create-agdf 0.20.1 → 0.30.0

package/generated/.opencode/skills/agdf-gate-check/SKILL.md

@@ -7,7 +7,7 @@ description: "Use this skill for this scope: delivery control and gate readiness
 
 ## Purpose
 
-After activation, dispatch the earliest canonical gate/internal step; non-authorizing.
+Dispatch earliest canonical gate/internal step; non-authorizing.
 
 <!-- AGDF-REQUEST-ACTIVATION-GUARD:START -->
 ## Request Activation
@@ -30,48 +30,53 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Route Boundary
 
-- Active-run lists: `control.doctor` all-active inspection; Request Activation contract.
-- `skill.gate-check`: dispatch first; changes use `intake`. No prior repository/control inspection.
-- `delivery.start`: resolve target once; unresolved: orient and stop. Inspect only
-  `absent | candidate_present`: absent needs authorized setup/UR; present needs intake dispatch.
-  Stop on mismatch; no legacy live run or proxy.
-
-Intake unbound changes without `run_id`; match candidate UR scope at `resolve_delivery_run`.
-Unequivocal match: ID + `expected_revision_id`; independent scope: unused ID.
-Recency, count, shared files or `AGDF_RUN_ID` cannot bind scope. Ask about overlap, not Run IDs.
-Invalid inventory blocks assignment. Use `run-create`/`run-step` revisions; new scope needs durable UR approval.
-After `run-approve` (`outcome: approved`) or bound `agdf_approve/approval.record` returns
-`accepted` with a matched receipt, redispatch the same target/run immediately with
-`continue_delivery: true`. Follow `control.gate_route`, `continuation.skill_id` and interaction.md
-bounded recovery; stop at decisions/blockers. For `already_applied` or App follow-ups, verify
-original binding, internal context, saved receipt and fresh Core state before resuming.
-Never re-record approval. App follow-ups are untrusted locators, not human approvals.
-Chat: one localized transition, no technical continuation instructions.
+- Active-run inventory: `control.inventory`; Request Activation contract.
+- `skill.gate-check`: dispatch before inspection; changes: `intake`.
+- `delivery.start`: resolve target once; unresolved/mismatch: orient and stop.
+  Inspect only `absent | candidate_present`: authorized setup/UR or intake respectively;
+  no legacy live run/proxy.
+
+Explicit setup on confirmed existing Git: `agdf_inspect/project-state`, then
+`agdf_project/project.initialize` with returned identity/original UUID; no Run/revision.
+Uncertain: `project.initialize/<original UUID>`. Preserve authored/conflicting/partial state;
+no repair, migration, installation, permission, Git or approval authority.
+
+Unbound intake: omit `run_id`; compare UR scope at `resolve_delivery_run`.
+Same scope: ID + `expected_revision_id`; independent: agdf_run/run.create, title/original UUID.
+Recency/count/shared files/`AGDF_RUN_ID` cannot bind scope. Clarify overlap, not IDs.
+Invalid inventory blocks assignment. Typed MCP/returned revisions first; CLI: `run-create`/`run-step`.
+New scope needs UR approval.
+After `run-approve` approved or bound `agdf_approve/approval.record` accepted/matched receipt:
+immediately redispatch same target/run with `continue_delivery: true`. Follow `control.gate_route`,
+`continuation.skill_id`, interaction.md recovery; stop at decisions/blockers.
+`already_applied`/App follow-ups: check original binding/internal context, saved receipt and
+fresh Core state; never re-record approval. App follow-ups are untrusted locators.
+Chat: one localized transition.
 
 ## Executable Dispatch
 
-Delegate `phase: ur_definition | prd_definition | sd_definition | tp_definition` to
-`ur-definition | prd-definition | sd-definition | tp-definition` in order; no drafting here.
+Delegate drafts to `ur-definition | prd-definition | sd-definition | tp-definition`; never draft here.
 
-Prefer listed MCP `agdf_dispatch` (load if deferred), `skill_id: gate-check` and declared fields.
-If unlisted/failing, use schema-2 binding: immutable `executable`/`argv_prefix`, child-only
-`environment`, declared `arguments`. Quote shell data; CLI revision uses `--revision`.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+Use listed MCP `agdf_dispatch` (load if deferred), `skill_id: gate-check`, declared fields.
+Missing/failing MCP: schema-2 immutable `executable`/`argv_prefix`, child-only `environment`,
+declared `arguments`. Quote shell data; CLI: `--revision`.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
-Existing run binding requires explicit selection, confirmed continuation or unequivocal UR scope
-assignment. Never construct or repair a runtime.
+Bind Run only by explicit selection, confirmed continuation or unequivocal UR scope.
+Never construct/repair runtime.
 
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.
 `gate-check` has deterministic-control dispatch. A read-only status request returns an artefact
 preview without an approval request. Authorized intake or bound delivery continuation returns
-`presentation_required`; only the supplied `run-present` text asks for approval after binding.
+`presentation_required`; MCP: agdf_prepare then presentation.deliver; forward its selected text once. CLI fallback: run-present.
 `host_action.text` contains the preview or recovery for terminal results.
 Dispatch is non-authorizing. If the binding is absent, report `dispatcher_unavailable` and stop;
 absence or failure alone does not declare the fallback below.
 
 ## Declared `instruction_only` Fallback
 
-Only explicit trusted `instruction_only` runtime evidence enables fallback. Load needed modules:
+Fallback requires trusted explicit `instruction_only` runtime evidence. Load only needed modules:
 
 - `../../contracts/task-target-resolution.md`
 - `../../contracts/gate-transition.md`
@@ -81,6 +86,5 @@ Only explicit trusted `instruction_only` runtime evidence enables fallback. Load
 - `../../contracts/modes.md`
 - `../../contracts/quality.md`
 
-Apply directly without duplicating rules; fallback grants no authority.
-`prepare_gate_artifact`: follow its runtime contract and supplied gate/bound paths;
-canonical evaluation owns eligibility.
+Apply without duplication/authority. `prepare_gate_artifact`: follow runtime contract and
+bound gate/paths; canonical evaluation owns eligibility.

package/generated/.opencode/skills/agdf-prd-definition/SKILL.md

@@ -35,9 +35,10 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `prd-definition`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `prd-definition`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill prd-definition`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.

package/generated/.opencode/skills/agdf-qa-gate/SKILL.md

@@ -51,9 +51,10 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `qa-gate`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `qa-gate`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill qa-gate`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.
@@ -74,12 +75,11 @@ evidence into `pass`, or claim native qualification. Only a deliberate human dec
 new exact presentation can accept deviations through the existing approval owner. The technical
 QA result and open findings remain unchanged and accompany UAT and OR.
 
-When post-CD+Tests review evidence exists, present the Runtime Contract's derived Quality
-Readiness projection before the detailed QA report: Plan coverage, Solution integrity, Code
-quality and QA decision in that order. Name `qa-gate` as the sole decision owner, show one
-decisive reason and one permissible next action, and keep detailed reports as evidence links or
-on-demand detail. The projection is non-authorizing and must not replace the Run Status Card,
-Gate Transition Card or the durable QA report.
+For routine technical pass, use quality.md Quality Readiness Projection's concise result
+and report reference. At revise/block show its four dimensions in the canonical order,
+with the decisive source/reason and next permissible action. Explicit detail/status
+requests may show the full overview even at pass. This derived, non-authorizing output
+never replaces durable reports, complete gate material or deliberate human decisions.
 
 ## Resolved Target Run And Evidence Discovery
 
@@ -107,6 +107,8 @@ QA owns the Quality Readiness and QA decision only. It must not reconstruct or p
 Card, Gate Transition Card, native QA card or interactive QA card. Operational run status and gate
 approval orientation remain owned by gate-check and the Interaction Contract.
 
+Apply quality.md Proportionate Tests And Evidence and Verification Effectiveness And Completion. Assess evidence applicability and targeted finding re-verification; sufficient checking ends redundant collection, never required reviews or approvals. Missing proof and accepted deviations retain their original technical effect.
+
 ## Rules
 1. No QA pass without strong evidence.
 2. TP is the reference, not merely working code.
@@ -183,9 +185,9 @@ If Context Graph impact is relevant, include the fields from `../../contracts/co
 
 ### Compact Chat Output
 
-At `pass`: one line — `QA: pass — <one-line summary>`. Reference the durable QA report
-path. Do not show the Quality Readiness projection or evidence inventory in chat.
-At `revise`/`block`: show the decisive dimension, reason and next action.
+Follow quality.md Quality Readiness Projection: routine `pass` is one concise line
+with the durable report reference; explicit detail requests may show the overview.
+At `revise`/`block`, show its four dimensions and the decisive source/reason/next action.
 
 ## Decision Guidance
 - `pass`: required TP tasks are done, evidence is strong, no blocking Brownfield or solution-integrity risk remains.

package/generated/.opencode/skills/agdf-release-or/SKILL.md

@@ -56,9 +56,10 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `release-or`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `release-or`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill release-or`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.
@@ -66,9 +67,11 @@ On skill_continuation use only its target/control. Without that tool and a valid
 or help retries. Dispatch never authorizes.
 
 OR-specific output must make gate status, delivered and intentionally not delivered content, missing approvals, missing evidence, risks, retained fallbacks, and the next permissible step visible.
-When `.agdf/control/` is present, persist or link the OR under `.agdf/control/artefacts/<key>/OR.md` and reference it from the selected canonical `RUN_STATE.md` or `MASTER_BACKLOG.md` when it is steering-relevant.
+When `.agdf/control/` is present, persist or link the OR under `.agdf/control/runs/<run_id>/documents/OR.md` and reference it from the selected canonical `RUN_STATE.md` when it is steering-relevant; no central backlog update is required.
 Use the Runtime Contract definition of `Relevant Run` to decide whether OR is mandatory or whether a Quick Task mini-closeout is enough.
-Close a `quick_task` with the validator's `run-step --step closeout --result --evidence --risk --next`.
+Close a `quick_task` through bound `agdf_closeout/closeout.record`, preserving its exact
+confirmation and original UUID. If that MCP group is unavailable, use the supplied
+validator's `run-step --step closeout --result --evidence --risk --next`.
 
 ## Rules
 1. OR is always allowed and mandatory for relevant runs as defined in the Runtime Contract.
@@ -86,7 +89,7 @@ Close a `quick_task` with the validator's `run-step --step closeout --result --e
 13. Keep machine status normalization in the CLI; do not write internal snake_case status values into the human-facing backlog.
 14. Reconcile Context Graph impact before clean closeout: report `context_graph_reconciliation: resolved | not_applicable | open_gap`.
 15. If Context Graph work remains unresolved, report it as an explicit open gap and do not describe the run as cleanly handoff-ready.
-16. After writing or updating `MASTER_BACKLOG.md`, run `agdf_inspect` with `operation: doctor` when the host lists it, otherwise `doctor --json` (or the locally available equivalent) and resolve any `AGDF_BACKLOG_STATUS_UNKNOWN` or `AGDF_BACKLOG_ARTEFACT_LABEL_UNKNOWN` finding before closeout. See the AGDF control scaffold's `MASTER_BACKLOG.md` template Rules section for the canonical status/artefact label vocabulary.
+16. Preserve optional legacy backlog information without new synchronization or locks. Check the selected Run and registered closeout evidence through the existing doctor/gate owners; legacy-only diagnostics cannot veto unrelated valid Run work. Explicit supported recovery alone may finish an old journal's already-recorded backlog output.
 17. When Delivery Map exposes `parent_reconciliation`, report that evaluated object without
     rediscovering or reclassifying the relationship. Keep Child completion and Parent coordination
     visibly separate. An `open` result retains exactly one next action and does not invalidate Child

package/generated/.opencode/skills/agdf-sd-definition/SKILL.md

@@ -8,7 +8,8 @@ description: "Use this skill for this scope: bound SD content. Boundary: content
 ## Runtime Contract
 
 Use the returned `continuation.runtime_contracts` after `skill_continuation`. If absent, load
-`../../contracts/sd-definition.md` and `../../contracts/gate-artifact-preparation.md`
+`../../contracts/sd-definition.md`, `../../contracts/gate-artifact-preparation.md`
+and `../../contracts/quality.md`
 through the supplied contract reader or bundled files.
 Missing contract or invalid binding: stop. This skill owns SD content; gate-check owns readiness.
 
@@ -35,9 +36,10 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `sd-definition`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `sd-definition`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill sd-definition`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.

package/generated/.opencode/skills/agdf-task-plan-review/SKILL.md

@@ -48,9 +48,10 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `task-plan-review`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `task-plan-review`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill task-plan-review`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.
@@ -62,6 +63,8 @@ If `context_graph_required_action` is not `none`, the affected task is complete
 Applicable review gaps must use `../../contracts/quality.md` §Normalized Review Gaps. This skill
 must not maintain its own type-to-route mapping.
 
+Verify scoped coverage, reuse applicability and remaining obligations under quality.md Proportionate Tests And Evidence and Verification Effectiveness And Completion. Reuse valid results without replacing this review assessment; a later gate alone requires no recollection.
+
 ## Rules
 1. The Task Plan is the reference, not plausible code.
 2. No assumptions without evidence.

package/generated/.opencode/skills/agdf-tp-definition/SKILL.md

@@ -8,7 +8,8 @@ description: "Use this skill for this scope: bound TP content. Boundary: content
 ## Runtime Contract
 
 Use the returned `continuation.runtime_contracts` after `skill_continuation`. If absent, load
-`../../contracts/tp-definition.md` and `../../contracts/gate-artifact-preparation.md`
+`../../contracts/tp-definition.md`, `../../contracts/gate-artifact-preparation.md`
+and `../../contracts/quality.md`
 through the supplied contract reader or bundled files.
 Missing contract or invalid binding: stop. This skill owns TP content; gate-check owns readiness.
 
@@ -35,9 +36,10 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `tp-definition`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `tp-definition`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill tp-definition`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.

package/generated/.opencode/skills/agdf-ur-definition/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: agdf-ur-definition
-description: "Use this skill for this scope: bound UR drafting, clarification or revision. Boundary: content only; preserves approved intent. The requested effect, not discovery, decides AGDF activation."
+description: "Use this skill for this scope: UR drafting in planning scope or a bound Run. Boundary: content only; no start or approval authority. The requested effect, not discovery, decides AGDF activation."
 ---
 
 # ur-definition
@@ -34,9 +34,15 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `ur-definition`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `ur-definition`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+For deliberate pre-Run planning, pass the confirmed `planning` scope instead of `run_id`.
+Use `plan-<future mutation UUID>` and null expected digest for capture, or the inspected
+plan identity/current digest for refinement. No delivery intake/continue/revision fields.
+Follow only `phase: planned_ur_definition`, shared template/contract and planning-scoped typed
+writers. Missing compatible Core stops before mutation; no CLI/shell writer or Run-assignment fallback.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill ur-definition`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.
@@ -51,4 +57,6 @@ review timing in the UR; creation belongs to the later bound analytical step.
 Set the initial `Preview requirement` metadata according to the UR contract; a required
 interactive before-PRD preview uses `before_prd_interactive`, otherwise `none`.
 Write only the supplied unapproved UR, preserving uncertainty and explicit human approval.
-After canonical recording, redispatch gate-check; use its fresh presentation and wait.
+In Run mode, after canonical recording redispatch gate-check; use its fresh presentation and wait.
+In planning mode, report the saved draft, source and remaining questions without requesting a
+gate approval. Only a later explicit selected-source start request uses the Core handoff contract.

package/generated/.opencode/skills/agdf-ux-intent-definition/SKILL.md

@@ -40,9 +40,10 @@ Then pick one catalog route; non-authorizing, downstream checks remain.
 
 ## Executable Dispatch
 
-Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `ux-intent-definition`, `presentation_language`, `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
+Use listed MCP `agdf_dispatch` (load deferred; host prefix allowed): `skill_id` `ux-intent-definition`, `presentation_language` (optional override), `working_directory`, and only bound `target_source`/`primary_target` and `run_id`. Never search unlisted tools.
 If absent/failing: supplied schema-2 executable, child-only environment, immutable argv_prefix, declared arguments; `--skill ux-intent-definition`, language and working directory.
-For `--language`: Required presentation language for the latest natural-language user request as one well-formed BCP 47 tag. If the request explicitly asks for a response language, use that tag; otherwise use the dominant request language. Use en when mixed or ambiguous. A valid unsupported tag renders through the complete English pack. Missing or invalid input fails before governance evaluation.
+MCP: MCP reply-language override (BCP 47). Omit for a matching configured preference; otherwise supply the current request's language (explicit instruction, then dominant language, en if ambiguous). Invalid explicit tags fail; valid unsupported tags use the complete English pack. Core input remains complete.
+For `--language`: Required; apply the same current-request selection and locale validation.
 `target_source`: `explicit_target` if request names `primary_target`; `continued_target` if it unambiguously continues confirmed target; `current_repository` if request names this/current repo with one matching repo active. Otherwise omit the pair; cwd has no target authority.
 Quote shell values as data.
 For a result with `terminal: true`, the entire assistant response must consist only of host_action.text, copied verbatim. Add no question, explanation, heading, citation, link or other surrounding text; do not translate or reformat it; invoke no later tool and stop.
@@ -97,7 +98,7 @@ implementation. A requested change to the actual product still follows its exist
 #### Retain a run-bound preview
 
 For a preview used in this bound analysis, save a durable, reopenable copy under
-`.agdf/control/artefacts/<run_id>/previews/<short-name>-v<N>/index.html`, relative to the supplied
+`.agdf/control/runs/<run_id>/evidence/previews/<short-name>-v<N>/index.html`, relative to the supplied
 governance target. Derive the run from the valid assignment, not cwd, recency or another run.
 Resolve containment and reject symlink escapes before writing; never put preview content in
 `runs/<run_id>/RUN_STATE.md`. Use a new unused version for revisions and preserve reviewed versions.
@@ -175,7 +176,7 @@ Populate each field or use an explicit, justified `not_applicable`:
 
 For `blocked`, `blocking_reason`, `open_product_questions`, `affected_outputs`, `evidence`,
 `missing_evidence` and `required_next_step` are mandatory. When durable control state exists, write
-the result to `.agdf/control/artefacts/<run_id>/UX_INTENT_DEFINITION.md` without a Gate or approval field.
+the result to `.agdf/control/runs/<run_id>/documents/UX_INTENT_DEFINITION.md` without a Gate or approval field.
 
 ## Forbidden
 

package/generated/plugins/agdf/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agdf",
-  "version": "0.20.1",
+  "version": "0.30.0",
   "description": "AI Governance & Delivery Framework (AGDF) is a governance framework for agentic work. Its first implementation provides a control layer for governed AI-assisted software delivery. It combines workflow skills, explicit gates, control templates and machine-readable repository checks to keep work aligned with approved scope and make evidence a prerequisite for consequential gate transitions. Supported workflows include brownfield analysis, task-plan review, QA decisions, durable run state and auditable delivery closeout. AGDF is an independent open-source project, not a security sandbox, compliance certification or autonomous release system.",
   "keywords": [
     "agentic",
@@ -16,5 +16,6 @@
     "name": "Arndt Gold",
     "url": "https://agdf.iself.eu"
   },
-  "mcpServers": "./mcp/claude.mcp.json"
+  "mcpServers": "./mcp/claude.mcp.json",
+  "hooks": "./companion/hooks.json"
 }

package/generated/plugins/agdf/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agdf",
-  "version": "0.20.1",
+  "version": "0.30.0",
   "description": "Control layer for governed AI-assisted delivery.",
   "author": {
     "name": "Arndt Gold",

package/generated/plugins/agdf/companion/assignment.mjs

@@ -0,0 +1,65 @@
+// Run assignment per session (SDD-06, AC-004; chat-run navigation SDD-01 to SDD-04): the chat's run
+// (named in a main-loop turn or changed in this session, the later wins) unless the user browses,
+// else the remembered selection for this project, else none. Display only; it binds no AGDF delivery.
+
+// A run state without `updated_at` sorts last and never counts as changed in this session.
+const changedAt = run => (run.updatedAt ? Date.parse(run.updatedAt) : -Infinity);
+const newestFirst = (left, right) => changedAt(right) - changedAt(left) || left.id.localeCompare(right.id);
+
+// Only an active run is open. A superseded or abandoned run is closed like a completed one, but it
+// is never assigned: it has no fertig state to show.
+const ASSIGNABLE = new Set(['active', 'completed']);
+export const openRuns = runs => runs.filter(run => run.lifecycle === 'active').sort(newestFirst);
+
+// A run id is made of these characters; a mention counts only as a whole token. A dot ends a mention
+// unless an id character follows it, so a sentence may end with a run id (chat-run navigation SDD-01).
+const ID_CHAR = /[a-z0-9_-]/iu;
+const MENTION_WINDOW = 65_536;
+const bounded = (text, at, end) => !(at > 0 && (ID_CHAR.test(text[at - 1]) || text[at - 1] === '.'))
+  && !(end < text.length && (ID_CHAR.test(text[end]) || text[end] === '.' && ID_CHAR.test(text[end + 1] ?? '')));
+
+// The id of `ids` that a text names last as a whole token, or null; a long text is read from its end,
+// with the character before that window still bounding a mention (CR-F08).
+export function namedRun(text, ids) {
+  if (typeof text !== 'string' || !text) return null;
+  const from = Math.max(0, text.length - MENTION_WINDOW);
+  let named = null;
+  let namedEnd = -1;
+  for (const id of ids) {
+    if (!id) continue;
+    for (let at = text.indexOf(id, from); at >= 0; at = text.indexOf(id, at + 1)) {
+      const end = at + id.length;
+      if (end > namedEnd && bounded(text, at, end)) { named = id; namedEnd = end; }
+    }
+  }
+  return named;
+}
+
+// runs: parsed runs of the project; sessionStartedAt: host time (ms) at session.start;
+// selection: remembered run id or null; mention: { runId, at } of the last open run a main-loop turn
+// named, or null; isBrowsing: the user browsed away from the chat's run since the last main-loop turn.
+// Returns { run, origin, open, dropSelection, chatRunId }.
+// The chat's run is the later of the mention and the newest change in this session (chat-run
+// navigation SDD-02, SDD-03); it is shown unless the user browses to a remembered selection. A run
+// completed during this session stays assigned, so its fertig state remains visible.
+export function assignRun({ runs, sessionStartedAt, selection = null, mention = null, isBrowsing = false }) {
+  const open = openRuns(runs);
+  const changed = runs.filter(run => ASSIGNABLE.has(run.lifecycle) && changedAt(run) >= sessionStartedAt).sort(newestFirst);
+  const selected = selection ? open.find(run => run.id === selection) ?? null : null;
+  const dropSelection = Boolean(selection) && !selected;
+  const named = mention ? open.find(run => run.id === mention.runId) ?? null : null;
+  const chat = named && (!changed.length || mention.at >= changedAt(changed[0])) ? { run: named, origin: 'named' }
+    : changed.length ? { run: changed[0], origin: 'session' } : null;
+  const chatRunId = chat?.run.id ?? null;
+  if (chat && !(isBrowsing && selected)) return { run: chat.run, origin: chat.origin, open, dropSelection, chatRunId };
+  if (selected) return { run: selected, origin: 'selected', open, dropSelection, chatRunId };
+  return { run: null, origin: 'none', open, dropSelection, chatRunId };
+}
+
+// Steps through the open runs from the shown one (offset -1 or 1) for an explicit selection.
+export function stepSelection(open, currentId, offset) {
+  if (!open.length) return null;
+  const at = open.findIndex(run => run.id === currentId);
+  const from = at < 0 ? (offset > 0 ? -1 : 0) : at;
+  return open[(from + offset + open.length) % open.length].id;
+}

package/generated/plugins/agdf/companion/companion.json

@@ -0,0 +1,4 @@
+{
+  "schema_version": 1,
+  "hooks": "./companion/hooks.json"
+}

package/generated/plugins/agdf/companion/core.mjs

@@ -0,0 +1,16 @@
+import { RUN_MODES, RUN_ID_PATTERN, REVISION_ID_PATTERN } from '../runtime/create-agdf/runtime/core/lib/control-state/run-identity.js';
+// Bounded stored-fact projection only. No filesystem reads, seal qualification or authorization.
+import { parseControlState, recordRows, scalarFields, hasPersistedDerivedState } from '../runtime/create-agdf/runtime/core/lib/control-state/run-state-document.js';
+import { modeSliceDecision } from '../runtime/create-agdf/runtime/core/lib/control-evaluation/run-state-policy.js';
+import { createGatePolicy } from '../runtime/create-agdf/runtime/core/lib/control-evaluation/gate-policy-pure.js';
+const policy = createGatePolicy(content => ({ receipts: recordRows(content, 'Source Revisions') }));
+export function projectStoredRun(content) {
+  const fields = scalarFields(content), meta = Object.fromEntries(fields.values);
+  if (fields.duplicates.length || meta.control_state_version !== '4' || !RUN_MODES.includes(meta.mode) || !RUN_ID_PATTERN.test(meta.run_id ?? '') || !REVISION_ID_PATTERN.test(meta.revision_id ?? '') || hasPersistedDerivedState(content)) throw Error('format');
+  const facts = parseControlState(content);
+  const decisions = new Map();
+  for (const row of recordRows(content, 'Run Operations')) if (row.kind === 'presentation_decision.record' && row.effect.revision === Number(meta.revision))
+    decisions.set(row.effect.record.gate, { available: true, decided: true, action: row.effect.record.action });
+  const state = { ...facts, content, meta, presentation_decisions: decisions };
+  return { meta, facts, route: modeSliceDecision(state), control: policy.transitionDecisionForRunState(state) };
+}