Adopt language-aware requirements intake governance

.agents/skills/speckit-intake-create-status/SKILL.md

@@ -38,9 +38,17 @@ This command is strictly read-only.
    response hash.
 7. Hash the receipt again and verify Git status for the inspected paths. Any
    mutation makes the status check fail.
-8. Classify the result as `Current`, `SourceDrift`, `TargetDrift`, `UrlDrift`,
-   `NeedsClarification`, `InvalidReceipt`, or `Missing`.
+8. If `requirements/intake-governance-config.json` exists, run the installed
+   configuration validator read-only. Resolve BCP-47 documentation language,
+   naming profile, four portable roles, collection paths, legacy aliases,
+   computed inventory, canonical index, and Series evidence. Do not infer
+   documentation language from implementation language or locale.
+9. Classify a requirements collection as exactly `Aligned`,
+   `MigrationRequired`, `NeedsClarification`, or `Blocked`. Preserve the
+   existing receipt-level classifications when no requirements collection is
+   selected.
 
 Finish with classification, target, receipt, source freshness, prompt state,
-delivery authority, validation exit, and exact next action. Never repair,
-review, run Specify, or start an autonomous command implicitly.
+delivery authority, resolved requirements roles when applicable, validation
+exit, and exact next action. Never repair, review, run Specify, or start an
+autonomous command implicitly.

.agents/skills/speckit-intake-delete/SKILL.md

@@ -16,7 +16,7 @@ metadata:
 $ARGUMENTS
 ```
 
-Delete is logical and reversible from versioned evidence. v0.2.1 has no purge.
+Delete is logical and reversible from versioned evidence. v0.3.0 has no purge.
 
 1. Resolve exactly one active intake or one complete series. Require the exact
    identity, a deletion reason, and explicit current delete authority.

.agents/skills/speckit-intake-review-status/SKILL.md

@@ -24,6 +24,9 @@ invoke the installed validator with the repository root, and report:
 - findings, questions, accepted-risk ownership, and re-evaluation triggers;
 - series and campaign coverage, including unique intake and worker counts;
 - for Series mode, request hash, root set, target order, edge count, and DAG validity;
+- for schema-2.0 requirements collections, language, profile, portable roles,
+  resolved paths, computed inventory, canonical index count, and eligible
+  candidate;
 - whether Preset 7 or Preset 8 may consume the result under current policy;
 - the exact next command.
 

.agents/skills/speckit-intake-review/SKILL.md

@@ -38,16 +38,21 @@ exist. Use the installed policy and checklist.
    intake, every worker applicability row, repository/language/stack fit,
    manifest DAG/base/handoff agreement, and separately governed operator
    exceptions.
-5. In interactive mode ask at most five material questions per pass, exactly
+5. When `requirements/intake-governance-config.json` is in scope, validate
+   schema 2.0 first. Review explicit BCP-47 documentation language, resolved
+   naming profile, four portable roles, collection paths, bounded legacy
+   aliases, computed inventory, canonical index, hashes, receipts, references,
+   and the distinction between `Eligible` and delivery authority.
+6. In interactive mode ask at most five material questions per pass, exactly
    one at a time. In batch mode record blockers and do not guess.
-6. Assign stable finding IDs, severity, category, owner, evidence, disposition,
+7. Assign stable finding IDs, severity, category, owner, evidence, disposition,
    and re-evaluation trigger. Outcomes are exactly `Ready`,
    `ReadyWithAcceptedRisks`, `NeedsClarification`, `NeedsRemediation`, or
    `Rejected`.
-7. Critical or High findings block. `ReadyWithAcceptedRisks` permits only
+8. Critical or High findings block. `ReadyWithAcceptedRisks` permits only
    Medium/Low risks with human owner, rationale, acceptance date, evidence, and
    re-evaluation trigger. An autonomous agent never accepts risk.
-8. Write the machine-readable result and readable report outside the reviewed
+9. Write the machine-readable result and readable report outside the reviewed
    target files, then validate the result with the installed Bash or PowerShell
    validator. Supersede older results explicitly; never silently overwrite
    their audit meaning.

.agents/skills/speckit-intake-series-next/SKILL.md

@@ -21,12 +21,16 @@ Select candidates only from a valid, current named series.
 1. Run the read-only status contract first.
 2. A target is eligible only when its lifecycle permits selection and every
    binding predecessor is `Completed`.
-3. Report all eligible targets in visible order. If none are eligible, report
+3. Require exactly one `Eligible` target when the manifest declares a preferred
+   next intake. More than one is invalid. An eligible result is ordering
+   evidence only and grants no implementation, push, PR, merge, bypass, or
+   provider authority.
+4. Report all eligible targets in visible order. If none are eligible, report
    each exact blocker and evidence path.
-4. Distinguish preferred order and shared-writer serialization from binding
+5. Distinguish preferred order and shared-writer serialization from binding
    functional dependencies.
-5. Revalidate downstream review freshness and user authority only when a later
+6. Revalidate downstream review freshness and user authority only when a later
    command is separately invoked.
-6. Never start Intake Review, Specify, Autonomous, or Parallel Autonomous.
+7. Never start Intake Review, Specify, Autonomous, or Parallel Autonomous.
 
 Finish with a copy-ready suggested command only; do not execute it.

.agents/skills/speckit-intake-series-status/SKILL.md

@@ -21,8 +21,11 @@ Inspect one named series read-only.
 2. Run Bash and PowerShell validators.
 3. Report identity, status, targets, roots, dependencies, eligible targets,
    blockers, receipt lineage, archive/tombstone state, and drift.
-4. Classify ambiguity or drift fail-closed. Do not repair it.
-5. Prove before/after hashes and Git status are unchanged.
-6. Do not stage, commit, push, review, or execute a target.
+4. When schema-2.0 requirements governance is configured, validate it first and
+   resolve target paths from its portable roles and collection paths. Confirm
+   that the human-readable order and manifest use the same resolved paths.
+5. Classify ambiguity or drift fail-closed. Do not repair it.
+6. Prove before/after hashes and Git status are unchanged.
+7. Do not stage, commit, push, review, or execute a target.
 
 Finish with one exact next action.

.agents/skills/speckit-intake-update/SKILL.md

@@ -42,9 +42,18 @@ Update is the only normal command that changes an active intake.
    crawl, temporary-source, prompt-safety, and context-limit rules as Create.
 9. Prepare and validate all outputs transactionally before publishing active
    changes. A failure leaves prior active artifacts unchanged.
-10. Supersede any old Intake Review result explicitly. Never report an old
-    review as current after target or series drift.
-11. Run both installed validators before completion.
+10. A requirements-collection migration uses
+    `requirements/intake-governance-config.json` schema 2.0 and an operation
+    journal. Resolve language, names, roles, collection paths, and legacy
+    aliases before mutation. Record before/after hashes, moves, reference
+    updates, validations, rollback, and repair boundary. Existing names remain
+    unchanged without separate rename authority.
+11. Publish configuration, index, manifest, receipts, prompts, agent guidance,
+    and links atomically. A partial failure must roll back completely or end as
+    `NeedsRepair`; never report a mixed state as success.
+12. Supersede any old Intake Review result explicitly. Never report an old
+   review as current after target or series drift.
+13. Run both installed validators before completion.
 
 Finish with updated identities, old/new hashes, lineage, archive paths,
 validation, review invalidation, and exactly one Intake Review next action.

.claude/skills/speckit-intake-create-status/SKILL.md

@@ -36,9 +36,17 @@ This command is strictly read-only.
    response hash.
 7. Hash the receipt again and verify Git status for the inspected paths. Any
    mutation makes the status check fail.
-8. Classify the result as `Current`, `SourceDrift`, `TargetDrift`, `UrlDrift`,
-   `NeedsClarification`, `InvalidReceipt`, or `Missing`.
+8. If `requirements/intake-governance-config.json` exists, run the installed
+   configuration validator read-only. Resolve BCP-47 documentation language,
+   naming profile, four portable roles, collection paths, legacy aliases,
+   computed inventory, canonical index, and Series evidence. Do not infer
+   documentation language from implementation language or locale.
+9. Classify a requirements collection as exactly `Aligned`,
+   `MigrationRequired`, `NeedsClarification`, or `Blocked`. Preserve the
+   existing receipt-level classifications when no requirements collection is
+   selected.
 
 Finish with classification, target, receipt, source freshness, prompt state,
-delivery authority, validation exit, and exact next action. Never repair,
-review, run Specify, or start an autonomous command implicitly.
+delivery authority, resolved requirements roles when applicable, validation
+exit, and exact next action. Never repair, review, run Specify, or start an
+autonomous command implicitly.

.claude/skills/speckit-intake-create/SKILL.md

@@ -92,4 +92,4 @@ Finish with status, targets, receipts or series manifest, source count, question
 and decision counts, delivery authority, validation results, and exactly one
 next action. For one `ReadyForReview` target, use
 `$speckit-intake-review <target>`. For a series, use the generated schema-1.1
-Series request. Report the command without executing it.
+Series request. Report the command without executing it.

.claude/skills/speckit-intake-delete/SKILL.md

@@ -14,7 +14,7 @@ metadata:
 $ARGUMENTS
 ```
 
-Delete is logical and reversible from versioned evidence. v0.2.1 has no purge.
+Delete is logical and reversible from versioned evidence. v0.3.0 has no purge.
 
 1. Resolve exactly one active intake or one complete series. Require the exact
    identity, a deletion reason, and explicit current delete authority.
@@ -40,4 +40,4 @@ Delete is logical and reversible from versioned evidence. v0.2.1 has no purge.
 
 Finish with deleted identities, archive and tombstone paths, validation,
 invalidated references, and exact next action. Never recreate, review, specify,
-or start an autonomous command implicitly.
+or start an autonomous command implicitly.

.claude/skills/speckit-intake-review-status/SKILL.md

@@ -22,6 +22,9 @@ invoke the installed validator with the repository root, and report:
 - findings, questions, accepted-risk ownership, and re-evaluation triggers;
 - series and campaign coverage, including unique intake and worker counts;
 - for Series mode, request hash, root set, target order, edge count, and DAG validity;
+- for schema-2.0 requirements collections, language, profile, portable roles,
+  resolved paths, computed inventory, canonical index count, and eligible
+  candidate;
 - whether Preset 7 or Preset 8 may consume the result under current policy;
 - the exact next command.
 

.claude/skills/speckit-intake-review/SKILL.md

@@ -36,20 +36,25 @@ exist. Use the installed policy and checklist.
    intake, every worker applicability row, repository/language/stack fit,
    manifest DAG/base/handoff agreement, and separately governed operator
    exceptions.
-5. In interactive mode ask at most five material questions per pass, exactly
+5. When `requirements/intake-governance-config.json` is in scope, validate
+   schema 2.0 first. Review explicit BCP-47 documentation language, resolved
+   naming profile, four portable roles, collection paths, bounded legacy
+   aliases, computed inventory, canonical index, hashes, receipts, references,
+   and the distinction between `Eligible` and delivery authority.
+6. In interactive mode ask at most five material questions per pass, exactly
    one at a time. In batch mode record blockers and do not guess.
-6. Assign stable finding IDs, severity, category, owner, evidence, disposition,
+7. Assign stable finding IDs, severity, category, owner, evidence, disposition,
    and re-evaluation trigger. Outcomes are exactly `Ready`,
    `ReadyWithAcceptedRisks`, `NeedsClarification`, `NeedsRemediation`, or
    `Rejected`.
-7. Critical or High findings block. `ReadyWithAcceptedRisks` permits only
+8. Critical or High findings block. `ReadyWithAcceptedRisks` permits only
    Medium/Low risks with human owner, rationale, acceptance date, evidence, and
    re-evaluation trigger. An autonomous agent never accepts risk.
-8. Write the machine-readable result and readable report outside the reviewed
+9. Write the machine-readable result and readable report outside the reviewed
    target files, then validate the result with the installed Bash or PowerShell
    validator. Supersede older results explicitly; never silently overwrite
    their audit meaning.
 
 Finish with outcome, target and worker counts, findings by severity, accepted
 risks, open questions, result path, and exact next action. Do not start Specify
-or an autonomous run implicitly.
+or an autonomous run implicitly.

.claude/skills/speckit-intake-series-create/SKILL.md

@@ -34,4 +34,4 @@ review or feature workflow.
 7. Publish all prepared files atomically. A failure leaves an explicit
    incomplete operation and no partially active series.
 8. Finish with series ID, cardinalities, validation results, eligible targets,
-   blockers, and exactly `$speckit-intake-series-status` as the next action.
+   blockers, and exactly `$speckit-intake-series-status` as the next action.

.claude/skills/speckit-intake-series-next/SKILL.md

@@ -19,12 +19,16 @@ Select candidates only from a valid, current named series.
 1. Run the read-only status contract first.
 2. A target is eligible only when its lifecycle permits selection and every
    binding predecessor is `Completed`.
-3. Report all eligible targets in visible order. If none are eligible, report
+3. Require exactly one `Eligible` target when the manifest declares a preferred
+   next intake. More than one is invalid. An eligible result is ordering
+   evidence only and grants no implementation, push, PR, merge, bypass, or
+   provider authority.
+4. Report all eligible targets in visible order. If none are eligible, report
    each exact blocker and evidence path.
-4. Distinguish preferred order and shared-writer serialization from binding
+5. Distinguish preferred order and shared-writer serialization from binding
    functional dependencies.
-5. Revalidate downstream review freshness and user authority only when a later
+6. Revalidate downstream review freshness and user authority only when a later
    command is separately invoked.
-6. Never start Intake Review, Specify, Autonomous, or Parallel Autonomous.
+7. Never start Intake Review, Specify, Autonomous, or Parallel Autonomous.
 
 Finish with a copy-ready suggested command only; do not execute it.

.claude/skills/speckit-intake-series-status/SKILL.md

@@ -19,8 +19,11 @@ Inspect one named series read-only.
 2. Run Bash and PowerShell validators.
 3. Report identity, status, targets, roots, dependencies, eligible targets,
    blockers, receipt lineage, archive/tombstone state, and drift.
-4. Classify ambiguity or drift fail-closed. Do not repair it.
-5. Prove before/after hashes and Git status are unchanged.
-6. Do not stage, commit, push, review, or execute a target.
+4. When schema-2.0 requirements governance is configured, validate it first and
+   resolve target paths from its portable roles and collection paths. Confirm
+   that the human-readable order and manifest use the same resolved paths.
+5. Classify ambiguity or drift fail-closed. Do not repair it.
+6. Prove before/after hashes and Git status are unchanged.
+7. Do not stage, commit, push, review, or execute a target.
 
 Finish with one exact next action.

.claude/skills/speckit-intake-update/SKILL.md

@@ -40,9 +40,18 @@ Update is the only normal command that changes an active intake.
    crawl, temporary-source, prompt-safety, and context-limit rules as Create.
 9. Prepare and validate all outputs transactionally before publishing active
    changes. A failure leaves prior active artifacts unchanged.
-10. Supersede any old Intake Review result explicitly. Never report an old
-    review as current after target or series drift.
-11. Run both installed validators before completion.
+10. A requirements-collection migration uses
+    `requirements/intake-governance-config.json` schema 2.0 and an operation
+    journal. Resolve language, names, roles, collection paths, and legacy
+    aliases before mutation. Record before/after hashes, moves, reference
+    updates, validations, rollback, and repair boundary. Existing names remain
+    unchanged without separate rename authority.
+11. Publish configuration, index, manifest, receipts, prompts, agent guidance,
+    and links atomically. A partial failure must roll back completely or end as
+    `NeedsRepair`; never report a mixed state as success.
+12. Supersede any old Intake Review result explicitly. Never report an old
+   review as current after target or series drift.
+13. Run both installed validators before completion.
 
 Finish with updated identities, old/new hashes, lineage, archive paths,
 validation, review invalidation, and exactly one Intake Review next action.

.github/agents/speckit.intake-create-status.agent.md

@@ -33,9 +33,17 @@ This command is strictly read-only.
    response hash.
 7. Hash the receipt again and verify Git status for the inspected paths. Any
    mutation makes the status check fail.
-8. Classify the result as `Current`, `SourceDrift`, `TargetDrift`, `UrlDrift`,
-   `NeedsClarification`, `InvalidReceipt`, or `Missing`.
+8. If `requirements/intake-governance-config.json` exists, run the installed
+   configuration validator read-only. Resolve BCP-47 documentation language,
+   naming profile, four portable roles, collection paths, legacy aliases,
+   computed inventory, canonical index, and Series evidence. Do not infer
+   documentation language from implementation language or locale.
+9. Classify a requirements collection as exactly `Aligned`,
+   `MigrationRequired`, `NeedsClarification`, or `Blocked`. Preserve the
+   existing receipt-level classifications when no requirements collection is
+   selected.
 
 Finish with classification, target, receipt, source freshness, prompt state,
-delivery authority, validation exit, and exact next action. Never repair,
-review, run Specify, or start an autonomous command implicitly.
+delivery authority, resolved requirements roles when applicable, validation
+exit, and exact next action. Never repair, review, run Specify, or start an
+autonomous command implicitly.

.github/agents/speckit.intake-create.agent.md

@@ -89,4 +89,4 @@ Finish with status, targets, receipts or series manifest, source count, question
 and decision counts, delivery authority, validation results, and exactly one
 next action. For one `ReadyForReview` target, use
 `$speckit-intake-review <target>`. For a series, use the generated schema-1.1
-Series request. Report the command without executing it.
+Series request. Report the command without executing it.

.github/agents/speckit.intake-delete.agent.md

@@ -11,7 +11,7 @@ description: Logically delete active intake artifacts through archive and tombst
 $ARGUMENTS
 ```
 
-Delete is logical and reversible from versioned evidence. v0.2.1 has no purge.
+Delete is logical and reversible from versioned evidence. v0.3.0 has no purge.
 
 1. Resolve exactly one active intake or one complete series. Require the exact
    identity, a deletion reason, and explicit current delete authority.
@@ -37,4 +37,4 @@ Delete is logical and reversible from versioned evidence. v0.2.1 has no purge.
 
 Finish with deleted identities, archive and tombstone paths, validation,
 invalidated references, and exact next action. Never recreate, review, specify,
-or start an autonomous command implicitly.
+or start an autonomous command implicitly.

.github/agents/speckit.intake-review-status.agent.md

@@ -19,6 +19,9 @@ invoke the installed validator with the repository root, and report:
 - findings, questions, accepted-risk ownership, and re-evaluation triggers;
 - series and campaign coverage, including unique intake and worker counts;
 - for Series mode, request hash, root set, target order, edge count, and DAG validity;
+- for schema-2.0 requirements collections, language, profile, portable roles,
+  resolved paths, computed inventory, canonical index count, and eligible
+  candidate;
 - whether Preset 7 or Preset 8 may consume the result under current policy;
 - the exact next command.