diff --git a/AGENT-SPEC.md b/AGENT-SPEC.md
index 366f526..7284702 100644
--- a/AGENT-SPEC.md
+++ b/AGENT-SPEC.md
@@ -245,6 +245,85 @@ task named `agent`. Network startup materializes active declarations for the sel
 reconciliation; retired and other-host declarations are skipped. A gating failure suppresses only that agent.
 Prefer a catalog-owned `.st2/` overlay and locally excluded tool loaders.
 
+## Experimental read-only plans
+
+This candidate contract is not part of the released `0fed14b` corpus
+conformance claim above. The paired model-free experiment is isolated to
+[`catalog-plan-vs-direct-brief`](cells/catalog-plan-vs-direct-brief/) and runs
+against st2 draft PR
+[#115](https://github.com/compoundingtech/st2/pull/115) at exact current-main source
+`8a76b6e71355140e5b89cd9313fcfd88c82b5cad`.
+
+The experiment shape descends from
+[source sketch revision `5c1d142`](https://gist.github.com/myobie/d5ecfac24cd3965e095a5031cd2e00cb/5c1d1427c0556d95d13890e5c5086cd85b25d994);
+the executable discovery and schema authority is the exact PR115 source above.
+An agent discovers a plan only through a childless Resource link:
+
+```kdl
+agent "app-web" {
+  resource "ship-remote-approvals" _tag="plan" uri="file:plans/ship-remote-approvals/plan.kdl"
+}
+```
+
+The positional Resource name is an agent-local role. The referenced
+`plan.kdl`, not the Resource envelope, owns all plan truth:
+
+```kdl
+plan "ship-remote-approvals" {
+  owner "app-web"
+  version "0000" content="file:versions/0000.md"
+  version "0001" content="file:versions/0001.md" {
+    parent "0000"
+    why "Browser proof exposed an approval race."
+  }
+}
+```
+
+Instead of external Markdown content, a version may keep its complete intent
+inline in the same referenced `plan.kdl`:
+
+```kdl
+plan "review-follow-up" {
+  owner "app-web"
+  version "0000" {
+    intent "Review every unresolved comment and report the exact final head."
+  }
+}
+```
+
+Plan identity is the explicit KDL value, never its directory. External plans
+require one `owner`. Each version requires exactly one `content="file:..."` or
+one child `intent`; both or neither fail. A version may repeat `parent` for
+multiple declared parents; parents must exist, be unique, and form an acyclic
+graph. A version with parents requires one non-empty `why`. Versions are sorted,
+and the frontier is derived as every version with no child, retaining concurrent
+siblings. Resource URIs resolve relative to the agent KDL, while content URIs
+resolve relative to the referenced `plan.kdl`; both must be relative `file:`
+references to regular files inside the selected catalog. The Resource adds the
+agent to `referencedBy` but owns no plan fields. Legacy `plan-ref`, childful plan
+Resources, and agent-owned inline plan truth are unsupported. The experiment
+stores no content digest or history. It cannot prove that an earlier declaration
+or content file stayed unchanged.
+
+The only supported CLI is read-only:
+
+```sh
+st2 plan validate [PATH] [--json]
+st2 plan list [PATH] [--json]
+st2 plan show <identity> [PATH] [--json]
+st2 plan inspect <identity> [PATH] [--json]
+```
+
+Global `--catalog` selects the same input. The experiment has no current
+pointer, execution, controller, schedule, step graph, retry, claim, receipt,
+event, reconciliation mutation, agent restart, CAS, merge, or deployment
+behavior. Direct KDL and direct human-to-agent planning remain supported.
+The current model-free A/B finds no plan advantage for cold resume, exact
+intent recovery, or acceptance evidence: both durable paths recover equally,
+and neither surface reports worker acceptance. Static validation and resolved
+provenance are plan-only authoring evidence; live correctness, traffic, cost,
+and duration remain unresolved.
+
 ## Validation, health, and lifecycle
 
 Canonical validation is:
diff --git a/CATALOG.md b/CATALOG.md
index a4f1270..a1ceef1 100644
--- a/CATALOG.md
+++ b/CATALOG.md
@@ -27,6 +27,7 @@ harness-native loader plus canonical hook file.
 | `assignment-contract-hot-assignment` | Codex | `gpt-5.6-sol` / medium | 2 | medium | `1200s` | 4 | — | **NO STRUCTURED RUN** |
 | `assignment-contract-hot-focus` | Codex | `gpt-5.6-sol` / medium | 2 | medium | `1200s` | 4 | — | **NO STRUCTURED RUN** |
 | `assignment-contract-hot-resources` | Codex | `gpt-5.6-sol` / medium | 2 | medium | `1200s` | 4 | — | **NO STRUCTURED RUN** |
+| `catalog-plan-vs-direct-brief` | model-free | — | 0 | none | `60s` | 22 | — | **NO STRUCTURED RUN** |
 | `context-resource-continuity` | model-free | — | 0 | none | `90s` | 4 | — | **NO STRUCTURED RUN** |
 | `crash-ding` | mixed | `claude-sonnet-5+gpt-5.6-sol` / medium | 4 | high | `180s` | 5 | — | **NO STRUCTURED RUN** |
 | `ding-mode` | Claude | `claude-sonnet-5` / medium | 2 | medium | `1200s` | 4 | — | **NO STRUCTURED RUN** |
@@ -95,6 +96,7 @@ while the last-run column makes a recorded failure distinct from a cell with no
 
 ## Harness hook exclusions
 
+- `catalog-plan-vs-direct-brief` / `cell` — deterministic plan-versus-brief fixture and recovery contract; no harness seat
 - `adopt-only-migration` / `cell` — deterministic native adoption/replacement lifecycle probe; no harness seat
 - `agent-spec-resource-bindings` / `cell` — deterministic native Agent Spec Resource-envelope probe; no harness seat
 - `context-resource-continuity` / `cell` — deterministic native context/resource restart probe; no harness seat
diff --git a/cells/catalog-plan-vs-direct-brief/README.md b/cells/catalog-plan-vs-direct-brief/README.md
new file mode 100644
index 0000000..0ef697f
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/README.md
@@ -0,0 +1,65 @@
+# catalog-plan-vs-direct-brief
+
+Model-free contract for the first experiment proposed by
+[`st2 plans: today → dream`](https://gist.github.com/myobie/d5ecfac24cd3965e095a5031cd2e00cb/5c1d1427c0556d95d13890e5c5086cd85b25d994):
+compare one realistic repository task delivered as a versioned local catalog
+plan (A) with the same task delivered as an ordinary durable direct brief (B).
+
+The frozen task, repository, tools, budgets, intent revisions, judges, and done
+condition are identical. The two intent documents are byte-identical at each
+revision. Both arms receive durable local input and use the same evaluator-owned
+receipt fields, so direct planning remains a valid control rather than an
+intentionally fragile baseline.
+
+The deterministic fixture now runs both durable product paths:
+
+- the seeded repository needs a real implementation and the arm-neutral
+  reference solution passes public and held-out correctness tests;
+- arm A recovers an initial and then revised catalog snapshot through exact
+  `st2 plan show`/`inspect` calls;
+- arm B receives the byte-identical revisions through real isolated
+  `st2 message send` deliveries with `inReplyTo` lineage, then recovers them
+  through `message ls`/`read`;
+- both arms recover the same intent after cold state loss with their remote
+  source offline, so cold resume, intent recovery, and steering are ties;
+- neither read-only plan inspection nor direct-message delivery reports worker
+  acceptance, so acceptance evidence is evaluator-owned in both arms and is
+  also a tie;
+- the plan adds native static validation and resolved provenance, but the
+  current model-free result is `no-measured-advantage`; and
+- no model or provider is launched.
+
+Correctness, coordination traffic, token use, cost, and wall duration remain
+unresolved live-run endpoints. The reference solution establishes a valid task
+and neutral judge; it is not substituted for two agent executions.
+
+Arm A uses the exact Resource-linked external `plan.kdl` contract from
+[st2 draft PR #115](https://github.com/compoundingtech/st2/pull/115) at source
+`8a76b6e71355140e5b89cd9313fcfd88c82b5cad`, while retaining the experiment
+shape from source gist revision
+`5c1d1427c0556d95d13890e5c5086cd85b25d994`. A childless Agent Spec Resource
+with `_tag="plan"` supplies only the agent-local role and source-relative file
+link. The referenced `plan.kdl` owns the plan identity, owner, versions, and
+intent truth. The comparison path keeps its external Markdown content, and one
+focused second target proves the supported inline-intent form without changing
+the A/B scenario.
+
+The accepted Linux artifact SHA256 is
+`214e08874720bc546d4adf7d7977e614237baf7989cc09f6932cd991f497a753`;
+hosted Nix run
+[30835684680](https://github.com/compoundingtech/st2/actions/runs/30835684680),
+job `91760161352`, passed. The fixture exercises only `plan validate`, `list`,
+`show`, and `inspect`, and proves they do not alter the catalog. Legacy
+`plan-ref` and agent-owned inline plan truth are not used.
+
+This plan model stores no content digest or history, so it cannot prove that an
+earlier declaration, parent link, or content file stayed unchanged.
+
+Arm B uses the product's ordinary durable message store rather than a synthetic
+thread sidecar. Neither arm receives a pre-authored acceptance receipt. The
+experiment uses plain copied folders, requires no server, and sets
+`casRequired` to false. The product experiment adds no current pointer,
+execution, scheduling, steps, retries, progress claims, events,
+reconciliation, or CAS. A provider-backed A/B remains a separate authorization
+with an exact model, effort, budget, run order, spend ceiling, cleanup, and
+rollback.
diff --git a/cells/catalog-plan-vs-direct-brief/catalog-plan-vs-direct-brief.kdl b/cells/catalog-plan-vs-direct-brief/catalog-plan-vs-direct-brief.kdl
new file mode 100644
index 0000000..e48d310
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/catalog-plan-vs-direct-brief.kdl
@@ -0,0 +1,81 @@
+// Neutral, model-free fixture contract for a catalog-plan versus direct-brief A/B.
+eval {
+  copy "./fixture"
+  max-timeout "60s"
+
+  run "fixture-contract" {
+    command "bash ./check-fixtures.sh"
+  }
+  run "plan-contract" {
+    command "bash ./check-plan-contract.sh"
+  }
+
+  judges {
+    judge "TASK — frozen repository starts red and its arm-neutral reference implementation passes" {
+      exec "grep -Fqx TASK-CONTRACT-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "PARITY — task intent, tools, budgets, judges, and done condition match across arms" {
+      exec "grep -Fqx ARM-PARITY-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "RESTART — catalog plan and real direct-message inbox both recover after cold state loss" {
+      exec "grep -Fqx COLD-RESTART-TIE-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "PARTITION — both arms recover locally while their remote sources are unavailable" {
+      exec "grep -Fqx LOCAL-PARTITION-TIE-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "INTENT — both arms recover byte-identical initial and revised intent" {
+      exec "grep -Fqx INTENT-RECOVERY-TIE-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "STEERING — a complete plan version and reply-linked direct brief recover equally" {
+      exec "grep -Fqx HUMAN-STEERING-TIE-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "ACCEPTANCE — neither read-only plan inspection nor message delivery proves worker acceptance" {
+      exec "grep -Fqx ACCEPTANCE-EVIDENCE-TIE-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "STATIC VALUE — plan validation and provenance are plan-only authoring evidence" {
+      exec "grep -Fqx PLAN-STATIC-EVIDENCE-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "DIRECT REALISM — the control uses durable st2 messages and reply lineage" {
+      exec "grep -Fqx DIRECT-MESSAGE-EVIDENCE-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "MEASUREMENT — correctness gates are separate from neutral traffic and cost observations" {
+      exec "grep -Fqx MEASUREMENT-CONTRACT-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "PROVENANCE — every authoritative input has a verified content hash" {
+      exec "grep -Fqx PROVENANCE-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "PAIRING — exact experimental source and artifact are frozen while live execution stays blocked" {
+      exec "grep -Fqx PRODUCT-PAIRING-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "STORAGE — plain local folders suffice and CAS is not required" {
+      exec "grep -Fqx PLAIN-FOLDER-NO-CAS-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "VERDICT — the model-free endpoints show no plan advantage and live endpoints stay unresolved" {
+      exec "grep -Fqx MODEL-FREE-OUTCOME-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+    judge "RUNNER — the exact experimental st2 binary matches the accepted source and SHA256" {
+      exec "grep -Fqx PLAN-RUNNER-PIN-GREEN-43af $RUNS_DIR/plan-contract.out"
+    }
+    judge "PLAN VALIDATE — Resource-linked external plans normalize without errors" {
+      exec "grep -Fqx PLAN-VALIDATE-GREEN-43af $RUNS_DIR/plan-contract.out"
+    }
+    judge "PLAN TARGETS — external Markdown content and inline intent are both exact" {
+      exec "grep -Fqx PLAN-TARGET-FORMS-GREEN-43af $RUNS_DIR/plan-contract.out"
+    }
+    judge "PLAN LIST — explicit identity, owner, and derived frontier are exact" {
+      exec "grep -Fqx PLAN-LIST-GREEN-43af $RUNS_DIR/plan-contract.out"
+    }
+    judge "PLAN SHOW — normalized intent exposes versions but not provenance paths" {
+      exec "grep -Fqx PLAN-SHOW-GREEN-43af $RUNS_DIR/plan-contract.out"
+    }
+    judge "PLAN INSPECT — provenance and resolved content are exact and read-only" {
+      exec "grep -Fqx PLAN-INSPECT-READONLY-GREEN-43af $RUNS_DIR/plan-contract.out"
+    }
+    judge "PLAN BOUNDARY — a mutable current pointer is rejected with a classified error" {
+      exec "grep -Fqx PLAN-BOUNDARY-GREEN-43af $RUNS_DIR/plan-contract.out"
+    }
+    judge "HERMETIC — the scaffold launches no model, provider, network client, or mutating product runtime" {
+      exec "grep -Fqx HERMETIC-SCAFFOLD-GREEN-43af $RUNS_DIR/fixture-contract.out"
+    }
+  }
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/agent.kdl b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/agent.kdl
new file mode 100644
index 0000000..66133e0
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/agent.kdl
@@ -0,0 +1,6 @@
+agent "receipt-worker" {
+  host "eval"
+  command "true"
+  resource "receipt-report" _tag="plan" uri="file:plans/receipt-report/plan.kdl"
+  resource "inline-intent" _tag="plan" uri="file:plans/inline-intent/plan.kdl"
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/inline-intent/plan.kdl b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/inline-intent/plan.kdl
new file mode 100644
index 0000000..7e37f03
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/inline-intent/plan.kdl
@@ -0,0 +1,6 @@
+plan "inline-intent" {
+  owner "receipt-worker"
+  version "0000" {
+    intent "Keep the complete inline intent in plan.kdl."
+  }
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/plan.kdl b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/plan.kdl
new file mode 100644
index 0000000..63844e9
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/plan.kdl
@@ -0,0 +1,4 @@
+plan "receipt-report" {
+  owner "receipt-worker"
+  version "0000" content="file:versions/0000.md"
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md
new file mode 100644
index 0000000..7f75d24
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md
@@ -0,0 +1,37 @@
+# Receipt report: latest accepted pass and last run
+
+Revision: `0000`
+
+## Goal
+
+Implement the report core in `src/report.mjs`. The report must retain the most
+recent accepted `PASS` for each cell independently from that cell's last run,
+which may be a failure.
+
+## Scope and owner
+
+You own the frozen `receipt-report` repository for this task. Change only
+`src/report.mjs`.
+
+## Required behavior
+
+- `parseJsonLines(text)` ignores blank lines and returns records in input order.
+- Malformed JSON, non-object JSON, or a record missing non-empty `run_id`,
+  `cell`, or `result` fields throws a `TypeError` that identifies the input line.
+- `summarize(records)` returns one row per cell, sorted by cell name.
+- Each row is `{ cell, accepted_pass, last_run }`.
+- `accepted_pass` is the complete most recent `PASS` record for the cell, or
+  `null` when the cell has no pass.
+- `last_run` is the complete last record for the cell.
+- Neither exported function mutates caller-owned records.
+
+## Invariants and allowed actions
+
+Use only the repository's existing Node, Bash, and Git tools. Do not access the
+network or add dependencies. Do not weaken or replace tests. No action outside
+the frozen repository is authorized.
+
+## Evidence and done condition
+
+Run `npm test`. Completion additionally requires the evaluator's held-out tests
+to pass and a diff containing only `src/report.mjs`.
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/agent.kdl b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/agent.kdl
new file mode 100644
index 0000000..66133e0
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/agent.kdl
@@ -0,0 +1,6 @@
+agent "receipt-worker" {
+  host "eval"
+  command "true"
+  resource "receipt-report" _tag="plan" uri="file:plans/receipt-report/plan.kdl"
+  resource "inline-intent" _tag="plan" uri="file:plans/inline-intent/plan.kdl"
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/inline-intent/plan.kdl b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/inline-intent/plan.kdl
new file mode 100644
index 0000000..7e37f03
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/inline-intent/plan.kdl
@@ -0,0 +1,6 @@
+plan "inline-intent" {
+  owner "receipt-worker"
+  version "0000" {
+    intent "Keep the complete inline intent in plan.kdl."
+  }
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/plan.kdl b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/plan.kdl
new file mode 100644
index 0000000..c556d30
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/plan.kdl
@@ -0,0 +1,9 @@
+plan "receipt-report" {
+  owner "receipt-worker"
+
+  version "0000" content="file:versions/0000.md"
+  version "0001" content="file:versions/0001.md" {
+    parent "0000"
+    why "Human steering tightens input validation without changing scope."
+  }
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md
new file mode 100644
index 0000000..7f75d24
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md
@@ -0,0 +1,37 @@
+# Receipt report: latest accepted pass and last run
+
+Revision: `0000`
+
+## Goal
+
+Implement the report core in `src/report.mjs`. The report must retain the most
+recent accepted `PASS` for each cell independently from that cell's last run,
+which may be a failure.
+
+## Scope and owner
+
+You own the frozen `receipt-report` repository for this task. Change only
+`src/report.mjs`.
+
+## Required behavior
+
+- `parseJsonLines(text)` ignores blank lines and returns records in input order.
+- Malformed JSON, non-object JSON, or a record missing non-empty `run_id`,
+  `cell`, or `result` fields throws a `TypeError` that identifies the input line.
+- `summarize(records)` returns one row per cell, sorted by cell name.
+- Each row is `{ cell, accepted_pass, last_run }`.
+- `accepted_pass` is the complete most recent `PASS` record for the cell, or
+  `null` when the cell has no pass.
+- `last_run` is the complete last record for the cell.
+- Neither exported function mutates caller-owned records.
+
+## Invariants and allowed actions
+
+Use only the repository's existing Node, Bash, and Git tools. Do not access the
+network or add dependencies. Do not weaken or replace tests. No action outside
+the frozen repository is authorized.
+
+## Evidence and done condition
+
+Run `npm test`. Completion additionally requires the evaluator's held-out tests
+to pass and a diff containing only `src/report.mjs`.
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md
new file mode 100644
index 0000000..299b7a5
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md
@@ -0,0 +1,41 @@
+# Receipt report: latest accepted pass and last run
+
+Revision: `0001`
+
+## Goal
+
+Implement the report core in `src/report.mjs`. The report must retain the most
+recent accepted `PASS` for each cell independently from that cell's last run,
+which may be a failure.
+
+## Scope and owner
+
+You own the frozen `receipt-report` repository for this task. Change only
+`src/report.mjs`.
+
+## Required behavior
+
+- `parseJsonLines(text)` ignores blank lines and returns records in input order.
+- Malformed JSON, non-object JSON, or a record missing non-empty `run_id`,
+  `cell`, or `result` fields throws a `TypeError` that identifies the input line.
+- Only `PASS` and `FAIL` results are valid. An unknown result throws a
+  `TypeError` that identifies its `run_id`.
+- A duplicate `run_id` is invalid. Both `parseJsonLines(text)` and
+  `summarize(records)` throw a `TypeError` that identifies the duplicate.
+- `summarize(records)` returns one row per cell, sorted by cell name.
+- Each row is `{ cell, accepted_pass, last_run }`.
+- `accepted_pass` is the complete most recent `PASS` record for the cell, or
+  `null` when the cell has no pass.
+- `last_run` is the complete last record for the cell.
+- Neither exported function mutates caller-owned records.
+
+## Invariants and allowed actions
+
+Use only the repository's existing Node, Bash, and Git tools. Do not access the
+network or add dependencies. Do not weaken or replace tests. No action outside
+the frozen repository is authorized.
+
+## Evidence and done condition
+
+Run `npm test`. Completion additionally requires the evaluator's held-out tests
+to pass and a diff containing only `src/report.mjs`.
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-b/inbox/brief-0000.md b/cells/catalog-plan-vs-direct-brief/fixture/arm-b/inbox/brief-0000.md
new file mode 100644
index 0000000..7f75d24
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-b/inbox/brief-0000.md
@@ -0,0 +1,37 @@
+# Receipt report: latest accepted pass and last run
+
+Revision: `0000`
+
+## Goal
+
+Implement the report core in `src/report.mjs`. The report must retain the most
+recent accepted `PASS` for each cell independently from that cell's last run,
+which may be a failure.
+
+## Scope and owner
+
+You own the frozen `receipt-report` repository for this task. Change only
+`src/report.mjs`.
+
+## Required behavior
+
+- `parseJsonLines(text)` ignores blank lines and returns records in input order.
+- Malformed JSON, non-object JSON, or a record missing non-empty `run_id`,
+  `cell`, or `result` fields throws a `TypeError` that identifies the input line.
+- `summarize(records)` returns one row per cell, sorted by cell name.
+- Each row is `{ cell, accepted_pass, last_run }`.
+- `accepted_pass` is the complete most recent `PASS` record for the cell, or
+  `null` when the cell has no pass.
+- `last_run` is the complete last record for the cell.
+- Neither exported function mutates caller-owned records.
+
+## Invariants and allowed actions
+
+Use only the repository's existing Node, Bash, and Git tools. Do not access the
+network or add dependencies. Do not weaken or replace tests. No action outside
+the frozen repository is authorized.
+
+## Evidence and done condition
+
+Run `npm test`. Completion additionally requires the evaluator's held-out tests
+to pass and a diff containing only `src/report.mjs`.
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arm-b/inbox/brief-0001.md b/cells/catalog-plan-vs-direct-brief/fixture/arm-b/inbox/brief-0001.md
new file mode 100644
index 0000000..299b7a5
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arm-b/inbox/brief-0001.md
@@ -0,0 +1,41 @@
+# Receipt report: latest accepted pass and last run
+
+Revision: `0001`
+
+## Goal
+
+Implement the report core in `src/report.mjs`. The report must retain the most
+recent accepted `PASS` for each cell independently from that cell's last run,
+which may be a failure.
+
+## Scope and owner
+
+You own the frozen `receipt-report` repository for this task. Change only
+`src/report.mjs`.
+
+## Required behavior
+
+- `parseJsonLines(text)` ignores blank lines and returns records in input order.
+- Malformed JSON, non-object JSON, or a record missing non-empty `run_id`,
+  `cell`, or `result` fields throws a `TypeError` that identifies the input line.
+- Only `PASS` and `FAIL` results are valid. An unknown result throws a
+  `TypeError` that identifies its `run_id`.
+- A duplicate `run_id` is invalid. Both `parseJsonLines(text)` and
+  `summarize(records)` throw a `TypeError` that identifies the duplicate.
+- `summarize(records)` returns one row per cell, sorted by cell name.
+- Each row is `{ cell, accepted_pass, last_run }`.
+- `accepted_pass` is the complete most recent `PASS` record for the cell, or
+  `null` when the cell has no pass.
+- `last_run` is the complete last record for the cell.
+- Neither exported function mutates caller-owned records.
+
+## Invariants and allowed actions
+
+Use only the repository's existing Node, Bash, and Git tools. Do not access the
+network or add dependencies. Do not weaken or replace tests. No action outside
+the frozen repository is authorized.
+
+## Evidence and done condition
+
+Run `npm test`. Completion additionally requires the evaluator's held-out tests
+to pass and a diff containing only `src/report.mjs`.
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/arms.tsv b/cells/catalog-plan-vs-direct-brief/fixture/arms.tsv
new file mode 100644
index 0000000..c67b5f4
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/arms.tsv
@@ -0,0 +1,3 @@
+arm	input_kind	initial_source	local_resume_source	steering_source	acceptance_surface	allowed_tools	wall_budget_seconds	model_token_budget	judge_profile	done_condition
+A	versioned-catalog-plan	arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md	local plan catalog inspected through st2 plan	arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md	evaluator-owned receipt only	node,bash,git	1200	40000	arm-neutral-public-plus-held-out	public and held-out tests pass; only src/report.mjs differs from the seed; no dependency or network change
+B	durable-direct-brief	arm-b/inbox/brief-0000.md	local st2 message inbox read through st2 message	arm-b/inbox/brief-0001.md	evaluator-owned receipt only	node,bash,git	1200	40000	arm-neutral-public-plus-held-out	public and held-out tests pass; only src/report.mjs differs from the seed; no dependency or network change
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/blockers.tsv b/cells/catalog-plan-vs-direct-brief/fixture/blockers.tsv
new file mode 100644
index 0000000..c5b3c69
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/blockers.tsv
@@ -0,0 +1,3 @@
+blocker_id	owner	dependency	required_evidence	status
+st2-plan-runtime	st2-owner	https://github.com/compoundingtech/st2/pull/115	exact Resource-link source 8a76b6e71355140e5b89cd9313fcfd88c82b5cad and binary SHA256 214e08874720bc546d4adf7d7977e614237baf7989cc09f6932cd991f497a753 with prior gates and hosted run 30835684680 job 91760161352 green	ready
+live-ab	evals-owner	future provider-backed execution	separate authorization naming exact plan artifact, model, effort, per-arm token and wall budgets, arm order or counterbalancing, maximum spend, cleanup, and rollback	blocked
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/check-fixtures.sh b/cells/catalog-plan-vs-direct-brief/fixture/check-fixtures.sh
new file mode 100755
index 0000000..42c338f
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/check-fixtures.sh
@@ -0,0 +1,228 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+root="${CATALOG:?CATALOG must be set}"
+experiment="$root/experiment.tsv"
+arms="$root/arms.tsv"
+scenario="$root/scenario.tsv"
+measurements="$root/measurement-schema.tsv"
+receipts="$root/receipt-schema.tsv"
+blockers="$root/blockers.tsv"
+runner="$root/runner.tsv"
+plan_declaration="$root/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/plan.kdl"
+agent_declaration="$root/arm-a/catalog/agents/eval/receipt-worker/agent.kdl"
+inline_plan_declaration="$root/arm-a/catalog/agents/eval/receipt-worker/plans/inline-intent/plan.kdl"
+initial_plan_declaration="$root/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/plan.kdl"
+initial_agent_declaration="$root/arm-a/catalog-initial/agents/eval/receipt-worker/agent.kdl"
+initial_inline_plan_declaration="$root/arm-a/catalog-initial/agents/eval/receipt-worker/plans/inline-intent/plan.kdl"
+initial_plan_0000="$root/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md"
+plan_0000="$root/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md"
+plan_0001="$root/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md"
+brief_0000="$root/arm-b/inbox/brief-0000.md"
+brief_0001="$root/arm-b/inbox/brief-0001.md"
+
+expect_header() {
+  file="$1"
+  expected="$2"
+  test "$(head -n 1 "$file")" = "$expected"
+}
+
+tree_hash() {
+  (
+    cd "$1"
+    find . -type f -print0 |
+      LC_ALL=C sort -z |
+      xargs -0 sha256sum |
+      sha256sum |
+      awk '{print $1}'
+  )
+}
+
+expect_header "$experiment" $'field\tvalue'
+expect_header "$arms" $'arm\tinput_kind\tinitial_source\tlocal_resume_source\tsteering_source\tacceptance_surface\tallowed_tools\twall_budget_seconds\tmodel_token_budget\tjudge_profile\tdone_condition'
+expect_header "$scenario" $'ordinal\tevent\tarm_a_input\tarm_b_input\tinvariant'
+expect_header "$measurements" $'metric\ttype\tunit\trole\tpreferred_direction\tmodel_free_arm_a\tmodel_free_arm_b\tmodel_free_verdict\tlive_source'
+expect_header "$receipts" $'field\ttype\trequired\tdescription'
+expect_header "$blockers" $'blocker_id\towner\tdependency\trequired_evidence\tstatus'
+expect_header "$runner" $'field\tvalue'
+
+test "$(awk -F '\t' 'NR > 1 { count++ } END { print count + 0 }' "$arms")" -eq 2
+test "$(awk -F '\t' 'NR > 1 { count++ } END { print count + 0 }' "$scenario")" -eq 6
+test "$(awk -F '\t' 'NR > 1 { count++ } END { print count + 0 }' "$measurements")" -eq 14
+test "$(awk -F '\t' 'NR > 1 { count++ } END { print count + 0 }' "$receipts")" -eq 24
+test "$(awk -F '\t' 'NR > 1 { count++ } END { print count + 0 }' "$blockers")" -eq 2
+test "$(awk -F '\t' 'NR > 1 { count++ } END { print count + 0 }' "$runner")" -eq 8
+
+grep -Fqx $'allowed_tools\tnode,bash,git' "$experiment"
+grep -Fqx $'network_policy\tdisabled' "$experiment"
+grep -Fqx $'wall_budget_seconds\t1200' "$experiment"
+grep -Fqx $'model_token_budget\t40000' "$experiment"
+grep -Fqx $'initial_revision\t0000' "$experiment"
+grep -Fqx $'steering_revision\t0001' "$experiment"
+grep -Fqx $'model_free_result\tno-measured-advantage-for-cold-resume-intent-recovery-or-acceptance-evidence' "$experiment"
+grep -Fqx $'source_gist\thttps://gist.github.com/myobie/d5ecfac24cd3965e095a5031cd2e00cb/5c1d1427c0556d95d13890e5c5086cd85b25d994' "$experiment"
+grep -Fqx $'source_gist_revision\t5c1d1427c0556d95d13890e5c5086cd85b25d994' "$experiment"
+grep -Fqx $'st2_plan_source\t8a76b6e71355140e5b89cd9313fcfd88c82b5cad' "$experiment"
+grep -Fqx $'st2_plan_binary_sha256\t214e08874720bc546d4adf7d7977e614237baf7989cc09f6932cd991f497a753' "$experiment"
+grep -Fqx $'live_run_status\tblocked-pending-separate-provider-authorization' "$experiment"
+grep -Fqx $'st2_pr\thttps://github.com/compoundingtech/st2/pull/115' "$runner"
+grep -Fqx $'source_full\t8a76b6e71355140e5b89cd9313fcfd88c82b5cad' "$runner"
+grep -Fqx $'source_short\t8a76b6e' "$runner"
+grep -Fqx $'binary_sha256\t214e08874720bc546d4adf7d7977e614237baf7989cc09f6932cd991f497a753' "$runner"
+grep -Fqx $'source_gist_revision\t5c1d1427c0556d95d13890e5c5086cd85b25d994' "$runner"
+grep -Fqx $'runtime_scope\tread-only-validate-list-show-inspect' "$runner"
+grep -Fqx $'hosted_run\thttps://github.com/compoundingtech/st2/actions/runs/30835684680' "$runner"
+grep -Fqx $'hosted_status\tpass' "$runner"
+
+cmp -s "$initial_plan_0000" "$plan_0000"
+cmp -s "$plan_0000" "$brief_0000"
+cmp -s "$plan_0001" "$brief_0001"
+test "$(sha256sum "$plan_0000" | awk '{print $1}')" != \
+  "$(sha256sum "$plan_0001" | awk '{print $1}')"
+
+awk -F '\t' '
+  NR == 2 {
+    tools = $7
+    wall = $8
+    tokens = $9
+    judge = $10
+    done = $11
+    next
+  }
+  NR == 3 {
+    if ($7 != tools || $8 != wall || $9 != tokens || $10 != judge || $11 != done) {
+      exit 1
+    }
+  }
+' "$arms"
+grep -Fqx $'A\tversioned-catalog-plan\tarm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md\tlocal plan catalog inspected through st2 plan\tarm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md\tevaluator-owned receipt only\tnode,bash,git\t1200\t40000\tarm-neutral-public-plus-held-out\tpublic and held-out tests pass; only src/report.mjs differs from the seed; no dependency or network change' "$arms"
+grep -Fqx $'B\tdurable-direct-brief\tarm-b/inbox/brief-0000.md\tlocal st2 message inbox read through st2 message\tarm-b/inbox/brief-0001.md\tevaluator-owned receipt only\tnode,bash,git\t1200\t40000\tarm-neutral-public-plus-held-out\tpublic and held-out tests pass; only src/report.mjs differs from the seed; no dependency or network change' "$arms"
+
+grep -Fqx 'plan "receipt-report" {' "$plan_declaration"
+grep -Fqx '  owner "receipt-worker"' "$plan_declaration"
+grep -Fqx '  version "0000" content="file:versions/0000.md"' "$plan_declaration"
+grep -Fqx '  version "0001" content="file:versions/0001.md" {' "$plan_declaration"
+grep -Fqx '    parent "0000"' "$plan_declaration"
+grep -Fqx '    why "Human steering tightens input validation without changing scope."' "$plan_declaration"
+grep -Fqx 'plan "receipt-report" {' "$initial_plan_declaration"
+grep -Fqx '  owner "receipt-worker"' "$initial_plan_declaration"
+grep -Fqx '  version "0000" content="file:versions/0000.md"' "$initial_plan_declaration"
+if grep -Fq 'version "0001"' "$initial_plan_declaration"; then
+  echo "initial plan snapshot already contains the steering revision" >&2
+  exit 1
+fi
+grep -Fqx 'agent "receipt-worker" {' "$agent_declaration"
+grep -Fqx '  resource "receipt-report" _tag="plan" uri="file:plans/receipt-report/plan.kdl"' "$agent_declaration"
+grep -Fqx '  resource "inline-intent" _tag="plan" uri="file:plans/inline-intent/plan.kdl"' "$agent_declaration"
+grep -Fqx 'plan "inline-intent" {' "$inline_plan_declaration"
+grep -Fqx '  owner "receipt-worker"' "$inline_plan_declaration"
+grep -Fqx '    intent "Keep the complete inline intent in plan.kdl."' "$inline_plan_declaration"
+cmp -s "$initial_inline_plan_declaration" "$inline_plan_declaration"
+cmp -s "$initial_agent_declaration" "$agent_declaration"
+if rg -n 'plan-ref|^[[:space:]]+plan[[:space:]]' "$initial_agent_declaration" "$agent_declaration"; then
+  echo "agent declaration still owns plan truth" >&2
+  exit 1
+fi
+
+if (
+  cd "$root/task-repo"
+  npm test >/dev/null 2>&1
+); then
+  echo "seed repository unexpectedly passes before implementation" >&2
+  exit 1
+fi
+
+scratch="$(mktemp -d)"
+cleanup() {
+  rm -rf -- "$scratch"
+}
+trap cleanup EXIT
+cp -a "$root/task-repo" "$scratch/candidate"
+cp "$root/reference/report.mjs" "$scratch/candidate/src/report.mjs"
+"$root/judges/judge-task.sh" "$scratch/candidate" >/dev/null
+echo "TASK-CONTRACT-GREEN-43af"
+
+if rg -n -i 'arm[-_ ]?[ab]|catalog[ -]?plan|direct[ -]?brief' \
+  "$root/judges/judge-task.sh" "$root/judges/report.test.mjs"; then
+  echo "arm-neutral correctness judge contains delivery-arm identity" >&2
+  exit 1
+fi
+echo "ARM-PARITY-GREEN-43af"
+
+matrix_output="$(CATALOG="$root" "$root/run-recovery-matrix.sh")"
+for marker in \
+  COLD-RESTART-TIE-43af \
+  LOCAL-PARTITION-TIE-43af \
+  INTENT-RECOVERY-TIE-43af \
+  HUMAN-STEERING-TIE-43af \
+  ACCEPTANCE-EVIDENCE-TIE-43af \
+  PLAN-STATIC-EVIDENCE-GREEN-43af \
+  DIRECT-MESSAGE-EVIDENCE-GREEN-43af \
+  MODEL-FREE-COMPARISON-GREEN-43af \
+  PLAIN-FOLDER-NO-CAS-GREEN-43af; do
+  grep -Fqx "$marker" <<<"$matrix_output"
+done
+jq -e '
+  .outcomes.coldResume.verdict == "tie" and
+  .outcomes.intentRecovery.verdict == "tie" and
+  .outcomes.humanSteering.verdict == "tie" and
+  .outcomes.acceptanceEvidence.verdict == "tie" and
+  .conclusion == "no-measured-advantage"
+' "$root/comparison-receipt.json" >/dev/null
+echo "COLD-RESTART-TIE-43af"
+echo "LOCAL-PARTITION-TIE-43af"
+echo "INTENT-RECOVERY-TIE-43af"
+echo "HUMAN-STEERING-TIE-43af"
+echo "ACCEPTANCE-EVIDENCE-TIE-43af"
+echo "PLAN-STATIC-EVIDENCE-GREEN-43af"
+echo "DIRECT-MESSAGE-EVIDENCE-GREEN-43af"
+echo "PLAIN-FOLDER-NO-CAS-GREEN-43af"
+echo "MODEL-FREE-OUTCOME-GREEN-43af"
+
+test "$(awk -F '\t' 'NR > 1 && $4 == "model-free-gating" { count++ } END { print count + 0 }' "$measurements")" -eq 3
+test "$(awk -F '\t' 'NR > 1 && $4 == "live-gating" { count++ } END { print count + 0 }' "$measurements")" -eq 2
+test "$(awk -F '\t' 'NR > 1 && $4 == "reporting" { count++ } END { print count + 0 }' "$measurements")" -eq 3
+test "$(awk -F '\t' 'NR > 1 && $4 == "live-reporting" { count++ } END { print count + 0 }' "$measurements")" -eq 6
+awk -F '\t' '
+  NR > 1 && ($4 == "model-free-gating" || $4 == "live-gating" || $4 == "reporting") && $5 != "pass" && $1 != "acceptance_evidence" { exit 1 }
+  NR > 1 && $4 == "live-reporting" && $5 != "lower" { exit 1 }
+' "$measurements"
+test "$(awk -F '\t' 'NR > 1 && $3 != "yes" { count++ } END { print count + 0 }' "$receipts")" -eq 0
+echo "MEASUREMENT-CONTRACT-GREEN-43af"
+
+checked=0
+while IFS=$'\t' read -r path expected; do
+  if test "$path" = "path"; then
+    test "$expected" = "sha256"
+    continue
+  fi
+  if test "$path" = "task-repo.tree"; then
+    test "$(tree_hash "$root/task-repo")" = "$expected"
+    checked=$((checked + 1))
+    continue
+  fi
+  test -f "$root/$path"
+  actual="$(sha256sum "$root/$path" | awk '{print $1}')"
+  test "$actual" = "$expected"
+  checked=$((checked + 1))
+done <"$root/provenance.tsv"
+expected_provenance="$(awk -F '\t' 'NR > 1 { count++ } END { print count + 0 }' "$root/provenance.tsv")"
+test "$checked" -eq "$expected_provenance"
+echo "PROVENANCE-GREEN-43af"
+
+test "$(awk -F '\t' 'NR > 1 && $5 == "ready" { count++ } END { print count + 0 }' "$blockers")" -eq 1
+test "$(awk -F '\t' 'NR > 1 && $5 == "blocked" { count++ } END { print count + 0 }' "$blockers")" -eq 1
+grep -Fq '8a76b6e71355140e5b89cd9313fcfd88c82b5cad' "$blockers"
+grep -Fq '214e08874720bc546d4adf7d7977e614237baf7989cc09f6932cd991f497a753' "$blockers"
+grep -Fq 'hosted run 30835684680 job 91760161352 green' "$blockers"
+grep -Fq 'separate authorization' "$blockers"
+test "$(find "$root/arm-a" -type f -name '*.kdl' | wc -l)" -eq 6
+echo "PRODUCT-PAIRING-GREEN-43af"
+
+if rg -n --pcre2 \
+  '(^|[;&|][[:space:]]*)(claude|codex|curl|wget|ssh|gh|st2[[:space:]]+(up|eval|ping)|pty[[:space:]]+(run|send)|eval)[[:space:]]' \
+  "$root" -g '*.sh' -g '*.kdl'; then
+  echo "fixture contains a provider, network, mutating product runtime, or nested eval command" >&2
+  exit 1
+fi
+echo "HERMETIC-SCAFFOLD-GREEN-43af"
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/check-plan-contract.sh b/cells/catalog-plan-vs-direct-brief/fixture/check-plan-contract.sh
new file mode 100755
index 0000000..cb08fea
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/check-plan-contract.sh
@@ -0,0 +1,139 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+root="${CATALOG:?CATALOG must be set}"
+runner="$root/runner.tsv"
+catalog="$root/arm-a/catalog"
+invalid="$root/invalid/current"
+st2_path="$(command -v st2)"
+
+value() {
+  awk -F '\t' -v key="$1" '$1 == key { print $2 }' "$runner"
+}
+
+source_short="$(value source_short)"
+expected_sha="$(value binary_sha256)"
+version_regex="^st2 0\\.1\\.0 — running from local source \\(${source_short}, .+ ago\\)$"
+actual_version="$("$st2_path" --version)"
+actual_sha="$(sha256sum "$st2_path" | awk '{print $1}')"
+[[ "$actual_version" =~ $version_regex ]]
+test "$actual_sha" = "$expected_sha"
+echo "PLAN-RUNNER-PIN-GREEN-43af"
+
+before="$(mktemp)"
+after="$(mktemp)"
+invalid_json="$(mktemp)"
+cleanup() {
+  rm -f -- "$before" "$after" "$invalid_json"
+}
+trap cleanup EXIT
+
+find "$catalog" -type f -print0 |
+  LC_ALL=C sort -z |
+  xargs -0 sha256sum >"$before"
+
+validate_json="$("$st2_path" plan validate --catalog "$catalog" --json)"
+jq -e '
+  .result == "valid" and
+  .plans == 2 and
+  .errors == 0
+' <<<"$validate_json" >/dev/null
+echo "PLAN-VALIDATE-GREEN-43af"
+
+list_json="$("$st2_path" plan list --catalog "$catalog" --json)"
+jq -e '
+  length == 2 and
+  .[0] == {
+    "identity": "inline-intent",
+    "owner": "receipt-worker",
+    "frontier": ["0000"]
+  } and
+  .[1] == {
+    "identity": "receipt-report",
+    "owner": "receipt-worker",
+    "frontier": ["0001"]
+  }
+' <<<"$list_json" >/dev/null
+echo "PLAN-LIST-GREEN-43af"
+
+show_json="$("$st2_path" plan show receipt-report --catalog "$catalog" --json)"
+jq -e '
+  .identity == "receipt-report" and
+  .owner == "receipt-worker" and
+  .frontier == ["0001"] and
+  (.versions | length) == 2 and
+  .versions[0] == {
+    "identity": "0000",
+    "parents": [],
+    "why": null,
+    "content": "file:versions/0000.md"
+  } and
+  .versions[1] == {
+    "identity": "0001",
+    "parents": ["0000"],
+    "why": "Human steering tightens input validation without changing scope.",
+    "content": "file:versions/0001.md"
+  } and
+  (has("source") | not) and
+  (.versions[0] | has("resolvedContent") | not)
+' <<<"$show_json" >/dev/null
+echo "PLAN-SHOW-GREEN-43af"
+
+inspect_json="$("$st2_path" plan inspect receipt-report --catalog "$catalog" --json)"
+jq -e '
+  .identity == "receipt-report" and
+  .owner == "receipt-worker" and
+  .sourceKind == "external" and
+  .referencedBy == ["receipt-worker"] and
+  .frontier == ["0001"] and
+  (.source | endswith("/agents/eval/receipt-worker/plans/receipt-report/plan.kdl")) and
+  (.versions[0].resolvedContent | endswith("/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md")) and
+  (.versions[1].resolvedContent | endswith("/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md"))
+' <<<"$inspect_json" >/dev/null
+
+inline_show_json="$("$st2_path" plan show inline-intent --catalog "$catalog" --json)"
+jq -e '
+  .identity == "inline-intent" and
+  .owner == "receipt-worker" and
+  .frontier == ["0000"] and
+  .versions == [{
+    "identity": "0000",
+    "parents": [],
+    "why": null,
+    "intent": "Keep the complete inline intent in plan.kdl."
+  }] and
+  (has("source") | not) and
+  (.versions[0] | has("content") | not) and
+  (.versions[0] | has("resolvedContent") | not)
+' <<<"$inline_show_json" >/dev/null
+
+inline_inspect_json="$("$st2_path" plan inspect inline-intent --catalog "$catalog" --json)"
+jq -e '
+  .identity == "inline-intent" and
+  .sourceKind == "external" and
+  .referencedBy == ["receipt-worker"] and
+  (.source | endswith("/agents/eval/receipt-worker/plans/inline-intent/plan.kdl")) and
+  .versions[0].intent == "Keep the complete inline intent in plan.kdl." and
+  (.versions[0] | has("content") | not) and
+  (.versions[0] | has("resolvedContent") | not)
+' <<<"$inline_inspect_json" >/dev/null
+echo "PLAN-TARGET-FORMS-GREEN-43af"
+
+find "$catalog" -type f -print0 |
+  LC_ALL=C sort -z |
+  xargs -0 sha256sum >"$after"
+cmp -s "$before" "$after"
+test ! -e "$catalog/.st2"
+echo "PLAN-INSPECT-READONLY-GREEN-43af"
+
+if "$st2_path" plan validate --catalog "$invalid" --json >"$invalid_json" 2>/dev/null; then
+  echo "unsupported current pointer unexpectedly validated" >&2
+  exit 1
+fi
+jq -e '
+  .result == "invalid" and
+  .code == "unsupported-plan-field" and
+  (.error | contains("unsupported-current")) and
+  (.error | contains("current"))
+' "$invalid_json" >/dev/null
+echo "PLAN-BOUNDARY-GREEN-43af"
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/experiment.tsv b/cells/catalog-plan-vs-direct-brief/fixture/experiment.tsv
new file mode 100644
index 0000000..e5eff76
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/experiment.tsv
@@ -0,0 +1,20 @@
+field	value
+experiment_id	catalog-plan-vs-direct-brief-v1
+source_gist	https://gist.github.com/myobie/d5ecfac24cd3965e095a5031cd2e00cb/5c1d1427c0556d95d13890e5c5086cd85b25d994
+source_gist_revision	5c1d1427c0556d95d13890e5c5086cd85b25d994
+st2_plan_pr	https://github.com/compoundingtech/st2/pull/115
+st2_plan_source	8a76b6e71355140e5b89cd9313fcfd88c82b5cad
+st2_plan_binary_sha256	214e08874720bc546d4adf7d7977e614237baf7989cc09f6932cd991f497a753
+task_id	receipt-report-latest-pass
+seed_repository	task-repo
+allowed_change	src/report.mjs
+allowed_tools	node,bash,git
+network_policy	disabled
+wall_budget_seconds	1200
+model_token_budget	40000
+judge_profile	arm-neutral-public-plus-held-out
+done_condition	public and held-out tests pass; only src/report.mjs differs from the seed; no dependency or network change
+initial_revision	0000
+steering_revision	0001
+model_free_result	no-measured-advantage-for-cold-resume-intent-recovery-or-acceptance-evidence
+live_run_status	blocked-pending-separate-provider-authorization
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/invalid/current/0000.md b/cells/catalog-plan-vs-direct-brief/fixture/invalid/current/0000.md
new file mode 100644
index 0000000..ef57ad9
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/invalid/current/0000.md
@@ -0,0 +1 @@
+# Unsupported mutable current pointer
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/invalid/current/plan.kdl b/cells/catalog-plan-vs-direct-brief/fixture/invalid/current/plan.kdl
new file mode 100644
index 0000000..999c07d
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/invalid/current/plan.kdl
@@ -0,0 +1,5 @@
+plan "unsupported-current" {
+  owner "receipt-worker"
+  current "0000"
+  version "0000" content="file:0000.md"
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/judges/judge-task.sh b/cells/catalog-plan-vs-direct-brief/fixture/judges/judge-task.sh
new file mode 100755
index 0000000..fbc47dc
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/judges/judge-task.sh
@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+fixture_root="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+candidate="${1:?usage: judge-task.sh CANDIDATE_REPOSITORY}"
+seed="$fixture_root/task-repo"
+
+test -f "$candidate/src/report.mjs"
+
+mapfile -t seed_files < <(
+  cd "$seed"
+  find . -type f -print | LC_ALL=C sort
+)
+mapfile -t candidate_files < <(
+  cd "$candidate"
+  find . -type f ! -path './.git/*' -print | LC_ALL=C sort
+)
+test "${seed_files[*]}" = "${candidate_files[*]}"
+
+while IFS= read -r relative; do
+  case "$relative" in
+    ./src/report.mjs) continue ;;
+  esac
+  cmp -s "$seed/${relative#./}" "$candidate/${relative#./}"
+done < <(printf '%s\n' "${seed_files[@]}")
+
+(
+  cd "$candidate"
+  npm test
+)
+CANDIDATE_REPORT="$candidate/src/report.mjs" \
+  node --test "$fixture_root/judges/report.test.mjs"
+
+echo "ARM-NEUTRAL-TASK-JUDGE-GREEN-43af"
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/judges/report.test.mjs b/cells/catalog-plan-vs-direct-brief/fixture/judges/report.test.mjs
new file mode 100644
index 0000000..c4abd76
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/judges/report.test.mjs
@@ -0,0 +1,111 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import { pathToFileURL } from "node:url";
+
+const candidate = process.env.CANDIDATE_REPORT;
+if (!candidate) {
+  throw new Error("CANDIDATE_REPORT is required");
+}
+
+const { parseJsonLines, summarize } = await import(pathToFileURL(candidate));
+
+test("retains complete records without mutating caller input", () => {
+  const records = [
+    { run_id: "r1", cell: "zeta", result: "PASS", usage: { input: 10 } },
+    { run_id: "r2", cell: "alpha", result: "FAIL", note: "first" },
+    { run_id: "r3", cell: "zeta", result: "FAIL", usage: { input: 11 } },
+    { run_id: "r4", cell: "alpha", result: "PASS", note: "accepted" },
+  ];
+  const original = structuredClone(records);
+
+  assert.deepEqual(summarize(records), [
+    {
+      cell: "alpha",
+      accepted_pass: {
+        run_id: "r4",
+        cell: "alpha",
+        result: "PASS",
+        note: "accepted",
+      },
+      last_run: {
+        run_id: "r4",
+        cell: "alpha",
+        result: "PASS",
+        note: "accepted",
+      },
+    },
+    {
+      cell: "zeta",
+      accepted_pass: {
+        run_id: "r1",
+        cell: "zeta",
+        result: "PASS",
+        usage: { input: 10 },
+      },
+      last_run: {
+        run_id: "r3",
+        cell: "zeta",
+        result: "FAIL",
+        usage: { input: 11 },
+      },
+    },
+  ]);
+  assert.deepEqual(records, original);
+});
+
+test("rejects duplicate run ids in parsed and direct records", () => {
+  const duplicateText = [
+    '{"run_id":"same","cell":"alpha","result":"PASS"}',
+    '{"run_id":"same","cell":"beta","result":"FAIL"}',
+  ].join("\n");
+
+  assert.throws(
+    () => parseJsonLines(duplicateText),
+    (error) =>
+      error instanceof TypeError &&
+      /duplicate/.test(error.message) &&
+      /same/.test(error.message),
+  );
+  assert.throws(
+    () =>
+      summarize([
+        { run_id: "same", cell: "alpha", result: "PASS" },
+        { run_id: "same", cell: "alpha", result: "FAIL" },
+      ]),
+    (error) =>
+      error instanceof TypeError &&
+      /duplicate/.test(error.message) &&
+      /same/.test(error.message),
+  );
+});
+
+test("rejects unknown results and identifies the run", () => {
+  assert.throws(
+    () =>
+      parseJsonLines(
+        '{"run_id":"mystery-7","cell":"alpha","result":"CANCELLED"}',
+      ),
+    (error) =>
+      error instanceof TypeError &&
+      /unknown result/.test(error.message) &&
+      /mystery-7/.test(error.message),
+  );
+});
+
+test("reports malformed and structurally invalid input lines", () => {
+  assert.throws(
+    () => parseJsonLines('\n{"run_id":"ok","cell":"alpha","result":"PASS"}\n{'),
+    (error) => error instanceof TypeError && /line 3/.test(error.message),
+  );
+  assert.throws(
+    () => parseJsonLines('["not","an","object"]'),
+    (error) => error instanceof TypeError && /line 1/.test(error.message),
+  );
+  assert.throws(
+    () => parseJsonLines('{"run_id":"","cell":"alpha","result":"PASS"}'),
+    (error) =>
+      error instanceof TypeError &&
+      /line 1/.test(error.message) &&
+      /run_id/.test(error.message),
+  );
+});
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/measurement-schema.tsv b/cells/catalog-plan-vs-direct-brief/fixture/measurement-schema.tsv
new file mode 100644
index 0000000..eb33629
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/measurement-schema.tsv
@@ -0,0 +1,15 @@
+metric	type	unit	role	preferred_direction	model_free_arm_a	model_free_arm_b	model_free_verdict	live_source
+correctness	boolean	pass/fail	live-gating	pass	not-run	not-run	unresolved-live	arm-neutral held-out judge
+cold_restart_recovery	boolean	pass/fail	model-free-gating	pass	pass	pass	tie	comparison receipt
+partition_recovery	boolean	pass/fail	model-free-gating	pass	pass	pass	tie	comparison receipt
+human_steering	boolean	pass/fail	model-free-gating	pass	pass	pass	tie	comparison receipt
+acceptance_evidence	enum	evaluator-only|product-native	live-gating	product-native	evaluator-only	evaluator-only	tie	comparison receipt plus future worker claim
+static_intent_validation	boolean	pass/fail	reporting	pass	pass	n/a	plan-only	st2 plan validate
+lineage_observation	boolean	pass/fail	reporting	pass	pass	pass	tie	st2 plan show and st2 message inReplyTo
+provenance_resolution	boolean	pass/fail	reporting	pass	pass	n/a	plan-only	st2 plan inspect
+coordination_messages	integer	count	live-reporting	lower	unmeasured	unmeasured	unresolved-live	st2 message receipt
+coordination_bytes	integer	bytes	live-reporting	lower	unmeasured	unmeasured	unresolved-live	st2 message and sync receipt
+model_input_tokens	integer	tokens	live-reporting	lower	0	0	model-free-only	harness usage receipt
+model_output_tokens	integer	tokens	live-reporting	lower	0	0	model-free-only	harness usage receipt
+cost_usd	decimal	USD	live-reporting	lower	0	0	model-free-only	harness usage receipt
+wall_duration	decimal	seconds	live-reporting	lower	unmeasured	unmeasured	unresolved-live	evaluator run receipt
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/provenance.tsv b/cells/catalog-plan-vs-direct-brief/fixture/provenance.tsv
new file mode 100644
index 0000000..0a0466c
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/provenance.tsv
@@ -0,0 +1,31 @@
+path	sha256
+experiment.tsv	7eb1c81bb4aa26c37aed62488602fc8ddda83fe946301832682252c440bc605e
+arms.tsv	bd7c41b75fc29920b5ba1d1298649f5b94e65cf85ed9b8192def51e22357e7b2
+scenario.tsv	884a9c83b983128c8e642bf3a945d6f7edb544ac224f5934a4468f17a48d282b
+measurement-schema.tsv	01763a0c090b9c2fd39e5f1b7995380e2dc0d2e5d4230227749ddd41a133f9ab
+receipt-schema.tsv	c996eaee81301640442b04d9fca31937bb840ed10ef78cca035e5990d1fe6441
+blockers.tsv	140a4fb38bae58ef5dbace2a10574000808d3d1fd5b8e609b829850dc5fb49ba
+runner.tsv	8114588d818d2306d944785953447cb6114bdb83e0e72c6ebf3bc060016b19b8
+arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/plan.kdl	08dca1863eadce1d1616d29d1f01de21435b07015d36066f42e5dbe8d68edbd1
+arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md	4f81d370fe6c09c5c199046cc948f68ebdc8a510a4c30fde8edc6993d1abef11
+arm-a/catalog-initial/agents/eval/receipt-worker/plans/inline-intent/plan.kdl	c1cf3ed40cc909edb8bcb69418dcbe21daf2ed074be34638e632cc31c4f0ae23
+arm-a/catalog-initial/agents/eval/receipt-worker/agent.kdl	e347308269970ab6abde6c1f99a96e0619783f55081ec348d9107c10a10a4e92
+arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/plan.kdl	9781d56aceee726cf26c639b195d100fb66229b0855fa9ba5ce2f1e56024e9ba
+arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md	4f81d370fe6c09c5c199046cc948f68ebdc8a510a4c30fde8edc6993d1abef11
+arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md	aa3075e6fcbec4648209f77cd0b3b9866d70f4a354bb16554d3e58babd7f04b4
+arm-a/catalog/agents/eval/receipt-worker/plans/inline-intent/plan.kdl	c1cf3ed40cc909edb8bcb69418dcbe21daf2ed074be34638e632cc31c4f0ae23
+arm-a/catalog/agents/eval/receipt-worker/agent.kdl	e347308269970ab6abde6c1f99a96e0619783f55081ec348d9107c10a10a4e92
+arm-b/inbox/brief-0000.md	4f81d370fe6c09c5c199046cc948f68ebdc8a510a4c30fde8edc6993d1abef11
+arm-b/inbox/brief-0001.md	aa3075e6fcbec4648209f77cd0b3b9866d70f4a354bb16554d3e58babd7f04b4
+invalid/current/plan.kdl	a7d5792410bcd4f6ba073a1fecaee536adec7a40f82298c2d3022a75f9a0b687
+invalid/current/0000.md	1a18971d45c564ee22de2496c1a086b6a799b9cf809a6d7351236d3c5b4e788a
+task-repo/package.json	8f756153985ab193ec21992e4adf874db4b02330db019006c5214550c2ff9618
+task-repo/README.md	c50a603eac8bb3dad994dfc96f682f7f8d9a56300e21a7f7be0d040308ff580b
+task-repo/src/report.mjs	e2bc26ea2e19e14290585388c4f7d2cc10f56294cb4547068af884847197f020
+task-repo/test/report.test.mjs	e9008004b55687df44025f66abe2e5ff739de8d8c7bec27d8d2815defd3a3a31
+reference/report.mjs	7343d91050754693f1ced37a97dae6f8701549af27ee78e08bab260a5d6550c9
+judges/report.test.mjs	24f41a2e596c940a8a3c5123e53b635bc66deb11d48bc0911a280d7a5485e670
+judges/judge-task.sh	23680852e5d3ffd744eb00b6edf12c85f5560cbe3b406da45169e710eff2a10b
+run-recovery-matrix.sh	29aab23eeca0c1af1f76c6745da11c37da4e8a1800592a8a53df4eb963941533
+check-plan-contract.sh	474a5b9ffa62d899ed81f4867f245202e6c161593d2e7cec5c2bcdfb969f5335
+task-repo.tree	22c368e3d2b9722c0fe53710dbd16e2010e98d7215db0aa42fe7532cfa9233c5
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/receipt-schema.tsv b/cells/catalog-plan-vs-direct-brief/fixture/receipt-schema.tsv
new file mode 100644
index 0000000..63d7983
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/receipt-schema.tsv
@@ -0,0 +1,25 @@
+field	type	required	description
+schemaVersion	integer	yes	comparison receipt schema version
+experimentId	string	yes	frozen experiment identity
+runner.source	git-sha	yes	exact experimental st2 source
+runner.sha256	sha256	yes	exact experimental st2 executable
+modelCalls	nonnegative-integer	yes	provider/model calls made by this run
+remoteSourcesOfflineDuringRecovery	boolean	yes	both recoveries occurred with remote sources unavailable
+storage.plainFoldersSufficient	boolean	yes	no service or content-addressed store is required
+storage.casRequired	boolean	yes	whether content-addressed storage is required
+arms.A.transport	enum:catalog-plan	yes	arm A durable transport
+arms.B.transport	enum:st2-direct-messages	yes	arm B durable transport
+arms.A.initialIntentSha256	sha256	yes	recovered initial intent hash
+arms.B.initialIntentSha256	sha256	yes	recovered initial intent hash
+arms.A.steeredIntentSha256	sha256	yes	recovered revised intent hash
+arms.B.steeredIntentSha256	sha256	yes	recovered revised intent hash
+outcomes.coldResume.verdict	enum:tie|arm-a|arm-b	yes	comparative cold-resume result
+outcomes.intentRecovery.verdict	enum:tie|arm-a|arm-b	yes	comparative exact-intent result
+outcomes.humanSteering.verdict	enum:tie|arm-a|arm-b	yes	comparative steering result
+outcomes.acceptanceEvidence.verdict	enum:tie|arm-a|arm-b	yes	comparative worker-acceptance evidence result
+planOnly.staticIntentValidation	enum:pass|fail	yes	read-only plan schema and graph validation
+planOnly.provenanceResolution	enum:pass|fail	yes	read-only resolved resource provenance
+directOnly.messageDelivery	enum:pass|fail	yes	durable direct-message delivery
+directOnly.replyLineage	enum:pass|fail	yes	direct-message inReplyTo lineage
+liveUnresolved	array	yes	endpoints requiring separately authorized model runs
+conclusion	enum:no-measured-advantage|arm-a-advantage|arm-b-advantage	yes	model-free comparative conclusion
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/reference/report.mjs b/cells/catalog-plan-vs-direct-brief/fixture/reference/report.mjs
new file mode 100644
index 0000000..6b8f744
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/reference/report.mjs
@@ -0,0 +1,76 @@
+function validateRecord(record, context) {
+  if (record === null || typeof record !== "object" || Array.isArray(record)) {
+    throw new TypeError(`${context}: expected an object record`);
+  }
+
+  for (const field of ["run_id", "cell", "result"]) {
+    if (typeof record[field] !== "string" || record[field].length === 0) {
+      throw new TypeError(`${context}: missing non-empty ${field}`);
+    }
+  }
+
+  if (record.result !== "PASS" && record.result !== "FAIL") {
+    throw new TypeError(
+      `${context}: unknown result ${record.result} for ${record.run_id}`,
+    );
+  }
+}
+
+function validateUnique(records, contextForIndex) {
+  const seen = new Set();
+  records.forEach((record, index) => {
+    const context = contextForIndex(index);
+    validateRecord(record, context);
+    if (seen.has(record.run_id)) {
+      throw new TypeError(`${context}: duplicate run_id ${record.run_id}`);
+    }
+    seen.add(record.run_id);
+  });
+}
+
+export function parseJsonLines(text) {
+  const records = [];
+  const sourceLines = text.split("\n");
+
+  sourceLines.forEach((line, index) => {
+    if (line.trim().length === 0) {
+      return;
+    }
+
+    let record;
+    try {
+      record = JSON.parse(line);
+    } catch {
+      throw new TypeError(`line ${index + 1}: malformed JSON`);
+    }
+    records.push({ record, lineNumber: index + 1 });
+  });
+
+  validateUnique(
+    records.map(({ record }) => record),
+    (index) => `line ${records[index].lineNumber}`,
+  );
+  return records.map(({ record }) => record);
+}
+
+export function summarize(records) {
+  validateUnique(records, (index) => `record ${index + 1}`);
+
+  const byCell = new Map();
+  for (const record of records) {
+    const current = byCell.get(record.cell) ?? {
+      cell: record.cell,
+      accepted_pass: null,
+      last_run: null,
+    };
+    if (record.result === "PASS") {
+      current.accepted_pass = { ...record };
+    }
+    current.last_run = { ...record };
+    byCell.set(record.cell, current);
+  }
+
+  return [...byCell.values()].sort((left, right) =>
+    left.cell.localeCompare(right.cell),
+  );
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/run-recovery-matrix.sh b/cells/catalog-plan-vs-direct-brief/fixture/run-recovery-matrix.sh
new file mode 100755
index 0000000..fc84238
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/run-recovery-matrix.sh
@@ -0,0 +1,226 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+root="${CATALOG:?CATALOG must be set}"
+runner="$root/runner.tsv"
+receipt="$root/comparison-receipt.json"
+st2_path="$(command -v st2)"
+scratch="$(mktemp -d)"
+
+cleanup() {
+  rm -rf -- "$scratch"
+}
+trap cleanup EXIT
+
+runner_value() {
+  awk -F '\t' -v key="$1" '$1 == key { print $2 }' "$runner"
+}
+
+hash_file() {
+  sha256sum "$1" | awk '{print $1}'
+}
+
+assert_no_acceptance_claim() {
+  jq -e '
+    (has("accepted") | not) and
+    (has("acceptance") | not) and
+    (has("completed") | not) and
+    (has("receipt") | not) and
+    (has("receipts") | not)
+  ' >/dev/null
+}
+
+source_full="$(runner_value source_full)"
+source_short="$(runner_value source_short)"
+expected_binary_sha="$(runner_value binary_sha256)"
+version_regex="^st2 0\\.1\\.0 — running from local source \\(${source_short}, .+ ago\\)$"
+actual_version="$($st2_path --version)"
+actual_binary_sha="$(hash_file "$st2_path")"
+[[ "$actual_version" =~ $version_regex ]]
+test "$actual_binary_sha" = "$expected_binary_sha"
+
+remote="$scratch/remote"
+local_store="$scratch/local"
+session="$scratch/session"
+plan_local="$local_store/plan-catalog"
+direct_bus="$local_store/direct-bus"
+mkdir -p "$remote" "$local_store" "$session" "$direct_bus"
+cp -a "$root/arm-a/catalog-initial" "$remote/plan-initial"
+cp -a "$root/arm-a/catalog" "$remote/plan-steered"
+cp -a "$root/arm-b/inbox" "$remote/direct-source"
+
+# Initial delivery: sync only the initial plan snapshot, and deliver only the
+# initial direct brief through the real st2 message store.
+cp -a "$remote/plan-initial" "$plan_local"
+initial_message="$($st2_path message send eval.worker \
+  --catalog "$direct_bus" \
+  --as eval.supervisor \
+  --subject 'intent revision 0000' \
+  <"$remote/direct-source/brief-0000.md")"
+test -n "$initial_message"
+
+# Cold restart and partition: volatile state disappears and both remote input
+# sources go offline. Recovery must use only each arm's local durable store.
+printf 'volatile\n' >"$session/state"
+rm -rf -- "$session"
+mv "$remote" "$remote.offline"
+test ! -e "$remote"
+
+plan_initial_show="$($st2_path plan show receipt-report --catalog "$plan_local" --json)"
+plan_initial_inspect="$($st2_path plan inspect receipt-report --catalog "$plan_local" --json)"
+jq -e '
+  .frontier == ["0000"] and
+  (.versions | length) == 1 and
+  .versions[0].identity == "0000"
+' <<<"$plan_initial_show" >/dev/null
+plan_initial_path="$(jq -r '.versions[0].resolvedContent' <<<"$plan_initial_inspect")"
+plan_initial_sha="$(hash_file "$plan_initial_path")"
+test "$plan_initial_sha" = "$(hash_file "$root/arm-a/catalog-initial/agents/eval/receipt-worker/plans/receipt-report/versions/0000.md")"
+assert_no_acceptance_claim <<<"$plan_initial_show"
+assert_no_acceptance_claim <<<"$plan_initial_inspect"
+
+direct_initial_list="$($st2_path message ls eval.worker --catalog "$direct_bus" --json)"
+jq -e --arg filename "$initial_message" '
+  length == 1 and
+  .[0].filename == $filename and
+  .[0].from == "eval.supervisor" and
+  .[0].subject == "intent revision 0000" and
+  .[0].inReplyTo == null
+' <<<"$direct_initial_list" >/dev/null
+direct_initial_read="$($st2_path message read eval.worker "$initial_message" --catalog "$direct_bus" --json)"
+jq -rj '.body' <<<"$direct_initial_read" >"$scratch/direct-initial.md"
+direct_initial_sha="$(hash_file "$scratch/direct-initial.md")"
+test "$direct_initial_sha" = "$(hash_file "$root/arm-b/inbox/brief-0000.md")"
+assert_no_acceptance_claim <<<"$direct_initial_read"
+echo "COLD-RESTART-TIE-43af"
+echo "LOCAL-PARTITION-TIE-43af"
+
+# Human steering: restore the remote source, sync a complete new plan snapshot,
+# and deliver a reply-linked direct brief. Then repeat cold recovery offline.
+mv "$remote.offline" "$remote"
+rm -rf -- "$plan_local"
+cp -a "$remote/plan-steered" "$plan_local"
+steered_message="$($st2_path message send eval.worker \
+  --catalog "$direct_bus" \
+  --as eval.supervisor \
+  --subject 'intent revision 0001' \
+  --in-reply-to "$initial_message" \
+  <"$remote/direct-source/brief-0001.md")"
+test -n "$steered_message"
+
+mkdir -p "$session"
+printf 'volatile\n' >"$session/state"
+rm -rf -- "$session"
+mv "$remote" "$remote.offline"
+test ! -e "$remote"
+
+plan_steered_show="$($st2_path plan show receipt-report --catalog "$plan_local" --json)"
+plan_steered_inspect="$($st2_path plan inspect receipt-report --catalog "$plan_local" --json)"
+jq -e '
+  .frontier == ["0001"] and
+  (.versions | length) == 2 and
+  .versions[0].identity == "0000" and
+  .versions[1].identity == "0001" and
+  .versions[1].parents == ["0000"]
+' <<<"$plan_steered_show" >/dev/null
+plan_steered_path="$(jq -r '.versions[] | select(.identity == "0001") | .resolvedContent' <<<"$plan_steered_inspect")"
+plan_steered_sha="$(hash_file "$plan_steered_path")"
+test "$plan_steered_sha" = "$(hash_file "$root/arm-a/catalog/agents/eval/receipt-worker/plans/receipt-report/versions/0001.md")"
+assert_no_acceptance_claim <<<"$plan_steered_show"
+assert_no_acceptance_claim <<<"$plan_steered_inspect"
+
+direct_steered_list="$($st2_path message ls eval.worker --catalog "$direct_bus" --json)"
+jq -e --arg initial "$initial_message" --arg steered "$steered_message" '
+  length == 2 and
+  any(.[]; .filename == $initial and .inReplyTo == null) and
+  any(.[]; .filename == $steered and .inReplyTo == $initial and .subject == "intent revision 0001")
+' <<<"$direct_steered_list" >/dev/null
+direct_steered_read="$($st2_path message read eval.worker "$steered_message" --catalog "$direct_bus" --json)"
+jq -rj '.body' <<<"$direct_steered_read" >"$scratch/direct-steered.md"
+direct_steered_sha="$(hash_file "$scratch/direct-steered.md")"
+test "$direct_steered_sha" = "$(hash_file "$root/arm-b/inbox/brief-0001.md")"
+test "$(jq -r '.inReplyTo' <<<"$direct_steered_read")" = "$initial_message"
+assert_no_acceptance_claim <<<"$direct_steered_read"
+
+test "$plan_initial_sha" = "$direct_initial_sha"
+test "$plan_steered_sha" = "$direct_steered_sha"
+echo "INTENT-RECOVERY-TIE-43af"
+echo "HUMAN-STEERING-TIE-43af"
+echo "ACCEPTANCE-EVIDENCE-TIE-43af"
+echo "PLAN-STATIC-EVIDENCE-GREEN-43af"
+echo "DIRECT-MESSAGE-EVIDENCE-GREEN-43af"
+
+jq -n \
+  --arg experiment_id "catalog-plan-vs-direct-brief-v1" \
+  --arg source "$source_full" \
+  --arg sha "$actual_binary_sha" \
+  --arg initial_sha "$plan_initial_sha" \
+  --arg steered_sha "$plan_steered_sha" \
+  '{
+    schemaVersion: 1,
+    experimentId: $experiment_id,
+    runner: {source: $source, sha256: $sha},
+    modelCalls: 0,
+    remoteSourcesOfflineDuringRecovery: true,
+    storage: {plainFoldersSufficient: true, casRequired: false},
+    arms: {
+      A: {
+        transport: "catalog-plan",
+        initialIntentSha256: $initial_sha,
+        steeredIntentSha256: $steered_sha
+      },
+      B: {
+        transport: "st2-direct-messages",
+        initialIntentSha256: $initial_sha,
+        steeredIntentSha256: $steered_sha
+      }
+    },
+    outcomes: {
+      coldResume: {A: "pass", B: "pass", verdict: "tie"},
+      intentRecovery: {A: "pass", B: "pass", verdict: "tie"},
+      humanSteering: {A: "pass", B: "pass", verdict: "tie"},
+      acceptanceEvidence: {
+        A: "evaluator-only",
+        B: "evaluator-only",
+        verdict: "tie",
+        reason: "Neither read-only plan inspection nor direct-message delivery reports worker acceptance."
+      }
+    },
+    planOnly: {
+      staticIntentValidation: "pass",
+      provenanceResolution: "pass"
+    },
+    directOnly: {
+      messageDelivery: "pass",
+      replyLineage: "pass"
+    },
+    liveUnresolved: [
+      "correctness",
+      "coordination_messages",
+      "coordination_bytes",
+      "model_input_tokens",
+      "model_output_tokens",
+      "cost_usd",
+      "wall_duration"
+    ],
+    conclusion: "no-measured-advantage"
+  }' >"$receipt"
+
+jq -e '
+  .modelCalls == 0 and
+  .remoteSourcesOfflineDuringRecovery == true and
+  .storage.plainFoldersSufficient == true and
+  .storage.casRequired == false and
+  .outcomes.coldResume.verdict == "tie" and
+  .outcomes.intentRecovery.verdict == "tie" and
+  .outcomes.humanSteering.verdict == "tie" and
+  .outcomes.acceptanceEvidence.verdict == "tie" and
+  .planOnly.staticIntentValidation == "pass" and
+  .planOnly.provenanceResolution == "pass" and
+  .directOnly.messageDelivery == "pass" and
+  .directOnly.replyLineage == "pass" and
+  .conclusion == "no-measured-advantage"
+' "$receipt" >/dev/null
+
+echo "MODEL-FREE-COMPARISON-GREEN-43af"
+echo "PLAIN-FOLDER-NO-CAS-GREEN-43af"
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/runner.tsv b/cells/catalog-plan-vs-direct-brief/fixture/runner.tsv
new file mode 100644
index 0000000..2a807f5
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/runner.tsv
@@ -0,0 +1,9 @@
+field	value
+st2_pr	https://github.com/compoundingtech/st2/pull/115
+source_full	8a76b6e71355140e5b89cd9313fcfd88c82b5cad
+source_short	8a76b6e
+binary_sha256	214e08874720bc546d4adf7d7977e614237baf7989cc09f6932cd991f497a753
+source_gist_revision	5c1d1427c0556d95d13890e5c5086cd85b25d994
+runtime_scope	read-only-validate-list-show-inspect
+hosted_run	https://github.com/compoundingtech/st2/actions/runs/30835684680
+hosted_status	pass
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/scenario.tsv b/cells/catalog-plan-vs-direct-brief/fixture/scenario.tsv
new file mode 100644
index 0000000..221af70
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/scenario.tsv
@@ -0,0 +1,7 @@
+ordinal	event	arm_a_input	arm_b_input	invariant
+1	initial_delivery	catalog snapshot with version 0000	st2 message carrying brief 0000	byte-identical intent; neither delivery claims worker acceptance
+2	cold_restart	st2 plan show on local snapshot	st2 message read from local inbox	no volatile conversation or process state
+3	simulated_partition	local catalog only	local message inbox only	remote input source unavailable during recovery
+4	human_steering	catalog snapshot adds version 0001	second st2 message replies to brief 0000	byte-identical complete revised intent
+5	restart_after_steering	frontier 0001 with version 0000 retained	brief 0001 with inReplyTo 0000 retained	revision 0001 recovered and lineage remains inspectable
+6	model_free_verdict	static validation and provenance only	message delivery and reply lineage only	cold resume, intent recovery, and acceptance evidence tie; correctness, traffic, cost, and duration remain live-run endpoints
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/task-repo/README.md b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/README.md
new file mode 100644
index 0000000..093850f
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/README.md
@@ -0,0 +1,11 @@
+# receipt-report
+
+A dependency-free report core for append-only eval run receipts.
+
+Run the visible test suite with:
+
+```sh
+npm test
+```
+
+The task brief is delivered separately. Only `src/report.mjs` is in scope.
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/task-repo/package.json b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/package.json
new file mode 100644
index 0000000..a34c746
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/package.json
@@ -0,0 +1,9 @@
+{
+  "name": "receipt-report",
+  "version": "0.0.0",
+  "private": true,
+  "type": "module",
+  "scripts": {
+    "test": "node --test"
+  }
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/task-repo/src/report.mjs b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/src/report.mjs
new file mode 100644
index 0000000..7b5ae6e
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/src/report.mjs
@@ -0,0 +1,13 @@
+export function parseJsonLines(text) {
+  return text
+    .split("\n")
+    .filter(Boolean)
+    .map((line) => JSON.parse(line));
+}
+
+export function summarize(records) {
+  return records.map((record) => ({
+    cell: record.cell,
+    last_run: record.result,
+  }));
+}
diff --git a/cells/catalog-plan-vs-direct-brief/fixture/task-repo/test/report.test.mjs b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/test/report.test.mjs
new file mode 100644
index 0000000..f3cacc1
--- /dev/null
+++ b/cells/catalog-plan-vs-direct-brief/fixture/task-repo/test/report.test.mjs
@@ -0,0 +1,54 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+
+import { parseJsonLines, summarize } from "../src/report.mjs";
+
+test("reports the latest accepted pass separately from the last run", () => {
+  const records = parseJsonLines(`
+{"run_id":"r1","cell":"alpha","result":"PASS","score":"2/2"}
+{"run_id":"r2","cell":"alpha","result":"FAIL","score":"1/2"}
+{"run_id":"r3","cell":"beta","result":"FAIL","score":"0/1"}
+`);
+
+  assert.deepEqual(summarize(records), [
+    {
+      cell: "alpha",
+      accepted_pass: {
+        run_id: "r1",
+        cell: "alpha",
+        result: "PASS",
+        score: "2/2",
+      },
+      last_run: {
+        run_id: "r2",
+        cell: "alpha",
+        result: "FAIL",
+        score: "1/2",
+      },
+    },
+    {
+      cell: "beta",
+      accepted_pass: null,
+      last_run: {
+        run_id: "r3",
+        cell: "beta",
+        result: "FAIL",
+        score: "0/1",
+      },
+    },
+  ]);
+});
+
+test("ignores blank lines and sorts cells", () => {
+  const records = parseJsonLines(`
+
+{"run_id":"r1","cell":"zeta","result":"PASS"}
+
+{"run_id":"r2","cell":"alpha","result":"PASS"}
+`);
+
+  assert.deepEqual(
+    summarize(records).map(({ cell }) => cell),
+    ["alpha", "zeta"],
+  );
+});
diff --git a/evidence/harness-exclusions.tsv b/evidence/harness-exclusions.tsv
index 020ab44..f417fd9 100644
--- a/evidence/harness-exclusions.tsv
+++ b/evidence/harness-exclusions.tsv
@@ -1,4 +1,5 @@
 cell	surface	reason
+catalog-plan-vs-direct-brief	cell	deterministic plan-versus-brief fixture and recovery contract; no harness seat
 adopt-only-migration	cell	deterministic native adoption/replacement lifecycle probe; no harness seat
 agent-spec-resource-bindings	cell	deterministic native Agent Spec Resource-envelope probe; no harness seat
 context-resource-continuity	cell	deterministic native context/resource restart probe; no harness seat