agentera 3.0.0-dev.28 → 3.0.0-dev.29

package/README.md

@@ -38,15 +38,17 @@ Preview has no side effects. The apply path is the explicit, one-way v2-to-v3
 migration described in [UPGRADE.md](../../UPGRADE.md); it does not run a native
 package installer.
 
-Retired Claude cleanup is intentionally separate:
+Native Agentera resource cleanup is intentionally separate from host support:
 
 ```bash
-npx -y agentera@next upgrade --legacy-cleanup claude --dry-run
+npx -y agentera@next upgrade --legacy-cleanup claude.agentera-skill-link --dry-run
 ```
 
-It can remove only the exact Agentera-owned legacy link. Historical transcript
-import is also explicit (`agentera report refresh --import-source claude`) and
-is excluded from default active-runtime analytics.
+It selects declared native Agentera resources only. Each removal needs a
+matching whole-resource ledger identity and fingerprint; shared Codex config
+keys require unavailable key-level evidence and remain action-required.
+Historical transcript import is also explicit (`agentera report refresh
+--import-source claude`) and is excluded from default active-runtime analytics.
 
 ## Bounded state retrieval
 

package/bundle/CHANGELOG.md

@@ -4,6 +4,11 @@
 
 ### Changed
 
+- Health audit appends now receive a CLI-owned UTC timestamp. Bounded health
+  reads and `prime` select timestamped records newest-first regardless of random
+  ID; legacy records retain deterministic date-and-ID ordering without inferred
+  chronology.
+- Removed bundled native capability descriptors and `agentera check validate descriptors`; Codex setup no longer installs descriptors or writes agent-dispatch settings, and the canonical shared skill and CLI are the only active integration seam.
 - `agentera route receipt` now requires the phase-one `semantic_capsule_sha256` as well as `request_sha256`; it fails closed when the current trigger-intent capsule differs, preventing startup from a receipt classified against stale routing authority.
 - Defined the portable routing boundary: `agentera route request` deterministically resolves bare `/agentera`, explicit capability names or aliases, and curated leading phrases to a capability; every other request remains host-owned open-ended semantic classification. `agentera route receipt` validates the host result before authorizing startup, while `agentera route evaluate` and `agentera check verify eval routing` gate implementation with the frozen offline synthetic corpus; Agentera ships no built-in model service or live benchmark.
 - Added ordinary progress, decision, and health reads plus startup handling for
@@ -44,7 +49,7 @@
 
 - Added an executable bounded-retrieval authority and reserved CLI grammar for plan tasks, plans, and objective-scoped experiments, including stable and legacy identity rules, experiment `0`, opaque cursors, explicit omissions, and structured compatibility errors; retrieval execution remains staged for subsequent tasks.
 - Added one shared skill at `~/.agents/skills/agentera` plus the CLI and entity-backed project state as the active integration contract; host-native plugin, hook, agent, and package surfaces are not shipped.
-- Retained runtime lifecycle contracts only as migration evidence and for explicit `agentera upgrade --legacy-cleanup claude --dry-run|--yes`, which can remove only the ownership-proven legacy Agentera skill link and excludes user data.
+- Defined native-resource cleanup evidence for supported hosts and explicit `agentera upgrade --legacy-cleanup RESOURCE_ID --dry-run|--yes` selection. Cleanup retains only ledger-proven whole resources; Codex descriptor reports preserve shared configuration keys as report-only action-required items without durable key-level ownership.
 - Added typed artifact writes under `agentera state`: progress/decision/audit appends, decision satisfaction updates, plan create/task/status/archive lifecycle operations, schema-derived `explain`, server-assigned identifiers, exact retry replay, dry-run diffs, project locking, strict mapped-path safety, staged 10/40/50 compaction, final-byte validation, and atomic publication.
 - Added `install_track` to `prime --format json` (`app_home.install_track`) and `doctor --format json`. Values are `v2`, `v3`, `source`, and `unknown`.
 - Added the TypeScript CLI on the npm development channel (`npx -y agentera@next`) as a self-contained package whose governed isolated construction stages compiled commands, registry-owned app data, and the `.agentera-npx-bundle.json` sentinel before packing with lifecycle scripts disabled.
@@ -128,7 +133,7 @@
 - Removed `hej`, `describe`, and `gate` from top-level `agentera --help` and dispatch; unsupported structured requests return runnable corrections to `agentera prime`, `agentera schema`, and `agentera check compact`.
 - Removed the requirement for `uv` or a git checkout on the development channel; `uvx --from git+https://github.com/jgabor/agentera@main agentera` remains the stable 2.x Python line only.
 - Removed `agentera state migrate`, `agentera state backfill`, projection repair, and automatic v1 conversion. Recognized v2 aggregate projects now have one forward mutation path: full `agentera upgrade --yes`; `state migrate entities --dry-run` remains read-only.
-- Removed host-native runtime selectors, repair actions, package manifests, plugins, hooks, agents, and Bun single-binary delivery from the v3 distribution. Migration fixtures, ownership evidence, and explicit Claude cleanup remain internal migration surfaces.
+- Removed host-native runtime selectors, repair actions, package manifests, plugins, hooks, agents, and Bun single-binary delivery from the v3 distribution. Migration fixtures, ownership evidence, and explicit native Agentera resource cleanup remain internal migration surfaces.
 
 ### Migration notes
 

package/bundle/README.md

@@ -41,19 +41,20 @@ skill at `~/.agents/skills/agentera` plus the Agentera CLI. Normal `upgrade`
 previews and applies only app/project migration; it has no current-runtime
 selector and creates no native plugin, hook, agent, command, descriptor, or
 marketplace file. See [UPGRADE.md](./UPGRADE.md) for the distinct one-way v2
-migration and explicit retired Claude cleanup route:
-`agentera upgrade --legacy-cleanup claude --dry-run|--yes`.
+migration and explicit native-resource cleanup route:
+`agentera upgrade --legacy-cleanup RESOURCE_ID --dry-run|--yes`.
 
 `doctor` reports read-only app, project-state, shared-skill, and CLI evidence.
 `prime`, status, and project-integration output use the app/project recommendation
 and shared-skill diagnosis to select the next action.
 
-Claude Code is retired from active support. Its only lifecycle command is a
-separate, explicit cleanup of the exact Agentera-owned legacy skill link:
+Supported hosts continue to use the canonical shared skill and CLI. Cleanup
+selects a native Agentera resource, never a host. Claude Code is retired from
+active support; its legacy link remains one selectable resource:
 
 ```bash
-npx -y agentera@next upgrade --legacy-cleanup claude --dry-run
-npx -y agentera@next upgrade --legacy-cleanup claude --yes
+npx -y agentera@next upgrade --legacy-cleanup claude.agentera-skill-link --dry-run
+npx -y agentera@next upgrade --legacy-cleanup claude.agentera-skill-link --yes
 ```
 
 Historical Claude transcripts are excluded by default. A local import requires

package/bundle/UPGRADE.md

@@ -1,8 +1,8 @@
 # Upgrade and migration
 
 Agentera 3.0 uses one upgrade command for app/project migration and explicit
-retired Claude cleanup. Preview and apply are separate operations. Current
-runtime installation and repair are not upgrade operations.
+native Agentera resource cleanup. Preview and apply are separate operations.
+Current runtime installation and repair are not upgrade operations.
 
 Before the npm stable dist-tag promotion, use `agentera@next`. The stable 2.x
 channel remains `agentera@latest` until that promotion; upgrade does not publish
@@ -24,8 +24,8 @@ runtime package managers, authentication, enablement, or trust operations.
 
 Passing `--runtime` fails before mutation. Remove `--runtime`, ensure the runtime
 can read `~/.agents/skills/agentera`, and invoke the CLI directly. The supported
-v2 migration and Claude cleanup routes below are separate from normal active
-integration.
+v2 migration and native Agentera resource cleanup routes below are separate from
+normal active integration.
 
 ## Preview and apply
 
@@ -55,20 +55,30 @@ Retries re-observe the current app/project state. Completed migration work
 converges to no change; interrupted v2 migration continues through the same full
 apply command.
 
-## Retired Claude cleanup
+## Native Agentera resource cleanup
 
-Claude cleanup is a distinct historical route, not a current-runtime selection:
+Cleanup is a distinct resource route, not a host selection. Codex, Cursor,
+OpenCode, and Copilot remain supported through the canonical shared skill and
+CLI. Accepted smoke evidence records that Codex and Cursor loaded Agentera's
+skill instructions, OpenCode listed the canonical skill, and Copilot's listed
+canonical skill is intentionally disabled.
+
+Use `agentera upgrade --legacy-cleanup RESOURCE_ID --dry-run|--yes` with one
+declared native Agentera resource ID.
 
 ```bash
-npx -y agentera@next upgrade --legacy-cleanup claude --dry-run
-npx -y agentera@next upgrade --legacy-cleanup claude --yes
+npx -y agentera@next upgrade --legacy-cleanup claude.agentera-skill-link --dry-run
+npx -y agentera@next upgrade --legacy-cleanup claude.agentera-skill-link --yes
 ```
 
-Cleanup is limited to the exact Agentera-owned legacy skill link recorded by the
-ownership journal. It never removes Claude projects, transcripts, settings,
-credentials, caches, or other user data. Ambiguous or unowned resources remain
-blocked. Preview is read-only; apply is idempotent and can be retried with the
-same explicit selection.
+The contract also inventories retired Codex descriptor files. Each file needs a
+matching whole-resource ledger identity and fingerprint before it can be
+removed. Codex descriptor previews report each shared configuration key as
+`action_required` without adding it to the selected resource's apply work.
+Agentera has no durable key-level ownership proof, so matching values, markers,
+names, or whole file contents never authorize a shared `config.toml` mutation.
+Preview is read-only; approved apply is idempotent and preserves ambiguous or
+unowned resources.
 
 Historical transcript import is independent of cleanup. Default extraction and
 analytics do not read Claude history. Import requires explicit local consent:
@@ -164,7 +174,7 @@ npx -y agentera@next prime --format json
 
 `upgrade --verify` without `--yes` is a read-only doctor and capability-context
 verification. With `--yes`, it runs those checks after approved migration.
-Explicit Claude cleanup with `--verify` requires `--yes`.
+Explicit native Agentera resource cleanup with `--verify` requires `--yes`.
 
 `--force` applies only where the owning migration contract explicitly permits
 replacement. It never adopts user-owned resources.
@@ -184,7 +194,8 @@ node packages/cli/dist/bin/agentera.js check validate release-metadata
 ## Mutation ownership
 
 Upgrade apply holds the required project lock before migration begins. Explicit
-Claude cleanup additionally uses its ownership journal and shared lock. Lock
-creation is atomic, and release succeeds only for the matching ownership token.
+native Agentera resource cleanup additionally uses its ownership journal and
+shared lock. Lock creation is atomic, and release succeeds only for the matching
+ownership token.
 A stale or malformed lock stops the operation; inspect the named lock file,
 remove only that file after confirming no operation owns it, then rerun.

package/bundle/references/adapters/package-surface-characterization.md

@@ -31,8 +31,7 @@ by [`docs/packaging/v3-packaging.md`](../../docs/packaging/v3-packaging.md).
 The self-contained npm package ships `dist/` plus `bundle/`. Bundle data
 includes the canonical `skills/` tree, required `references/`, `registry.json`,
 the frozen routing evaluation corpus, and package guidance. Compiled commands
-ship in `dist/`. The shared-skill tree includes its capability schemas, artifact schemas, and canonical
-`skills/agentera/agents/*.toml` descriptors.
+ship in `dist/`. The shared-skill tree includes its capability schemas and artifact schemas.
 
 The package lane's isolated package fixture invokes `npm pack --json
 --ignore-scripts` after governed construction and fails if the CLI or required

package/bundle/references/adapters/runtime-lifecycle-authority.yaml

@@ -3,7 +3,7 @@ status: migration_only_authority
 decision: 92
 operation_contract: references/adapters/runtime-lifecycle-operation-contract.yaml
 adapter_contract: references/adapters/runtime-lifecycle-adapters.yaml
-retired_cleanup_contract: references/adapters/runtime-retired-resources.yaml
+native_resource_cleanup_contract: references/adapters/runtime-retired-resources.yaml
 canonical_skill:
   path: ~/.agents/skills/agentera
   required_for_support_floor: true
@@ -52,11 +52,10 @@ support_floor:
     not_applicable_scope: unobserved_conditional_surface_only
   unmet_blocks_release: true
 active_runtimes: []
-retired_runtime_inputs:
+historical_inputs:
   - id: claude
     purposes:
       - consent_gated_historical_import
-      - owned_legacy_resource_cleanup
 migration_aliases:
   cursor-agent:
     runtime_id: cursor

package/bundle/references/adapters/runtime-retired-resources.yaml

@@ -1,25 +1,58 @@
-schema_version: agentera.retiredRuntimeResources.v1
-status: retired_migration_contract
-decision: 92
+schema_version: agentera.nativeResourceCleanup.v1
+status: resource_retirement_contract
+decision: nksvqmnevm
 authority: references/adapters/runtime-lifecycle-authority.yaml
 operation_contract: references/adapters/runtime-lifecycle-operation-contract.yaml
 policy:
-  active_inventory_exposure: forbidden
+  host_inventory_exposure: evidence_only
+  selection: native_agentera_resource_only
   preview: strictly_read_only
   apply_requires: explicit_approval
-  ownership: matching_whole_resource_legacy_ledger_identity_and_fingerprint
+  ownership: matching_whole_resource_ledger_identity_and_fingerprint
+  shared_configuration: action_required_without_key_level_ownership
+  forbidden_ownership_evidence:
+    - value_equality
+    - managed_marker
+    - resource_name
+    - file_equality
   unsupported_platform_result: action_required
-retired_runtimes:
+accepted_host_evidence:
+  - host: codex
+    state: supported
+    proof: Codex loaded Agentera's specific canonical skill instructions and bootstrap command.
+  - host: cursor
+    state: supported
+    proof: Cursor loaded Agentera's specific canonical skill instructions and bootstrap command.
+  - host: opencode
+    state: supported
+    proof: OpenCode's native CLI listed /home/jgabor/.agents/skills/agentera/SKILL.md.
+  - host: copilot
+    state: supported_disabled
+    proof: Copilot's native CLI listed the canonical personal-agents skill; its disabled state is intentional.
+hosts:
+  - id: codex
+    display_name: Codex
+    support_status: supported
+  - id: cursor
+    display_name: Cursor
+    support_status: supported
+  - id: opencode
+    display_name: OpenCode
+    support_status: supported
+  - id: copilot
+    display_name: Copilot
+    support_status: supported_disabled
   - id: claude
     display_name: Claude Code historical migration
-    active_runtime: false
-    source_product: claude-code
-    resources:
-      - id: claude.agentera-skill-link
-        kind: symlink
-        intent: remove
-        destination: "{home}/.claude/skills/agentera"
-        ledger_status: legacy
+    support_status: retired_historical
+resources:
+  - id: claude.agentera-skill-link
+    host: claude
+    kind: symlink
+    intent: remove
+    destination: "{home}/.claude/skills/agentera"
+    ledger_status: legacy
+    durable_proof: matching whole-resource legacy ledger identity and fingerprint
     never_touch:
       - "{home}/.claude/projects"
       - "{home}/.claude/statsig"
@@ -31,3 +64,34 @@ retired_runtimes:
       - "{home}/.claude/conversations"
       - "{home}/.claude/credentials"
     safety_note: Only the declared Agentera skill symlink may be removed, and only when a matching legacy ownership ledger still binds its identity and target fingerprint. Transcript, settings, credential, conversation, stats, cache, and ambiguous paths are never cleanup targets.
+  - id: codex.agent-descriptor
+    host: codex
+    kind: file
+    intent: remove
+    destination: "{home}/.codex/agents/{descriptor}.toml"
+    descriptors: [status, vision, discuss, research, plan, build, optimize, audit, document, profile, design, orchestrate]
+    ledger_status: managed
+    durable_proof: matching whole-resource ledger identity and fingerprint for each declared descriptor file
+    safety_note: Each descriptor is an independently owned file. A managed marker, descriptor name, or matching content never establishes ownership.
+configuration_inventory:
+  - id: codex.config.shell_environment_policy.set.AGENTERA_HOME
+    host: codex
+    destination: "{home}/.codex/config.toml"
+    key: shell_environment_policy.set.AGENTERA_HOME
+    durable_proof: key-level ownership ledger identity and fingerprint
+    ownership_available: false
+    result_without_proof: action_required
+  - id: codex.config.agents.max_depth
+    host: codex
+    destination: "{home}/.codex/config.toml"
+    key: agents.max_depth
+    durable_proof: key-level ownership ledger identity and fingerprint
+    ownership_available: false
+    result_without_proof: action_required
+  - id: codex.config.features.multi_agent_v2
+    host: codex
+    destination: "{home}/.codex/config.toml"
+    key: features.multi_agent_v2
+    durable_proof: key-level ownership ledger identity and fingerprint
+    ownership_available: false
+    result_without_proof: action_required

package/bundle/references/artifacts/state-storage-authority.yaml

@@ -298,10 +298,19 @@ entity_target:
         required_fields: [date, dimensions, findings_summary, trajectory, grades]
         required_paths: []
         forbidden_fields: [id, artifact, number, stable_id, artifact_id, entry_number]
+        cli_owned_append_fields: [appended_at]
+        legacy_optional_fields: [appended_at]
+        field_contract:
+          appended_at:
+            format: canonical_utc_iso8601
+            assignment: cli_writer_at_append
+            compatibility: >-
+              Legacy canonical health audits omit appended_at and remain valid without
+              reconstructed chronology.
       retrieval:
         exact: agentera state health get --id ID --format json
         list: agentera state health list [--limit N] [--cursor TOKEN] --format json
-        ordering: date_desc_then_id_asc
+        ordering: appended_at_desc_then_id_asc_then_legacy_date_desc_then_id_asc
         default_limit: 20
         maximum_limit: 100
         max_utf8_bytes: 32768

package/bundle/references/cli/audience-namespace-cli-migration.yaml

@@ -174,7 +174,7 @@ audience_namespaces:
       validate:
         canonical_shape: agentera check validate <family> [target options]
         delegates_from: validate
-        notes: Preserves existing validate families (capability, artifact, descriptors, etc.).
+        notes: Preserves current validate families including capability and artifact validation.
       verify:
         canonical_shape: agentera check verify <family> <target> [options]
         delegates_from: verify

package/bundle/references/cli/bundle-skill-vocabulary.yaml

@@ -2,7 +2,7 @@ schema_version: agentera.bundle_skill_vocabulary.v1
 status: active_authority
 purpose: >-
   Classify surviving `bundle` and `SKILL.md` terminology so docs, package
-  metadata, runtime descriptors, tests, historical records, and compatibility
+  metadata, tests, historical records, and compatibility
   identifiers share one bounded vocabulary source without forcing mechanical
   renames of shipped surfaces.
 sources:

package/bundle/references/cli/capability-instruction-contract.yaml

@@ -87,8 +87,6 @@ implementation_state:
   first_invocation_read_cli_metadata: true
   first_invocation_read_schema_metadata: true
   runtime_first_invocation_behavior: true
-  descriptors_rewritten_to_prime_command: true
-  package_metadata_rewritten_to_prime_command: true
 compatibility_preservation:
   preserve_legacy_references_only_when_classified:
     - skills/agentera/capabilities/*/instructions.md paths (historical)
@@ -111,7 +109,6 @@ surface_inventory:
   docs_vocabulary: delegates_to_this_authority
   readme: concise_user_facing_current_and_future_authoring_seam
   capability_schema_contract: current_canonical_validator_reference
-  runtime_descriptors: current_canonical_instructions_md_pointers
   capability_instruction_files: current_canonical_instruction_files
 evaluator_handoff:
   status: implemented

package/bundle/references/cli/routing-execution-vocabulary.yaml

@@ -237,10 +237,6 @@ compatibility_identifiers:
   route_aliases:
     - ROUTE_ALIASES
     - primary_aliases
-  command_and_schema_surfaces:
-    - agentera validate descriptors
-    - descriptor_sources
-    - invocation_pattern
 code_identifiers:
   - subagent_dispatch
   - test_runtime_adapter_subagent_dispatch_metadata_is_complete

package/bundle/references/cli/vocabulary.md

@@ -100,8 +100,7 @@ Use this prose as guidance only: the default is that the first capability
 invocation shells out to `agentera prime --context <name> --format json` and
 reads the returned `instructions` field. Today, capability directories carry
 the `schemas/` files only, the instructions module lives at
-`packages/cli/src/capabilities/<name>/instructions.ts`, runtime descriptors
-invoke the prime command, and the `agentera prime --context <name> --format
+`packages/cli/src/capabilities/<name>/instructions.ts`, and the `agentera prime --context <name> --format
 json` response emits `capability_context.instructions` plus
 `first_invocation_read` metadata.
 

package/bundle/skills/agentera/capabilities/orchestrate/schemas/artifacts.yaml

@@ -23,13 +23,17 @@ ARTIFACTS:
   3:
     id: A3
     artifact: health
-    local_role: consumes
+    local_role: produces_and_consumes
     description: >-
-      Orchestrate consumes returned health state first after plan completion to
-      decide whether to start a new plan cycle; bounded `agentera state health
-      list --limit 20 --format json` and exact `state health get --id ID`
-      are the fallback contracts. Raw reads are diagnostic-only for corruption
-      or CLI-defect investigation.
+      For an active plan whose tasks are all terminal, Orchestrate consumes
+      returned health state and delegates a limited staleness closure check.
+      On PASS it publishes one typed `artifact_freshness` health record through
+      `agentera state health append --input PATH --format json` before setting
+      the still-open plan complete. WARN or FAIL leaves the plan open and
+      surfaces the finding. Bounded `agentera state health list --limit 20
+      --format json` and exact `state health get --id ID` are fallback
+      contracts. Raw reads are diagnostic-only for corruption or CLI-defect
+      investigation.
   4:
     id: A4
     artifact: todo

package/bundle/skills/agentera/capabilities/orchestrate/schemas/exit.yaml

@@ -3,8 +3,9 @@ EXIT_CONDITIONS:
     id: E1
     condition: complete
     description: >-
-      All plan tasks are complete or validly superseded, the health check passed, and the
-      orchestration session concluded with all planned work finished.
+      All plan tasks are complete or validly superseded, the limited post-plan
+      health check passed, its typed health record was published, and the
+      still-open plan was then set complete without invoking archive.
     exit_signal: complete
   2:
     id: E2

package/bundle/skills/agentera/capabilities/orchestrate/schemas/validation.yaml

@@ -46,14 +46,15 @@ VALIDATION:
     id: V5
     rule: orchestrator_read_only_artifacts
     description: >-
-      The orchestrator reads artifacts and updates plan task statuses and
-      TODO.md blocked entries. It does NOT write to progress, changelog,
-      or any other capability-owned artifact. This rule verifies the
-      artifacts schema reflects this constraint.
+      The orchestrator updates plan task statuses and TODO.md blocked entries.
+      During terminal-open closure only, it writes the typed limited-scope
+      health record before plan completion. It does NOT write progress or
+      changelog. This rule verifies the artifacts schema reflects this
+      constraint.
     severity: warning
     checks:
       - "progress is consumes only (not produces)"
-      - "health is consumes only (not produces)"
+      - "health is produces_and_consumes for terminal-open closure only"
       - "decisions is consumes only (not produces)"
       - "vision is consumes only (not produces)"
   6:

package/bundle/skills/agentera/schemas/artifacts/health.yaml

@@ -1,9 +1,10 @@
 # Health Artifact Schema
 #
-# Audit log maintained by audit. Each audit entry records assessed
-# dimensions, findings (critical/warning/info counts), overall trajectory,
-# per-dimension grades (A-F), per-dimension findings with confidence, and
-# trend analysis against the prior audit.
+# Audit normally owns health records. Orchestrate may append only the limited
+# artifact_freshness record for terminal-plan closure after Audit passes. Each
+# audit entry records assessed dimensions, findings (critical/warning/info
+# counts), overall trajectory, per-dimension grades (A-F), per-dimension
+# findings with confidence, and trend analysis against the prior audit.
 #
 # The active authority is one health_audit entity per writer-owned file. The
 # numbered projection model below is retained only as an exact legacy migration
@@ -15,6 +16,7 @@
 # ─────────────────────────────────────────────────────────────────────
 # ## Audit N · date                     AUDIT.number           HA1
 # (date from heading)                   AUDIT.date             HA2
+# (CLI append time)                     AUDIT.appended_at      HA7
 # **Dimensions assessed**: <list>       AUDIT.dimensions       HA3
 # **Findings**: N crit, N warn, N info  AUDIT.findings_summary HA4
 # **Overall trajectory**: <text>        AUDIT.trajectory       HA5
@@ -50,12 +52,13 @@ meta:
   name: health
   version: "1.0.0"
   description: >-
-    Audit log maintained by audit. Consumed by capabilities that need
-    codebase health context.
+    Audit normally owns health records. Orchestrate may append only the
+    limited artifact_freshness record for terminal-plan closure after Audit
+    passes. Consumed by capabilities that need codebase health context.
   artifact_type: agent_facing
   path: ".agentera/health.yaml"
   authority: legacy_migration_input_only
-  producer: audit
+  producer: [audit, orchestrate]
   consumers: [document, status, audit, orchestrate, plan, build, vision]
   format: yaml
 
@@ -238,6 +241,21 @@ AUDIT:
     validation:
       - "Every assessed dimension has a grade"
       - "Grades are single characters A through F"
+  7:
+    id: HA7
+    field: appended_at
+    type: string
+    format: "YYYY-MM-DDTHH:mm:ss.sssZ"
+    required: false
+    cli_owned: true
+    description: >-
+      Canonical UTC timestamp assigned by the health append writer. Full records
+      with this field sort newest-first by append time, then stable ID. Legacy
+      records without it retain date-then-ID ordering; migration never fabricates
+      append chronology.
+    validation:
+      - "When present, must be canonical UTC ISO-8601 with millisecond precision"
+      - "Must be omitted from caller input and assigned by the CLI writer"
 
 DIMENSION:
   1:

package/bundle/skills/agentera/schemas/artifacts/plan.yaml

@@ -645,7 +645,6 @@ LIFECYCLE_CONTRACT:
       - packages/cli/src/capabilities/*/instructions.ts
       - packages/cli/README.md
       - skills/agentera/SKILL.md
-      - skills/agentera/agents/*
       - skills/agentera/references/*
       - references/**/*.md
       - docs/**