);
+            }
+        }
+        check_id_redaction('cross_source.id', entry['id']);
+        const type_val = entry['type'];
+        if (typeof type_val !== 'string' || !(ALLOWED_CROSS_SOURCE_TYPES as readonly string[]).includes(type_val)) {
+            throw new EngagementSchemaError(
+                `cross_source.type must be one of ${_reprTuple(ALLOWED_CROSS_SOURCE_TYPES)}, `
+                + `got ${_reprScalar(type_val)}`,
+            );
+        }
+        if (typeof entry['asked'] !== 'boolean') {
+            throw new EngagementSchemaError('cross_source.asked must be a bool');
+        }
+    }
+}
+
 export function parse_event(line: string): EngagementEvent {
     if (typeof line !== 'string' || !line.trim()) {
         throw new EngagementSchemaError('line must be a non-empty JSONL record');
@@ -342,6 +418,7 @@ export function parse_event(line: string): EngagementEvent {
             : _orEmptyDict(r['loaded']),
         outcomes: (r['outcomes'] as string[] | null | undefined) ?? null,
         tokens_estimate: (r['tokens_estimate'] as Record<string, number> | null | undefined) ?? null,
+        cross_source: (r['cross_source'] as CrossSourceEntry[] | null | undefined) ?? null,
         schema_version: r['schema_version'] === undefined
             ? SCHEMA_VERSION
             : (r['schema_version'] as number),

package/src/agent-src/templates/scripts/telemetry_record.ts

@@ -21,6 +21,7 @@ import * as path from 'node:path';
 import { record_event } from './telemetry/boundary.js';
 import {
     ALLOWED_OUTCOMES,
+    type CrossSourceEntry,
     EngagementEvent,
     EngagementSchemaError,
     now_utc_iso,
@@ -58,6 +59,30 @@ function _reprStr(s: string): string {
     return `'${s}'`;
 }
 
+/** Turn `["d1:text-image:ask", ...]` into cross_source entries. Unknown
+ *  `type` values are left to schema validation (consistent with how
+ *  `_parse_kv_list` leaves unknown artefact kinds to the schema); only
+ *  the CLI-only `id:type:ask|warn` shape is checked here. */
+function _parse_cross_source_list(values: string[]): CrossSourceEntry[] {
+    const out: CrossSourceEntry[] = [];
+    for (const raw of values) {
+        const parts = raw.split(':');
+        if (parts.length !== 3 || !parts[0] || !parts[1] || !parts[2]) {
+            throw new SystemExitError(
+                `❌  --cross-source must be 'id:type:ask|warn', got ${_reprStr(raw)}`,
+            );
+        }
+        const [id, type, verdict] = parts as [string, string, string];
+        if (verdict !== 'ask' && verdict !== 'warn') {
+            throw new SystemExitError(
+                `❌  --cross-source verdict must be 'ask' or 'warn', got ${_reprStr(verdict)}`,
+            );
+        }
+        out.push({ id, type, asked: verdict === 'ask' });
+    }
+    return out;
+}
+
 interface RecordArgs {
     task_id: string;
     boundary: string;
@@ -65,6 +90,7 @@ interface RecordArgs {
     applied: string[] | null;
     loaded: string[] | null;
     outcome: string[] | null;
+    cross_source: string[] | null;
     ts: string;
     payload_file: string | null;
     stdin: boolean;
@@ -81,6 +107,9 @@ function _build_event_from_args(args: RecordArgs): EngagementEvent {
         applied: _parse_kv_list(args.applied ?? []),
         loaded: args.loaded && args.loaded.length > 0 ? _parse_kv_list(args.loaded) : null,
         outcomes: args.outcome && args.outcome.length > 0 ? [...args.outcome] : null,
+        cross_source: args.cross_source && args.cross_source.length > 0
+            ? _parse_cross_source_list(args.cross_source)
+            : null,
     });
 }
 
@@ -106,6 +135,7 @@ function _build_event_from_payload(raw: string): EngagementEvent {
         loaded: d['loaded'] === undefined || d['loaded'] === null ? null : _orEmpty(d['loaded']),
         outcomes: (d['outcomes'] as string[] | null | undefined) ?? null,
         tokens_estimate: (d['tokens_estimate'] as Record<string, number> | null | undefined) ?? null,
+        cross_source: (d['cross_source'] as CrossSourceEntry[] | null | undefined) ?? null,
     });
 }
 
@@ -126,6 +156,7 @@ function _parseArgs(argv: string[]): RecordArgs {
         applied: null,
         loaded: null,
         outcome: null,
+        cross_source: null,
         ts: '',
         payload_file: null,
         stdin: false,
@@ -170,7 +201,8 @@ function _parseArgs(argv: string[]): RecordArgs {
                 );
             }
             (a.outcome ??= []).push(v);
-        } else if (tok === '--ts') a.ts = next();
+        } else if (tok === '--cross-source') (a.cross_source ??= []).push(next());
+        else if (tok === '--ts') a.ts = next();
         else if (tok === '--payload-file') a.payload_file = next();
         else if (tok === '--stdin') a.stdin = true;
         else if (tok === '--settings') a.settings = next();

package/src/agent-src/templates/scripts/tier_usage_report.ts

@@ -234,7 +234,7 @@ function _within_window(ts_bucket: string, window_days: number | null): boolean
     return ms >= Date.now() - window_days * 24 * 60 * 60 * 1000;
 }
 
-type BucketKey = string; // `${tier}${command}` — composite (tier, command) key.
+type BucketKey = string; // `${tier}\0${command}` — composite (tier, command) key.
 interface RowStats {
     count: number;
     distinct_users: number;
@@ -274,7 +274,7 @@ function aggregate(log_path: string, window_days: number): AggregateResult {
         }
         kept += 1;
         const tier = Math.trunc(rec.tier);
-        const key = `${tier}${rec.command}`;
+        const key = `${tier}\0${rec.command}`;
         if (!keyMeta.has(key)) {
             keyMeta.set(key, { tier, command: rec.command });
             counts.set(key, 0);

package/src/agent-src/templates/scripts/work_engine/directives/ui/polish.ts

@@ -336,7 +336,7 @@ function _classify_token_violations(
     tokens: Record<string, Any>,
 ): [Record<string, Any>[], Record<string, Any>[]] {
     const matched: Record<string, Any>[] = [];
-    // Insertion-ordered map keyed by `${category}${value}` to mirror the
+    // Insertion-ordered map keyed by `${category}\0${value}` to mirror the
     // Python `dict[tuple[str, str], int]` insertion order.
     const unmatched_counts = new Map<string, { category: string; value: string; count: number }>();
     for (const finding of findings) {
@@ -356,7 +356,7 @@ function _classify_token_violations(
             matched.push(finding);
             continue;
         }
-        const key = `${category}${value}`;
+        const key = `${category}\0${value}`;
         const existing = unmatched_counts.get(key);
         if (existing) {
             existing.count += 1;

package/src/agent-src/templates/scripts/work_engine/directives/ui/review.ts

@@ -294,7 +294,7 @@ function _filter_known(violations: Any[], known: Any[]): Any[] {
  * scalars, mirroring Python tuple-equality in a JS `Set`.
  */
 function _keyOf(rule: Any, selector: Any): string {
-    return `${_tag(rule)}${_tag(selector)}`;
+    return `${_tag(rule)}\0${_tag(selector)}`;
 }
 
 function _tag(v: Any): string {

package/src/config/agent-settings.template.yml

@@ -40,18 +40,25 @@ profile:
 # trees (.claude/ .cursor/ .windsurf/) or only the active profile + packs'
 # artefacts. See docs/decisions/ADR-040 and docs/contracts/capability-packs.md.
 #
-# legacy-all = (DEFAULT, non-breaking) project the full surface, exactly as
-#              5.x did. An `npm update` to 6.0.0 changes nothing here.
-# scoped     = project only the active profile's `packs` unioned with the
-#              runtime.active_packs overlay, expanded over the `requires`
-#              graph. Opt in with `agent-config use --profile=<id>`.
+# scoped     = (DEFAULT for new installs — road-to-credible-install Phase 2,
+#              2026-07-27) project only the active profile's `packs` unioned
+#              with the runtime.active_packs overlay, expanded over the
+#              `requires` graph. With the default `profile.id: developer`
+#              this ships the engineering surface only; the full catalog is
+#              one explicit opt-in away (`legacy-all` below, or
+#              `agent-config use --profile=<id>` / packs overlay).
+# legacy-all = project the full surface, exactly as 5.x did. EXISTING
+#              installs keep their recorded setting on upgrade (this file is
+#              never rewritten by `npm update`); a missing key also still
+#              means legacy-all, so only fresh installs get the scoped
+#              default. See MIGRATION.md for the flip note.
 #
-# Scoping is opt-in by THIS key alone — never inferred from profile.id (every
-# install has `developer` by default). A failed scoped projection restores the
-# full tree. The committed plugin marketplace (.claude-plugin/) and the Augment
-# tree (.augment/) always project the full set in 6.0.0.
+# Scoping is controlled by THIS key alone — never inferred from profile.id.
+# A failed scoped projection restores the full tree. The committed plugin
+# marketplace (.claude-plugin/) and the Augment tree (.augment/) always
+# project the full set.
 projection:
-  mode: legacy-all
+  mode: scoped
   # --- Consumer-scoped RULE projection (road-to-request-scoped-rule-load P1,
   #     opt-in) ---
   # Workspace scope for the RULE layer only (independent of `mode` above,
@@ -138,6 +145,36 @@ rule_loading_tier: __RULE_LOADING_TIER__
 lean_projection:
   mode: eager-all
 
+# --- Telegraph condensation (output-side) ---
+#
+# speak_scope: off | prose_only | aggressive   (QUOTE the value — bare `off` is a
+#                                              YAML 1.1 boolean, not the string)
+#
+# `off` is the shipped default per docs/adrs/telegraph/0001-default-off-until-bench.md
+# (Status: accepted) and docs/contracts/condensation-default-kill-criterion.md.
+# Reason it is off: the only output-side bench (telegraph-v1, n=10, 2026-05-16)
+# measured median vs_terse -9.27% (API counts) / -5.47% (exact cl100k_base
+# re-analysis) — telegraph emits MORE tokens than a plain "be terse" instruction.
+# The kill-criterion table therefore lands on "criterion not met — defer" and the
+# telemetry multiplier stays suspended (0.9155 < 1.0).
+#
+# NOTE: `speak_scope: off` disables the BEHAVIOUR but not the COST — the projector
+# gates this rule on `telegraph.speak`, never on speak_scope, so under
+# `lean_projection.mode: eager-all` the ~982-token body still ships every session.
+# Set `speak: false` for zero-cost dormancy.
+#
+# Stated explicitly here because until 2026-07-29 this default existed only in
+# prose — and two prose sources disagreed (the rule said prose_only, the accepted
+# ADR said off). Do not flip without a passing output-side bench.
+telegraph:
+  # speak: false = DORMANT (the shipped default per docs/adrs/telegraph/0002).
+  # The rule is omitted from dist/router.json entirely, so its ~982-token body
+  # never ships. This — not speak_scope — is the lever that stops the cost:
+  # compile_router gates on `speak` and never reads speak_scope.
+  # Set true only after an output-side bench clears the kill-criterion bar.
+  speak: false
+  speak_scope: "off"
+
 # --- Token-rich skills governance ---
 #
 # Skills marked `token_budget_class: rich` in their frontmatter are exempt
@@ -214,6 +251,14 @@ personal:
   # false = silently investigate, only report the conclusion
   play_by_play: false
 
+  # Session canary — the name the agent addresses you with at the start of
+  # every new task (e.g. "Alex"). Like a canary in a coal mine: when the
+  # greeting silently disappears, the context window is degrading and it is
+  # time for a fresh conversation. Also keeps the reply-close markers alive
+  # (ONE end-summary; PR URL as the literal last line).
+  # Empty = feature off. See rules/session-canary.md.
+  canary_name: ""
+
   # Prefix PR comment replies with a bot icon 🤖 (true, false)
   # Personal preference — each developer decides for themselves.
   pr_comment_bot_icon: false
@@ -627,7 +672,7 @@ subagents:
   # delegate frontier-priced or speculative work. `resolveShippedDefault()`
   # remains the telemetry-driven DEMOTION gate: a measured regression demotes
   # back to `ask`.
-  auto: on
+  auto: "on"
 
   # Route delegable sub-tasks to the lowest-capable model tier (cost + speed
   # via model downshift). true (default) = downshift per declared model_tier;
@@ -740,9 +785,12 @@ ai_team:
 
   # Per-day cap on team calls. Team calls count into the EXISTING
   # cli_call_budget openai bucket (~/.event4u/agent-config/cli-calls.json,
-  # daily UTC reset) — one subscription, one counter, never a parallel
-  # count. This key is the team-side ceiling read against that shared
-  # bucket. 0 = block all team calls; raise if you review heavily.
+  # daily UTC reset) — the counter tracks MACHINE-WIDE spend across ALL
+  # subscriptions/profiles on this machine (an `agent-switch` profile
+  # switch does NOT get its own counter; the bucket is deliberately one
+  # per machine, never parallel counts). This key is the team-side
+  # ceiling read against that shared bucket. 0 = block all team calls;
+  # raise if you review heavily.
   max_calls_per_day: 50
 
   # Suppress the one-line wizard/init recommendation to set up the codex
@@ -915,6 +963,17 @@ memory:
   #   on  = inject the compact index (spotlighted as DATA).
   session_index: "off"
 
+  # session_end learning-sidecar aggregation (road-to-reachable-code-memory
+  # Phase 4). When true, the session_end hook aggregates
+  # agents/memory/intake/*.jsonl through the learning sidecar (30-day decay,
+  # ≥2-origin corroboration) into the gitignored .agent-learning.json +
+  # LESSONS.md and emits the one-line memory-visibility marker. Local-only,
+  # budget-capped (2 s), fail-open; promotion into curated memory stays
+  # human via /memory:propose. SHIPS OFF (council 2026-07-27): the default
+  # flip is proposed only after the 30-day dogfood shows non-trivial signal
+  # AND session-end p95 < 2 s.
+  learn_on_session_end: false
+
 # --- Knowledge cards: global cross-project sharing (ADR-100) ---
 #
 # Structure-grounding v2. Promote *expensive* (remote) knowledge cards from
@@ -1046,7 +1105,16 @@ hooks:
   # the agent is about to Grep/Glob or Read a source file: query the graph
   # first for "who calls / where used / impact" questions (or rebuild if the
   # index is stale, or build one if absent). Source G's strict block-first-read
-  # mode is deliberately un-ported. Build the cache with `code_graph build`.
+  # mode is deliberately un-ported. Build the cache with
+  # `agent-config code-graph build`.
+  #
+  # DEPRECATED (honest null, 2026-07-28): the pre-registered 2-arm retrieval
+  # benchmark measured the native graph at recall 0.365 vs grep 0.797 on
+  # graph-shaped questions across 3 real repos (internal/bench/reports/
+  # code-graph-vs-grep.md; claim code-graph-retrieval-null). `enabled: false`
+  # is permanent; deprecation notice lands at the next major, removal the
+  # major after — unless external evidence (a consumer case the graph answers
+  # and grep cannot) appears first.
   code_graph:
     enabled: false
 
@@ -1076,7 +1144,7 @@ decision_engine:
   #   low    = refuse to advance when band is below low
   #   medium = refuse when below medium
   #   high   = refuse when below high
-  min_confidence: off
+  min_confidence: "off"
 
   # Risk-class ceiling for Phase=Implement.
   #   off    = no ceiling (default)

package/src/config/ci-local-parity.yml

@@ -0,0 +1,142 @@
+# CI ↔ local parity manifest — the input to `check_ci_local_parity.ts`.
+#
+# WHY THIS EXISTS
+# ---------------
+# The local aggregate (`task ci`) and the CI workflows drifted apart in BOTH
+# directions, and each direction cost real defects:
+#
+#   CI-only  — a gate CI runs that no local chain reaches. The contributor pushes,
+#              CI goes red, and the fix costs a round trip. Measured 2026-07-30:
+#              check_portability and check_adr_frontmatter turned a PR red that a
+#              local run would have caught, twice in one session.
+#
+#   local-only — a gate `task ci` runs that no workflow does. Nothing enforces it
+#              remotely, so a defect merges. Measured: check-index had verified the
+#              artefact index for a long time and ran in no workflow, which is how
+#              16 stale rows in agents/index.md + docs/catalog.md reached main.
+#              check_bridge_derivation is the same shape.
+#
+# A gate that exists on one side only drifts silently on the other. This manifest
+# plus the checker makes each direction a declared, reviewable choice instead of an
+# accident.
+#
+# THE CONTRACT
+# ------------
+# The checker DERIVES both sides — it never trusts a hand-maintained list of gates:
+#
+#   CI side     parsed from .github/workflows/*.yml — every `./scripts-run
+#               src/scripts/<name>`, `npx tsx src/scripts/<name>`, `node <path>` and
+#               `task <name>` invocation.
+#   local side  the transitive closure of the `task ci` / `task consistency` chains
+#               plus the pre-push hook, resolved through taskfiles/*.yml.
+#
+# Anything present on one side and absent on the other must appear below with a
+# reason, or the checker fails. That is the whole mechanism: drift is allowed, but
+# only on the record.
+#
+# Parity is by gate ID, not by argv. Three gates run locally in a cheap variant and
+# in CI in a strict one (check_release_published bare vs --strict --check-npm,
+# build_cloud_bundle --check vs --all --clean, build_linear_digest offline vs live).
+# They need no declaration: the parity question is "can a contributor run this before
+# pushing at all", and the answer is yes. Declaring them anyway produced exactly the
+# dead exemption the stale-declaration check exists to catch — it was removed rather
+# than kept, because an exemption nothing needs is one the next reader trusts.
+#
+# WHY NOT JUST "RUN task ci BEFORE EVERY PUSH"
+# --------------------------------------------
+# Because a pre-push gate that is red for a reason the contributor did not cause
+# either blocks every push or trains people to ignore it, and the second is worse
+# than no gate. Two measured bounds from the same sweep:
+#
+#   bucket A (repo-content gates) sum to ~14.5s  → affordable before a push
+#   check_enforcement_coverage alone is 30.7s    → belongs in the post-CI loop
+#
+# So `pre_push_budget_seconds` below is a real budget, not a wish. A gate declared
+# `local_only` or `both` that exceeds it is reported, not silently added to the hook.
+
+# Wall-clock ceiling for the pre-push gate set, in seconds. Derived from the
+# measured bucket-A sum (14.5s) plus headroom; the existing hook already costs
+# `task consistency` + eslint + tsc, so this is a bounded increase, not a new tax.
+pre_push_budget_seconds: 25
+
+# Gates CI runs that no local chain reaches — each needs a reason it cannot or
+# should not run locally. `reason` is free text; `class` is one of:
+#   network        needs a token, a registry, or a live API
+#   release        verifies tag / release / published-package state
+#   artifact       needs something only CI builds (site, tarball, bundle)
+#   benchmark      it is a measurement, timing-sensitive or paid
+#   matrix         needs an OS or Node version this machine is not
+ci_only:
+  - id: adoption_snapshot
+    class: network
+    reason: "live api.npmjs.org + api.github.com calls, reads GITHUB_TOKEN"
+  - id: sync_github_metadata
+    class: network
+    reason: "hard-fails without GITHUB_TOKEN; mutates repo topics/description via the API"
+  - id: check_release_pr_shape
+    class: release
+    reason: "shells out to `gh pr diff <n>` — needs a live PR and a token"
+  - id: check_release_adjacent_health
+    class: release
+    reason: "queries GitHub Actions run history via gh"
+  - id: bench_drift_check
+    class: benchmark
+    reason: "consumes rolling bench reports that only a bench run produces"
+  - id: bench_baseline_ready
+    class: benchmark
+    reason: "same — no local baseline to compare against"
+  - id: bench_hook_latency
+    class: benchmark
+    reason: "a latency measurement; 24.0s measured and timing-sensitive on a loaded machine"
+  - id: check_site_links
+    class: artifact
+    reason: "needs the built Starlight site; exits 2 unbuilt or stale (registered in gate-coverage.yml with unavailable_exit: 2)"
+  - id: pack_mcp_content
+    class: artifact
+    reason: "writes the tracked internal/workers/mcp/content.json deploy artifact — a local run dirties the tree"
+  - id: consumer_matrix
+    class: matrix
+    reason: "consumes the npm pack tarball across a Node version matrix"
+  - id: cross_model_smoke
+    class: network
+    reason: "paid cross-model canary calls, secret-gated"
+  - id: trigger_eval_rotation
+    class: network
+    reason: "paid model calls"
+  - id: self_review_gate
+    class: network
+    reason: "--dry-run in the PR job; the live variant is a paid model call on the PR diff"
+  - id: bench_run
+    class: benchmark
+    reason: "paid model calls against a corpus"
+  - id: bench_runner
+    class: benchmark
+    reason: "the runner behind bench_run — same paid calls"
+  - id: skill_trigger_eval
+    class: benchmark
+    reason: "paid cross-model trigger-activation eval; the offline structural half is check_trigger_evals, which IS local"
+  - id: evaluator_umbrella
+    class: artifact
+    reason: "shell umbrella over CI-only evaluator fixtures"
+  - id: smoke_host_loadability
+    class: artifact
+    reason: "temp-HOME install + `claude plugin validate`; measured >100s, did not finish inside a 2-minute budget"
+  - id: install
+    class: matrix
+    reason: "the installer end-to-end against a scratch HOME on the OS matrix; a local run would rewrite this machine's own agent config"
+  - id: release
+    class: release
+    reason: "tags, bumps, publishes, opens the release PR — the release itself"
+
+# Gates a local chain runs that no workflow does — each needs a reason it is not
+# worth remote enforcement, OR it should simply be added to a workflow. Prefer
+# adding it: a local-only gate is the direction that let real defects merge.
+local_only:
+  - id: check_knowledge_sharing
+    class: staging-scoped
+    reason: >
+      Inspects the STAGED set (a gitignored intake file staged anyway via `git add -f`),
+      and CI has no staged set — the commit already happened by the time a workflow
+      runs. The committed-tree equivalent ("a file that should be gitignored is
+      tracked") is a different invariant and would need its own mode, so this is not
+      the local-only-by-neglect case the manifest warns about.

package/src/config/discovery/packs.yml

@@ -133,6 +133,28 @@
   domain: engineering
   size_class: medium
 
+- id: scale-discipline
+  label: Scale Discipline
+  description: Scale-safe persistence — index parity, bounded reads, migration safety, growth budgets, thin request path, durable async; deterministic pattern lints gate, heuristics advise.
+  workspaces: [engineering]
+  requires: [engineering-base]
+  trust_level_default: experimental
+  surface_tier: lab
+  domain: engineering
+  size_class: small
+  default_install: false
+
+- id: history-discipline
+  label: History Discipline
+  description: Audit/change history done right — audit coverage on declared scopes, cheapest-sufficient tier selection (default row-level audit log; event sourcing only by waiver), audit-table hygiene, privacy + reliability interlocks.
+  workspaces: [engineering]
+  requires: [scale-discipline]
+  trust_level_default: experimental
+  surface_tier: lab
+  domain: engineering
+  size_class: small
+  default_install: false
+
 - id: python
   label: Python
   description: Python-language patterns.

package/src/config/discovery/workspaces.yml

@@ -13,7 +13,7 @@
   description: Code, tests, CI, reviews, architecture.
   example_roles: [Developer, CTO, Tech Lead]
   default_packs: [engineering-base, git]
-  optional_packs: [php, laravel, symfony, javascript, typescript, react, nextjs, python, frontend-design, brand, analysis-workbench]
+  optional_packs: [php, laravel, symfony, javascript, typescript, react, nextjs, python, frontend-design, brand, analysis-workbench, scale-discipline, history-discipline]
 
 - id: product
   label: Product

package/src/config/evaluator-budgets.json

@@ -0,0 +1,47 @@
+{
+    "_comment": "Outside-in evaluator budgets (road-to-credible-install Phase 6). VALUES ARE FROZEN FROM MEASURED ACCEPTANCE NUMBERS of Phases 0-3 (2026-07-27, packed tarball @ 9.8.0 tree, --omit=dev consumer install), never invented. The umbrella harness (src/scripts/evaluator_umbrella.sh) measures the packed artifact and check_evaluator_budgets.ts enforces: absolute budget AND >regression_pct growth vs last_measured fails even under budget. Owner + review date are lint-enforced (lint_budget_ownership). Timing metrics carry a per-entry regression_pct override (200) — wall-clock noise would flap a 10% creep rule; their absolute max is the binding gate. Sizes/counts keep the tight 10% creep.",
+    "schema_version": 1,
+    "owner": "maintainer",
+    "review_by": "2027-07-27",
+    "regression_pct": 10,
+    "budgets": {
+        "unpacked_size_mb": {
+            "max": 28,
+            "last_measured": 26.05,
+            "method": "npm pack --json → dist.unpackedSize / 1e6 — includes the Phase-1 precompiled bundles (dist/hooks/dispatch.js + dist/mcp/server.mjs, ~2.1 MB) that bought the 23x hook-latency win and the tsx removal"
+        },
+        "node_modules_mb": {
+            "max": 140,
+            "last_measured": 127,
+            "method": "du -sm node_modules after `npm install --omit=dev <tarball>` in a clean project"
+        },
+        "runtime_dep_count": {
+            "max": 200,
+            "last_measured": 187,
+            "method": "package-lock.json packages count (minus root) of the clean --omit=dev install"
+        },
+        "cli_version_cold_ms": {
+            "max": 1500,
+            "last_measured": 370,
+            "method": "wall-clock of the FIRST `agent-config --version` after install (cold fs cache run counts)",
+            "regression_pct": 200
+        },
+        "mcp_boot_to_initialize_ms": {
+            "max": 2000,
+            "last_measured": 264,
+            "method": "initialize request piped to `node dist/mcp/server.mjs`, wall-clock to response",
+            "regression_pct": 200
+        },
+        "cli_help_command_count": {
+            "max": 80,
+            "last_measured": 80,
+            "method": "registry enumeration — count of `{ name: '...', disposition` entries in src/cli/registry.ts (NOT --help prose parsing; methods historically disagreed 74 vs 76 vs 79)",
+            "baseline_note": "79 → 80 on 2026-07-29 (9.9.0). This RECORDS growth that already shipped through its own reviewed PRs; it does not authorize new surface. Net +6 registry entries since the 9.8.0 tag (7 added — code-graph, knowledge, analytics, memory:get, memory:learn, rtk:detect, dispatch:hook; 1 removed — the old dispatch:hook entry), and the frozen 79 was never re-measured after them. The gate did its job: it made the drift visible at release time rather than letting it accrue silently. Next addition breaches again by design."
+        },
+        "mcp_public_tool_count": {
+            "max": 19,
+            "last_measured": 19,
+            "method": "tools/list length on the stdio server (post-ADR-132: implemented tools only)"
+        }
+    }
+}