@akagilnc/pi-workflow-roles 0.1.5406 → 0.1.5475

package/dist/public-cli/terminal.js

@@ -1,3 +1,4 @@
+import { serializeThrownValue } from "../serialize-thrown-value.js";
 /** Encode one free-text Terminal cell. JSON string form cannot embed raw tab/newline. */
 export function encodeTerminalField(value) {
     return JSON.stringify(value);
@@ -10,36 +11,15 @@ export function exitCodeForTerminalOutcome(outcome) {
     return isLawfulTypedTerminalOutcome(outcome) ? 0 : 1;
 }
 /**
- * One admitted Role run's typed Terminal aggregate.
- * Resumable failures carry `resume` and must not re-disclose the run ID via
- * top-level `runId` or public artifact path components — only `resume.command`.
- * #416: autoResumeCount (0..2) is a call-local observation of how many in-place
- * auto-resumes occurred during this single LLM call; it is not persisted to
- * run-state.json and does not participate in limit decisions.
+ * Parent-facing current reply: a court's own accepted/audit payloads only.
+ * Run-scoped history stays on TerminalResult.submissions (#879 / #836), and a
+ * failure face never answers as this court's reply (#953).
  */
-/**
- * Current-result payloads on a terminal — accepted/audit role result, or a
- * rare intentional failure face. Run history is TerminalResult.submissions
- * (#836 / #953), not this helper.
- */
-export function roleResultPayloads(outcome) {
-    if (outcome.kind === "accepted" || outcome.kind === "audit_escalation")
-        return outcome.payloads ?? [];
-    if (outcome.kind === "failure")
-        return outcome.payloads ?? [];
-    return [];
-}
-/**
- * Prefer non-empty current-result payloads; an empty array must not shadow
- * top-level submissions (#953 / #836). Callers that need failure history must
- * read TerminalResult.submissions directly — do not coalesce into failure.payloads.
- */
-export function coalesceSubmissionRows(payloads, submissions) {
-    if (payloads !== undefined && payloads.length > 0)
-        return payloads;
-    if (submissions !== undefined && submissions.length > 0)
-        return submissions;
-    return payloads ?? submissions ?? [];
+export function currentReplyRows(terminal) {
+    const outcome = terminal?.roleOutcome;
+    return outcome?.kind === "accepted" || outcome?.kind === "audit_escalation"
+        ? outcome.payloads ?? []
+        : [];
 }
 /**
  * Build an advice navigator fact (#959). Prose is presented as submitted;
@@ -87,7 +67,7 @@ export function formatTerminalResult(result) {
         for (const [key, value] of Object.entries(facts)) {
             if (value === undefined)
                 continue;
-            const rendered = typeof value === "string" ? value : JSON.stringify(value);
+            const rendered = typeof value === "string" ? value : serializeThrownValue(value);
             lines.push(`fact\t${encodeTerminalField(key)}\t${encodeTerminalField(rendered)}`);
         }
     }
@@ -126,11 +106,7 @@ export function formatTerminalResult(result) {
             }
         }
     }
-    if (result.resume !== undefined) {
-        // Resumable failure: run ID is revealed only inside the complete resume command.
-        lines.push(`resume\t${encodeTerminalField(result.resume.command)}`);
-    }
-    else if (result.runId !== undefined) {
+    if (result.runId !== undefined) {
         lines.push(`run\t${encodeTerminalField(result.runId)}`);
     }
     if (result.autoResumeCount !== undefined) {
@@ -138,25 +114,19 @@ export function formatTerminalResult(result) {
     }
     // Role-result block: original payloads, newest first for humans (#961).
     // Typed payloads/submissions stay ledger order; only this presentation reverses.
-    // #953: failure history is the top-level submissions carrier (#836) — present
-    // as recorded-submission, never as this-turn submission/receipt.
-    if (result.roleOutcome.kind === "failure") {
-        const recorded = result.submissions ?? [];
-        for (let i = recorded.length - 1; i >= 0; i -= 1) {
-            const payload = recorded[i];
-            const rendered = typeof payload === "string" ? payload : JSON.stringify(payload);
-            lines.push(`recorded-submission\t${encodeTerminalField(rendered)}`);
-        }
-    }
-    else {
-        const payloads = result.roleOutcome.kind === "accepted" ||
-            result.roleOutcome.kind === "audit_escalation"
-            ? coalesceSubmissionRows(result.submissions, result.roleOutcome.payloads)
-            : result.submissions ?? [];
-        for (let i = payloads.length - 1; i >= 0; i -= 1) {
-            const payload = payloads[i];
+    // Present the current result and the run history on distinct faces (#879 / #836).
+    // Neither a prior court's submission nor a failed turn's history is this reply.
+    const current = result.roleOutcome.kind === "failure"
+        ? result.roleOutcome.payloads ?? []
+        : currentReplyRows(result);
+    for (const [label, rows] of [
+        ["submission", current],
+        ["recorded-submission", result.submissions ?? []],
+    ]) {
+        for (let i = rows.length - 1; i >= 0; i -= 1) {
+            const payload = rows[i];
             const rendered = typeof payload === "string" ? payload : JSON.stringify(payload);
-            lines.push(`submission\t${encodeTerminalField(rendered)}`);
+            lines.push(`${label}\t${encodeTerminalField(rendered)}`);
         }
     }
     return `${lines.join("\n")}\n`;

package/dist/public-role-summons.js

@@ -4,6 +4,7 @@ import { realpath } from "node:fs/promises";
 import { join } from "node:path";
 import { promisify } from "node:util";
 import { isRecord, errorText } from "./unknown-value.js";
+import { serializeThrownValue } from "./serialize-thrown-value.js";
 const AK_ROLE_PACKAGE_ROOT_ENV = "AK_ROLE_PACKAGE_ROOT";
 const execFileAsync = promisify(execFile);
 function createCapturingIo() {
@@ -288,15 +289,9 @@ async function resolveReviewerGitToplevel(projectRoot) {
   )).stdout.trim());
 }
 async function summonParallelReviewerLenses(options) {
-  const describeFailure = (error) => {
-    if (error instanceof AggregateError) {
-      return [error.message, ...error.errors.map(describeFailure)].join("\n");
-    }
-    return errorText(error);
-  };
   const failedResult = (error) => ({
     exitCode: 1,
-    stderr: describeFailure(error)
+    stderr: serializeThrownValue(error)
   });
   const dualFailure = (error) => {
     const failure = failedResult(error);
@@ -399,7 +394,7 @@ async function summonParallelReviewerLenses(options) {
     }
   } catch (error) {
     sealDiagnostic = `Reviewer target final seal failed:
-${describeFailure(error)}`;
+${serializeThrownValue(error)}`;
   }
   if (sealDiagnostic !== void 0) {
     const withSealFailure = (result) => ({

package/dist/receipt-delivery-policy.js

@@ -30,18 +30,13 @@ function parseNoReceiptLifecycleFacts(input) {
   if (!isRecord(input) || typeof input.terminalToolCalled !== "boolean" || input.sessionCompletion !== "settled-without-accepted-receipt" || input.acceptedReceipt !== false || typeof input.runPointer !== "string" || input.runPointer.trim() === "" || typeof input.attemptPointer !== "string" || input.attemptPointer.trim() === "" || !Array.isArray(input.rejectedReceipts) || !input.rejectedReceipts.every((item) => isRecord(item) && typeof item.reason === "string")) {
     throw new TypeError("malformed no-receipt lifecycle facts");
   }
-  return {
+  return noReceiptLifecycleFacts({
     terminalToolCalled: input.terminalToolCalled,
-    rejectedReceipts: input.rejectedReceipts.map((item) => ({
-      reason: item.reason,
-      diagnosticAvailable: item.reason.trim() !== ""
-    })),
+    rejectedReceipts: input.rejectedReceipts,
     deliveryTurns: deliveryTurnsFact(input.deliveryTurns),
-    sessionCompletion: "settled-without-accepted-receipt",
     runPointer: input.runPointer,
-    attemptPointer: input.attemptPointer,
-    acceptedReceipt: false
-  };
+    attemptPointer: input.attemptPointer
+  });
 }
 function noReceiptLifecycleFacts(input) {
   return {
@@ -106,9 +101,13 @@ function createReceiptDeliveryPolicy(limit) {
      * already sent, and adding the two counts would bill one send twice.
      */
     recordNestedNoReceipt(facts) {
-      terminalToolCalled = terminalToolCalled || facts.terminalToolCalled;
-      rejectedReceipts.push(...facts.rejectedReceipts);
-      deliveryTurns = Math.max(deliveryTurns, facts.deliveryTurns);
+      if (typeof facts.terminalToolCalled === "boolean") {
+        terminalToolCalled = terminalToolCalled || facts.terminalToolCalled;
+      }
+      if (facts.rejectedReceipts !== void 0) {
+        rejectedReceipts.push(...facts.rejectedReceipts);
+      }
+      if (facts.deliveryTurns !== void 0) deliveryTurns = Math.max(deliveryTurns, facts.deliveryTurns);
       closed = true;
     },
     nextAction() {

package/dist/run-terminal-artifacts.js

@@ -1,18 +1,23 @@
 /**
- * Canonical reader for run-directory typed terminal artifacts.
- * Layout owner is settlement publish*Artifacts (report.json / error.json /
- * audit-incomplete.json under artifacts/, plus the same publisher's durable
- * failure fallbacks). This module only reads presence and structural
+ * Canonical layout and reader for run-directory typed terminal artifacts.
+ * Settlement publish*Artifacts uses these names (report.json / error.json /
+ * audit-incomplete.json under artifacts/, plus durable failure fallbacks).
+ * The reader only checks presence and structural
  * readability — it does not re-derive role outcomes or invent a second
  * candidate algorithm.
  */
+import { randomUUID } from "node:crypto";
 import { readdir, readFile } from "node:fs/promises";
 import { basename, dirname, join } from "node:path";
 import { roleRunArtifactsDirectory } from "./role-run-placement.js";
 import { isRecord, errorText, isMissingPathError } from "./unknown-value.js";
+export const RUN_TERMINAL_REPORT_FILE = "report.json";
+export const RUN_TERMINAL_ERROR_FILE = "error.json";
+export const RUN_TERMINAL_EVIDENCE_FILE = "evidence.json";
+export const RUN_TERMINAL_ERROR_SETTLEMENT_FILE = "error.settlement.json";
 export const RUN_TERMINAL_ARTIFACT_FILES = [
-    "report.json",
-    "error.json",
+    RUN_TERMINAL_REPORT_FILE,
+    RUN_TERMINAL_ERROR_FILE,
     "audit-incomplete.json",
 ];
 /**
@@ -22,11 +27,15 @@ export const RUN_TERMINAL_ARTIFACT_FILES = [
  * Relative to the run directory.
  */
 export const RUN_TERMINAL_ERROR_FALLBACK_RELATIVE_PATHS = [
-    "artifacts/error.settlement.json",
-    "error.settlement.json",
+    `artifacts/${RUN_TERMINAL_ERROR_SETTLEMENT_FILE}`,
+    RUN_TERMINAL_ERROR_SETTLEMENT_FILE,
 ];
-/** Unique open-ended failure names: error.<uuid>.json (publisher stem + uuid). */
-const UNIQUE_ERROR_FALLBACK_NAME = /^error\.[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}\.json$/i;
+/** Publisher and reader share the open-ended error face stem. */
+export const UNIQUE_ERROR_FALLBACK_STEM = "error";
+export function uniqueErrorFallbackName() {
+    return `${UNIQUE_ERROR_FALLBACK_STEM}.${randomUUID()}.json`;
+}
+const UNIQUE_ERROR_FALLBACK_NAME = new RegExp(`^${UNIQUE_ERROR_FALLBACK_STEM}\\.[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}\\.json$`, "i");
 /**
  * Minimum producer-owned face shared by settlement terminal artifacts
  * (report / error / audit-incomplete). Consumer-driven: enough to identify a
@@ -219,7 +228,3 @@ export async function readRunTerminalArtifact(runDirectory) {
     }
     return { status: "absent" };
 }
-/** Test/helper: basename face of a unique fallback path, if any. */
-export function isUniqueErrorFallbackName(name) {
-    return UNIQUE_ERROR_FALLBACK_NAME.test(basename(name));
-}

package/dist/serialize-thrown-value.js

@@ -1,5 +1,11 @@
 import { inspect } from "node:util";
 /** Record the whole thrown value without invoking its getters or custom formatter. */
 export function serializeThrownValue(value) {
-    return inspect(value, { depth: 10, showHidden: true, customInspect: false });
+    return inspect(value, {
+        depth: null,
+        maxArrayLength: null,
+        maxStringLength: null,
+        showHidden: true,
+        customInspect: false,
+    });
 }

package/dist/submission-ledger.js

@@ -79,11 +79,6 @@ function resolveReadScope(homeOrScope) {
         return { home: homeOrScope };
     return homeOrScope;
 }
-function recordsForAttempt(owned, attemptId) {
-    // #836: courtAttempt is a recording tag for new courts, not a visibility gate.
-    void attemptId;
-    return owned;
-}
 /** Recording tag a row was written under — subject first, historical payload fallback. */
 function recordAttemptId(record) {
     if (typeof record.subject === "object" && record.subject !== null) {
@@ -98,30 +93,6 @@ function recordAttemptId(record) {
     }
     return undefined;
 }
-/**
- * True when the given attemptId itself already produced a sealed or
- * audit-escalation submission (#836 r12 class 2). This is a separate
- * freshness signal, not a presentation filter — `recordsForAttempt` above
- * stays a pass-through so every original payload keeps presenting honestly
- * (#836: attemptId is a recording tag, not a visibility gate). Settlement
- * consumes this only to stop a prior attempt's stale acceptance from
- * outranking the current attempt's own real host-turn failure signal
- * (#637 original intent, restored narrowly).
- */
-export async function hasFreshAttemptSubmission(cwd, runId, attemptId, homeOrScope) {
-    const { owned } = await readOwnedSubmissionRecords(cwd, runId, resolveReadScope(homeOrScope));
-    return owned.some((record) => {
-        if (recordAttemptId(record) !== attemptId)
-            return false;
-        if (record.kind === "sealed")
-            return true;
-        if (record.kind === "outcome") {
-            const payload = record.payload;
-            return payload?.type === "outcome" && payload.outcome === "audit-escalation";
-        }
-        return false;
-    });
-}
 function isTerminalRoleName(value) {
     return typeof value === "string" && value.length > 0;
 }
@@ -215,49 +186,31 @@ function mapOwnedToSubmissionRows(owned) {
         out.push(row);
     };
     for (const record of scoped) {
-        const attemptId = recordAttemptId(record);
-        if (record.kind === "candidate") {
-            const payload = record.payload;
-            if (payload?.type !== "candidate" || payload.params === undefined)
-                continue;
-            const callKey = typeof payload.toolCallId === "string"
-                ? submissionCallKey(attemptId, payload.toolCallId)
-                : undefined;
-            const fallback = callKey !== undefined ? roleByCall.get(callKey) : undefined;
-            take(rowFromPayload("candidate", payload, payload.params, fallback), callKey);
-            continue;
+        const payload = record.payload;
+        let kind;
+        let original;
+        if (record.kind === "candidate" && payload?.type === "candidate") {
+            kind = "candidate";
+            original = payload.params;
         }
-        if (record.kind === "sealed") {
-            const payload = record.payload;
-            if (payload?.type !== "sealed" || payload.accepted === undefined)
-                continue;
-            const callKey = typeof payload.toolCallId === "string"
-                ? submissionCallKey(attemptId, payload.toolCallId)
-                : undefined;
-            const fallback = callKey !== undefined ? roleByCall.get(callKey) : undefined;
-            take(rowFromPayload("accepted", payload, payload.accepted, fallback), callKey);
-            continue;
+        else if (record.kind === "sealed" && payload?.type === "sealed") {
+            kind = "accepted";
+            original = payload.accepted;
         }
-        if (record.kind !== "outcome")
-            continue;
-        const payload = record.payload;
-        if (payload?.type !== "outcome" || payload.accepted === undefined)
-            continue;
-        const outcome = payload.outcome;
-        const kind = outcome === "audit-escalation"
-            ? "audit-escalation"
-            : outcome === "correctable-rejection"
-                ? "correctable-rejection"
-                : outcome === "infrastructure"
-                    ? "infrastructure"
-                    : undefined;
-        if (kind === undefined)
+        else if (record.kind === "outcome" && payload?.type === "outcome") {
+            kind = payload.outcome === "audit-escalation" ? "audit-escalation"
+                : payload.outcome === "correctable-rejection" ? "correctable-rejection"
+                    : payload.outcome === "infrastructure" ? "infrastructure"
+                        : undefined;
+            original = payload.accepted;
+        }
+        if (kind === undefined || original === undefined || payload === undefined)
             continue;
         const callKey = typeof payload.toolCallId === "string"
-            ? submissionCallKey(attemptId, payload.toolCallId)
+            ? submissionCallKey(recordAttemptId(record), payload.toolCallId)
             : undefined;
-        const fallback = callKey !== undefined ? roleByCall.get(callKey) : undefined;
-        take(rowFromPayload(kind, payload, payload.accepted, fallback), callKey);
+        const fallback = callKey === undefined ? undefined : roleByCall.get(callKey);
+        take(rowFromPayload(kind, payload, original, fallback), callKey);
     }
     return out;
 }
@@ -270,8 +223,7 @@ function mapOwnedToSubmissionRows(owned) {
 export async function readRecordedSubmissionRows(cwd, runId, homeOrScope) {
     const scope = resolveReadScope(homeOrScope);
     const { owned } = await readOwnedSubmissionRecords(cwd, runId, scope);
-    const scoped = recordsForAttempt(owned, scope.attemptId);
-    return mapOwnedToSubmissionRows(scoped);
+    return mapOwnedToSubmissionRows(owned);
 }
 /**
  * Settlement-only this-court rows (#879 return path).
@@ -291,26 +243,6 @@ export async function readAttemptScopedSubmissionRows(cwd, runId, attemptId, hom
 export async function readRecordedSubmissions(cwd, runId, homeOrScope) {
     return (await readRecordedSubmissionRows(cwd, runId, homeOrScope)).map((row) => row.accepted);
 }
-/** True when the run has at least one recorded original payload (any outcome class). */
-export async function hasRecordedSubmission(cwd, runId, homeOrScope) {
-    return (await readRecordedSubmissionRows(cwd, runId, homeOrScope)).length > 0;
-}
-/** Latest non-final outcome on the run ledger (settlement residual precedence). */
-export async function readLatestSubmissionOutcome(cwd, runId, homeOrScope) {
-    const scope = resolveReadScope(homeOrScope);
-    const { owned } = await readOwnedSubmissionRecords(cwd, runId, scope);
-    const scoped = recordsForAttempt(owned, scope.attemptId);
-    for (let index = scoped.length - 1; index >= 0; index -= 1) {
-        const record = scoped[index];
-        if (record?.kind !== "outcome")
-            continue;
-        const payload = record.payload;
-        if (payload?.type === "outcome" && typeof payload.outcome === "string") {
-            return payload;
-        }
-    }
-    return undefined;
-}
 async function restoreState(cwd, runId, scope) {
     const located = await readOwnedSubmissionRecords(cwd, runId, scope);
     if (located.file === undefined) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@akagilnc/pi-workflow-roles",
-  "version": "0.1.5406",
+  "version": "0.1.5475",
   "description": "Soul-bound workflow roles for Pi",
   "type": "module",
   "license": "Apache-2.0",

package/packets/fixer-repair.md

@@ -2,4 +2,6 @@
 
 Repair and verify the caller-assigned findings. Treat this document as opaque prose: its headings and wording have no machine semantics.
 
-Search the authorized scope for every same-pattern occurrence, preserve unrelated behavior, run the required checks, and report each finding class through the typed Fixer output.
+Preserve unrelated behavior.
+
+Repair method: [Fixer Soul](../souls/fixer.md). Report fields: [Fixer output contract](../src/package-contracts/fixer-output.ts).

package/packets/judge-apply.md

@@ -1,18 +1,13 @@
 # Apply evidence packet (contributor template)
 
-This file is a **repository-contributor template** for a manually supplied Apply
-(executable-proof) artifact. The filename identifies an **evidence burden**, not
-a verdict, phase flag, transition, or required predecessor/successor.
-
-Selection, composition, and use are **caller-owned** (ADR 0010). Apply posture
-semantics remain solely in `souls/judge.md`. Git commit identity owns the code
-snapshot and any reviewed range. This template creates no receipt envelope,
-audit, runtime enforcement, trust tier, production hook, or package surface.
+Repository-contributor template for manually supplied Apply evidence.
+Template scope and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md).
+Composition: [ADR 0010](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/adr/0010-callers-own-role-composition-and-repetition.md).
+Adjudication: [Judge Soul](../souls/judge.md).
 
 ## Sealed upstream identities
 
-Digests seal **artifact-byte identity only**. They do not prove truth,
-acceptance, or freshness.
+Artifact identity and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md#artifact-preservation-rules).
 
 | Artifact | Repository-relative path | SHA-256 of exact bytes |
 | --- | --- | --- |
@@ -36,6 +31,8 @@ Distinguish carefully:
 
 None of these alone proves truth or acceptance.
 
+Adjudication duties: [Judge Soul](../souls/judge.md).
+
 ## Construction evidence
 
 ### Live code and tests
@@ -69,9 +66,3 @@ otherwise record `N/A — no guardrail added or approved` with disposition.
 | 1. Which real, reproducible failure proves this guardrail is needed? | |
 | 2. Which seam owns the invariant it protects? | |
 | 3. Why is deleting or simplifying the root cause insufficient for this failure class? | |
-
-## Explicit non-claims
-
-- Filename is not a verdict and does not imply Apply convergence.
-- This template does not define or alter role receipts, Soul audit, or runtime gates.
-- Instantiation does not route work to any role or require a successor packet.

package/packets/judge-authority.md

@@ -1,12 +1,9 @@
 # Authority evidence packet (contributor template)
 
-This file is a **repository-contributor template** for a manually supplied
-Authority-evidence artifact. The filename identifies an **evidence burden**, not
-a verdict, phase flag, transition, or required predecessor/successor.
-
-Selection, composition, and use are **caller-owned** (ADR 0010). This template
-does not restate or amend Judge law in `souls/judge.md`. It creates no routing,
-next-role, package memory, resume semantics, timeout budget, or trust tier.
+Repository-contributor template for manually supplied Authority evidence.
+Template scope and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md).
+Composition: [ADR 0010](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/adr/0010-callers-own-role-composition-and-repetition.md).
+Adjudication: [Judge Soul](../souls/judge.md).
 
 ## Identity seal
 
@@ -17,15 +14,14 @@ Every supplied artifact is identified by:
 | Repository-relative path | |
 | SHA-256 of exact bytes | |
 
-A digest seals **identity only**. It does not prove truth, acceptance, or
-freshness. Judge independently checks claims against the supplied current
-target. This packet does **not** require Apply-level executable proof.
+Artifact identity and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md#artifact-preservation-rules).
+Adjudication duties: [Judge Soul](../souls/judge.md).
 
 ## Authority items
 
 Record each item the contributor wants examined under the Authority burden.
 Do not invent Apply fixtures, blanket `file:line` demands, or implementation
-recipes here.
+recipes here. This packet does **not** require Apply-level executable proof.
 
 ### Clauses
 
@@ -56,9 +52,3 @@ recipes here.
 | ID | Open choice | Options | Blocking? |
 | --- | --- | --- | --- |
 | U1 | | | |
-
-## Explicit non-claims
-
-- Filename is not a verdict and does not imply convergence.
-- Instantiation does not start, order, or require any role call.
-- No mechanical schema/runtime enforcement is claimed by this template.

package/packets/judge-plan.md

@@ -1,19 +1,15 @@
 # Plan evidence packet (contributor template)
 
-This file is a **repository-contributor template** for a manually supplied Plan
-(construction-readiness) artifact. The filename identifies an **evidence
-burden**, not a verdict, phase flag, transition, or required
-predecessor/successor.
-
-Selection, composition, and use are **caller-owned** (ADR 0010). Plan posture
-semantics remain solely in `souls/judge.md`. This template does not restate or
-amend that law. It creates no routing, topology, role-order, package memory, or
-proof that construction occurred.
+Repository-contributor template for manually supplied Plan evidence.
+This packet does not claim construction already happened.
+Template scope and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md).
+Composition: [ADR 0010](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/adr/0010-callers-own-role-composition-and-repetition.md).
+Adjudication: [Judge Soul](../souls/judge.md).
 
 ## Sealed authority identity
 
-Bind the authority materials this plan consumes. Digests seal identity only—not
-truth, acceptance, or freshness.
+Bind the authority materials this plan consumes. Artifact identity and preservation:
+[development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md#artifact-preservation-rules).
 
 | Artifact | Repository-relative path | SHA-256 of exact bytes |
 | --- | --- | --- |
@@ -46,10 +42,3 @@ fixture pseudocode or blanket `file:line` here.
 
 Add P3… as needed. Every planned change must be reconcilable to the sealed
 authority identity above and must carry all five facts.
-
-## Explicit non-claims
-
-- Filename is not a verdict and does not authorize or prove Apply success.
-- This packet does not claim construction already happened.
-- No orchestration, next-role, or stage-machine semantics arise from this file.
-- No mechanical schema/runtime enforcement is claimed by this template.

package/packets/judge-review.md

@@ -1,15 +1,10 @@
 # Review evidence packet (contributor template)
 
-This file is a **repository-contributor template** for a manually supplied
-Review (finding-adjudication) artifact. The filename identifies an **evidence
-burden**, not a verdict, phase flag, transition, or required
-predecessor/successor.
-
-Selection, composition, and use are **caller-owned** (ADR 0010). This template
-records adjudication evidence only. It does **not** define Reviewer protocol,
-alter Reviewer method/audit, or alter Collector ledger semantics. Judge Review
-posture meaning remains in `souls/judge.md`. No generic Reviewer/Collector
-provider law is introduced (ADR 0011).
+Repository-contributor template for manually supplied Review evidence.
+Template scope and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md).
+Composition: [ADR 0010](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/adr/0010-callers-own-role-composition-and-repetition.md).
+Adjudication: [Judge Soul](../souls/judge.md).
+External reviewer protocols: [ADR 0011](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/adr/0011-collector-adapts-to-documented-reviewer-protocols.md).
 
 ## Sealed authority identity
 
@@ -17,13 +12,12 @@ provider law is introduced (ADR 0011).
 | --- | --- | --- |
 | Authority materials | | |
 
-A digest seals identity only—not truth, acceptance, or freshness.
+Artifact identity and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md#artifact-preservation-rules).
 
 ## Fixed reviewed range
 
-Every finding and disposition in this packet concerns **one immutable range**
-and current target facts. A changed target or range requires a **new**
-artifact/digest; do not silently mutate this packet.
+Identify the reviewed range and current target below. Amendment procedure:
+[development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md#artifact-preservation-rules).
 
 | Field | Full SHA |
 | --- | --- |
@@ -39,11 +33,3 @@ current facts. Every finding needs an explicit disposition.
 | Finding ID | Claim | Authority binding | Evidence on current target/range | Disposition | Disposition evidence |
 | --- | --- | --- | --- | --- | --- |
 | F1 | | | | sustained / rejected / deferred / other | |
-
-## Explicit non-claims
-
-- Filename is not a verdict and does not imply Review convergence.
-- This packet defines no required review order, repetition count, or return path
-  to any role.
-- No routing, next-role, stage machine, or provider-universal protocol arises here.
-- No mechanical schema/runtime enforcement is claimed by this template.

package/packets/judge-submission.md

@@ -1,12 +1,9 @@
 # Submission-case evidence template (contributor template)
 
-This file is a **repository-contributor template** for the evidence record of a
-manually supplied Submission case (ticket face / plan / disposition ledger /
-constitutional review). The submission-case duty law is solely in
-`souls/judge.md`; this file supplies its evidence-record format.
-
-Selection, composition, and use are **caller-owned** (ADR 0010). This template
-creates no routing, topology, role-order, or package memory.
+Repository-contributor template for manually supplied Submission-case evidence.
+Template scope and preservation: [development closure](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/development-closure.md).
+Composition: [ADR 0010](https://github.com/Akagilnc/ak-pi-workflow-roles/blob/main/docs/adr/0010-callers-own-role-composition-and-repetition.md).
+Adjudication: [Judge Soul](../souls/judge.md).
 
 ## Identity seal
 

package/resources/collector-bot-handbook.md

@@ -4,7 +4,7 @@
 内容是角色阅读与修正的工作记忆，不是代码状态规则；临时额度/故障不得固化为永久行为。
 参与者依本仓使用与现场活动辨认，不把下列名称当作全仓固定名单。
 
-#1088：用宿主 CLI（如 `gh`）自行取证与请求；代码不代收、不选择、不归并 finding。取证受限时如实报告。
+#1088：用宿主 CLI（如 `gh`）自行取证与请求；代码不代收、不选择、不归并 finding。
 
 ## Codex（hosted GitHub connector）
 
@@ -31,7 +31,7 @@
 
 ## 工序
 
-从任务材料与 `--pr` / `--request-manifest` 材料判定目标后，用宿主 CLI 读取有评审的 PR，按各 reviewer 已记录的完成形态收集 findings，整理并交卷。未绑定 PR 时用 `gh` 等自行定位；无法确定时不猜，交给调用方明确目标。交卷交按机器身份归组的出席、材料与 findings；没有 finding 就不要编一条。未完成写现场原因，不写成没问题。
+从任务材料与 `--pr` / `--request-manifest` 材料判定目标后，用宿主 CLI 读取有评审的 PR，按各 reviewer 已记录的完成形态收集 findings，整理并交卷。未绑定 PR 时用 `gh` 等自行定位；无法确定时不猜，交给调用方明确目标。证据归属与不确定性的报告原则见 [通进司 Soul](../souls/collector.md)，字段见 [交卷契约](../src/package-contracts/collector-output.ts)。
 
 ## 修正规则