Add model-free configured DING activity matrix

AGENT-SPEC.md

@@ -1,8 +1,8 @@
 # Canonical st2 agent specification
 
 This is the sole agent-authoring specification for this repository. It is pinned to st2
-[`9887b2842222def0838c2cd82e6c24c218f7efa6`](https://github.com/compoundingtech/st2/commit/9887b2842222def0838c2cd82e6c24c218f7efa6)
-(`0.1.0`, source `9887b28`). It documents the hand-authored KDL accepted at that commit. Do not infer
+[`c6846f6239329f0803142afc06c15a07b93937c1`](https://github.com/compoundingtech/st2/commit/c6846f6239329f0803142afc06c15a07b93937c1)
+(`0.1.0`, source `c6846f6`). It documents the hand-authored KDL accepted at that commit. Do not infer
 additional fields or commands from older corpus fixtures.
 
 st2 runs long-lived `service` agents made of interactive `pty` tasks and terminal-free `exec` tasks.
@@ -248,6 +248,20 @@ idempotently clears a restored inbox copy. New post-start unread work sends:
 Subject and sender are normalized to one bounded printable line. Delivery is one bracketed-paste PTY send,
 500 ms, then Return. DING does not inspect terminal pixels or classify modal state.
 
+The issue #57 experimental matrix separately pins corrected st2 PR #123 at
+`d7500b0fcad8bb268da9da96c0226d9caddbe305`. In that opt-in path, a compact
+`ding { adapter { argv ... } }` lowers to a generated exec sidecar whose argv is
+expanded only after the complete managed task environment is known. Core
+consumes a provider-neutral activity/input-buffer lease, requires exact PTY
+generation/activity sequence plus I/O revision, and performs one atomic
+guarded send only for fresh `idle` + `empty`. A generic `ding-control
+hook-owned` receipt removes exact unread filenames from PTY work when an
+already-occurring post-turn hook owns their next-context delivery. The
+integrated model-free A/B covers 11 durable messages: the aggressive arm makes
+six non-idle/colliding PTY writes and the configured hook-plus-idle arm makes
+zero. Real-provider fixtures, live-pane acceptance, merge, and release remain
+outside that evidence.
+
 Arrivals remain FIFO. `busy` and `away` do not suppress delivery; only fresh `dnd` does. Non-DND presence is
 refreshed every five minutes, while abandoned DND ages to derived `unknown` after 15 minutes and delivery
 resumes. Failed PTY sends retain the queue head. Startup backlog produces one generic unread-work DING rather
@@ -432,8 +446,7 @@ Inspect the declaration, every referenced template, and every workspace destinat
 materialization command. Materialization is byte-idempotent and does not imply hook installation. Starting
 the network is a separate, explicitly authorized action.
 
-For source `9887b28`, the accepted Linux executable has SHA256
-`d49d44fd4f3f6f655455c212353a469fefa956082bedf22163deb767d8a36a0d`; its published archive has SHA256
-`32ee103bd17ccb3e155ac63d816a3906c2470a3c98e3cc04b56e5a67138b9927`. `bin/check-corpus.sh` verifies
-the variable-age version contract, exact installed binary, embedded full source commit, strict semantic
+For source `c6846f6239329f0803142afc06c15a07b93937c1`, the accepted local-source Linux executable has
+SHA256 `2bba8d58be24250bc262f75f835ce2d780369add275774f3f2135c623d23d29c`. `bin/check-corpus.sh` verifies
+the variable-age version contract, exact installed binary, shared source-pin consistency, strict semantic
 validation, fixture resets, and the rest of the model-free corpus gate before an eval may run.

CATALOG.md

@@ -1,7 +1,7 @@
 <!-- Generated by bin/generate-catalog.sh from cell KDL and evidence/*.tsv. Do not edit. -->
 # Current eval catalog and evidence
 
-This is the sole current corpus surface: **45 included cells** (14 model-free,
+This is the sole current corpus surface: **51 included cells** (20 model-free,
 31 model-backed) and **4 retired exclusions**. Inventory fields are derived from the active
 KDL and executable judge scripts. Latest accepted PASS evidence remains distinct from the append-only run
 history: a failed last run is visible without being advertised as accepted evidence, and cells with no
@@ -18,7 +18,13 @@ harness-native loader plus canonical hook file.
 |---|---|---|---:|---|---|---:|---|---|
 | `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-adapter-fixture-contract` | model-free | — | 0 | none | `30s` | 7 | — | **NO STRUCTURED RUN** |
+| `ding-delivery-a-aggressive-control` | model-free | — | 0 | none | `45s` | 7 | — | **NO STRUCTURED RUN** |
+| `ding-delivery-pty-ab` | model-free | — | 0 | none | `120s` | 8 | — | **NO STRUCTURED RUN** |
+| `ding-delivery-st2-rich-ab` | model-free | — | 0 | none | `120s` | 10 | — | **NO STRUCTURED RUN** |
 | `ding-mode` | Claude | `claude-sonnet-5` / medium | 2 | medium | `1200s` | 4 | — | **NO STRUCTURED RUN** |
+| `ding-pty-activity-lease` | model-free | — | 0 | none | `60s` | 8 | — | **NO STRUCTURED RUN** |
+| `ding-pty-guarded-send` | model-free | — | 0 | none | `90s` | 8 | — | **NO STRUCTURED RUN** |
 | `ding-reply` | Claude | `claude-sonnet-5` / medium | 1 | low | `420s` | 2 | — | **NO STRUCTURED RUN** |
 | `docs` | Claude | `claude-sonnet-5` / medium | 3 | high | `1200s` | 5 | — | **NO STRUCTURED RUN** |
 | `exec-lifecycle-logging` | model-free | — | 0 | none | `90s` | 5 | — | **NO STRUCTURED RUN** |
@@ -82,6 +88,12 @@ while the last-run column makes a recorded failure distinct from a cell with no
 ## Harness hook exclusions
 
 - `context-resource-continuity` / `cell` — deterministic native context/resource restart probe; no harness seat
+- `ding-adapter-fixture-contract` / `cell` — deterministic fixture provenance and blocker contract; no bus-connected agent
+- `ding-delivery-a-aggressive-control` / `cell` — deterministic unconfigured DING ordinary-input control; no harness or model agent
+- `ding-delivery-st2-rich-ab` / `cell` — deterministic exact-source configured generic-adapter versus aggressive DING A/B; no harness or model agent
+- `ding-delivery-pty-ab` / `cell` — deterministic PTY-boundary aggressive-versus-guarded delivery A/B; no harness or model agent
+- `ding-pty-activity-lease` / `cell` — deterministic exact-source PTY activity-lease probe; no harness or model agent
+- `ding-pty-guarded-send` / `cell` — deterministic exact-source PTY compare-and-send race probe; no harness or model agent
 - `docs` / `judge:cold-reader` — one-shot offline Claude print grader; no bus identity, DING, or hook surface
 - `exec-lifecycle-logging` / `cell` — deterministic native exec lifecycle probe; no harness seat
 - `hook-integrity` / `cell` — deterministic native hook-installation probe; no live harness seat
@@ -97,7 +109,7 @@ while the last-run column makes a recorded failure distinct from a cell with no
 - `targeted-reconcile-isolation` / `cell` — deterministic selected reconcile isolation probe; no bus-connected agent
 - `two-networks-coexist` / `cell` — deterministic catalog-isolation probe; no harness seat
 
-These **15** rows are generated from structured exclusions. A cell-level row is required for
+These **21** rows are generated from structured exclusions. A cell-level row is required for
 every derived model-free cell. The one-shot docs cold reader is a model-backed grader, but not a bus-connected
 agent: it has no st2 identity, DING sidecar, or lifecycle hooks.
 

README.md

@@ -96,10 +96,26 @@ VERDICT: PASS
 ```
 
 Requirements are `st2 0.1.0` from source
-[`9887b28`](https://github.com/compoundingtech/st2/commit/9887b2842222def0838c2cd82e6c24c218f7efa6),
+[`c6846f6`](https://github.com/compoundingtech/st2/commit/c6846f6239329f0803142afc06c15a07b93937c1),
+using the accepted local-source Linux binary SHA256
+`2bba8d58be24250bc262f75f835ce2d780369add275774f3f2135c623d23d29c`;
 `pty`, Bash, Git, `jq`, Rust/Cargo for the pinned KDL parser gate, and Node for JavaScript fixtures. A paid cell
 also needs every harness named by its dry-run row.
 
+The issue #57 PTY draft experiments are separately pinned to activity PR #131
+`46c71d31c0d6daee43adf568061b2b84a65ae8c0` and stacked guarded-send PR #133
+`743ceb796a41a3282e31382575bff0d0e3826d59`. Run their model-free cells with
+`EVALS_PTY_PR133_ROOT` pointing to a clean exact-head checkout built by
+`npm ci && npm run build`; the cells reject any other Git head or package lock.
+
+The integrated configured-DING A/B additionally pins corrected st2 PR #123
+`d7500b0fcad8bb268da9da96c0226d9caddbe305` and its accepted local-source Linux
+release binary SHA256
+`705ad3ebd0bce497a4117c7c7993505c0579fc1f7df2421e0525a22208f6949f`.
+Set `EVALS_ST2_PR123_ROOT` to a clean exact-head checkout built with
+`cargo build --release --locked`; the cell launches the generated configured
+sidecar from that artifact and rejects a different source, lockfile, or binary.
+
 ## Cell layout
 
 ```text

bin/check-corpus.sh

@@ -5,26 +5,20 @@ set -euo pipefail
 repo_root="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
 cd "$repo_root"
 
-expected_source="9887b28"
-expected_source_full="9887b2842222def0838c2cd82e6c24c218f7efa6"
-expected_binary_sha256="d49d44fd4f3f6f655455c212353a469fefa956082bedf22163deb767d8a36a0d"
-expected_version_regex='^st2 0\.1\.0 — running from local source \(9887b28, .+ ago\)$'
+source bin/st2-pin.sh
+bin/check-st2-pin-consistency.sh
 st2_path="$(command -v st2)"
 actual_version="$(st2 --version)"
-[[ "$actual_version" =~ $expected_version_regex ]] || {
-  echo "FAIL: expected st2 0.1.0 from pinned source $expected_source, found $actual_version" >&2
+[[ "$actual_version" =~ $ST2_VERSION_REGEX ]] || {
+  echo "FAIL: expected st2 0.1.0 from pinned source $ST2_SOURCE_SHORT, found $actual_version" >&2
   exit 1
 }
 actual_binary_sha256="$(sha256sum "$st2_path" | awk '{ print $1 }')"
-[ "$actual_binary_sha256" = "$expected_binary_sha256" ] || {
-  echo "FAIL: expected st2 binary sha256 $expected_binary_sha256, found $actual_binary_sha256 at $st2_path" >&2
+[ "$actual_binary_sha256" = "$ST2_BINARY_SHA256" ] || {
+  echo "FAIL: expected st2 binary sha256 $ST2_BINARY_SHA256, found $actual_binary_sha256 at $st2_path" >&2
   exit 1
 }
-LC_ALL=C grep -aFq "$expected_source_full" "$st2_path" || {
-  echo "FAIL: st2 binary at $st2_path does not embed full pinned source $expected_source_full" >&2
-  exit 1
-}
-echo "PASS: pinned runner source $expected_source ($actual_version; sha256 $actual_binary_sha256)"
+echo "PASS: pinned runner source $ST2_SOURCE_FULL ($actual_version; sha256 $actual_binary_sha256)"
 
 mapfile -d '' shell_files < <(
   find bin cells -type f -name '*.sh' -not -path '*/_git/*' -print0 | sort -z

bin/check-preflight-safety.sh

@@ -20,13 +20,15 @@ reachable=(
   bin/check-preflight-safety.sh
   bin/check-run-history.sh
   bin/check-retired-surfaces.sh
+  bin/check-st2-pin-consistency.sh
   bin/check-st2-semantic.sh
   bin/check-vrs-scope-drift.sh
   bin/check-vrs-variations.sh
   bin/check-weird-git-setup.sh
   bin/corpus-inventory.sh
   bin/generate-catalog.sh
   bin/model-seat-inventory.sh
+  bin/st2-pin.sh
 )
 materializers=(
   cells/signal-rename/fixture/materialize.sh
@@ -68,6 +70,7 @@ expected_direct=(
   bin/check-preflight-safety.sh
   bin/check-retired-surfaces.sh
   bin/check-run-history.sh
+  bin/check-st2-pin-consistency.sh
   bin/check-st2-semantic.sh
   bin/check-vrs-scope-drift.sh
   bin/check-vrs-variations.sh

bin/check-st2-pin-consistency.sh

@@ -0,0 +1,21 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+repo_root="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$repo_root"
+source bin/st2-pin.sh
+
+grep -Fq "$ST2_SOURCE_FULL" AGENT-SPEC.md
+grep -Fq "$ST2_SOURCE_FULL" README.md
+grep -Fq "$ST2_BINARY_SHA256" AGENT-SPEC.md
+grep -Fq "$ST2_BINARY_SHA256" README.md
+grep -Fq 'source bin/st2-pin.sh' bin/check-corpus.sh
+
+if rg -n \
+  '9887b28|d49d44fd4f3f6f655455c212353a469fefa956082bedf22163deb767d8a36a0d' \
+  AGENT-SPEC.md README.md bin/check-corpus.sh bin/st2-pin.sh; then
+  echo "FAIL: active st2 pin surfaces retain the superseded source or binary hash" >&2
+  exit 1
+fi
+
+echo "PASS: active spec, README, and executable preflight share st2 $ST2_SOURCE_FULL / $ST2_BINARY_SHA256"

bin/st2-pin.sh

@@ -0,0 +1,6 @@
+#!/usr/bin/env bash
+
+ST2_SOURCE_SHORT="c6846f6"
+ST2_SOURCE_FULL="c6846f6239329f0803142afc06c15a07b93937c1"
+ST2_BINARY_SHA256="2bba8d58be24250bc262f75f835ce2d780369add275774f3f2135c623d23d29c"
+ST2_VERSION_REGEX='^st2 0\.1\.0 — running from local source \(c6846f6, .+ ago\)$'

cells/ding-adapter-fixture-contract/README.md

@@ -0,0 +1,59 @@
+# ding-adapter-fixture-contract
+
+Hermetic, model-free contract inventory for the optional DING activity-lease
+matrix tracked by [evals#57](https://github.com/compoundingtech/evals/issues/57).
+
+This cell validates only inputs that are safe to establish before the product
+seams exist:
+
+- nine shared, synthetic ASCII screen diagnostics;
+- exact content hashes and provenance metadata;
+- symmetric, explicitly blocked activity-event and post-turn-hook fixture
+  slots for the two maintained provider adapters;
+- matched A/B outcome and generic conditional-send race inventories; and
+- the external dependency/evidence ledger, including exact runnable-draft PTY
+  heads and the still-blocked adapter/core seams.
+
+The screen text and requested terminal diagnostics are never activity
+authority. No provider event payload or adapter interface is guessed here.
+The exact PTY activity and guarded-send product surfaces are now exercised by
+`ding-pty-activity-lease` and `ding-pty-guarded-send` against stacked draft head
+`743ceb796a41a3282e31382575bff0d0e3826d59`, which contains activity head
+`46c71d31c0d6daee43adf568061b2b84a65ae8c0`. Those runs are local exact-source
+experiment evidence, not release artifacts. Corrected st2 PR #123 at
+`d7500b0fcad8bb268da9da96c0226d9caddbe305` now supplies the generic configured
+sidecar, managed-task-env argv expansion, guarded-send integration, DND/error
+holds, typed ownership receipts, and generic hook-control ingress. The
+`ding-delivery-st2-rich-ab` cell exercises that draft end to end with a
+synthetic adapter; merge/release and real-provider acceptance remain blocked.
+
+The intended opt-in shape is an external adapter command selected by DING
+through a generic protocol. Provider commands may differ, but core st2 and the
+shared eval judge must not branch on provider identity. The generic
+selection/envelope and hook ownership contract are pinned by st2 PR #123.
+Provider-specific activity translation, composer evidence, and hook payload
+fixtures remain blocked on immutable owner artifacts.
+
+The A/B inventory freezes the intended comparison without pretending it can
+execute unfinished behavior. The conditional-send inventory names races and
+outcomes only: a provider adapter must establish fresh idle plus an empty
+composer, while PTY must atomically reject a DING candidate if its live
+generation or generic I/O/snapshot revision changed. PTY does not classify
+submit, newline, or human intent.
+
+## Validation boundary
+
+The scaffold cell, parser, semantic loader, fixture-reset, harness-exclusion,
+preflight-reachability, generated-catalog, and full corpus gates are
+model-free. The accepted successor runner is exact source
+`c6846f6239329f0803142afc06c15a07b93937c1` with local-source Linux binary
+SHA256 `2bba8d58be24250bc262f75f835ce2d780369add275774f3f2135c623d23d29c`.
+Do not rebuild or substitute a different binary to claim the full corpus gate.
+
+The PTY cells additionally require `EVALS_PTY_PR133_ROOT` to point to a
+clean exact-head checkout after `npm ci && npm run build`. They verify the Git
+head, activity-base ancestry, package-lock hash, and runtime artifact hashes in
+their receipts. The integrated cell also requires `EVALS_ST2_PR123_ROOT` at
+exact corrected head `d7500b0fcad8bb268da9da96c0226d9caddbe305` after
+`cargo build --release --locked` and pins binary SHA256
+`705ad3ebd0bce497a4117c7c7993505c0579fc1f7df2421e0525a22208f6949f`.

cells/ding-adapter-fixture-contract/ding-adapter-fixture-contract.kdl

@@ -0,0 +1,33 @@
+// Seam-independent fixture and provenance contract for the optional DING activity lease.
+eval {
+  copy "./fixture"
+  max-timeout "30s"
+
+  run "fixture-contract" {
+    command "bash ./check-fixtures.sh"
+  }
+
+  judges {
+    judge "DIAGNOSTICS — nine shared ASCII screens are exact, hashed, and non-authoritative" {
+      exec "grep -Fqx DIAGNOSTIC-FIXTURES-GREEN-57d1 $RUNS_DIR/fixture-contract.out"
+    }
+    judge "PARITY SLOTS — every declared provider reserves the same activity and post-turn hook cases" {
+      exec "grep -Fqx PROVIDER-SLOTS-GREEN-57d1 $RUNS_DIR/fixture-contract.out"
+    }
+    judge "A/B CONTRACT — matched workloads preserve durability while B eliminates active-turn PTY input" {
+      exec "grep -Fqx AB-INVENTORY-GREEN-57d1 $RUNS_DIR/fixture-contract.out"
+    }
+    judge "CONDITIONAL SEND — generic I/O races reject the DING candidate without terminal heuristics" {
+      exec "grep -Fqx CONDITIONAL-SEND-INVENTORY-GREEN-57d1 $RUNS_DIR/fixture-contract.out"
+    }
+    judge "PROVENANCE — every ready asset has reviewed metadata and every provider asset is explicitly blocked" {
+      exec "grep -Fqx PROVENANCE-INVENTORY-GREEN-57d1 $RUNS_DIR/fixture-contract.out"
+    }
+    judge "SEAMS — exact draft PTY/st2 surfaces and remaining provider blockers are classified without invention" {
+      exec "grep -Fqx PRODUCT-BLOCKERS-GREEN-57d1 $RUNS_DIR/fixture-contract.out && grep -Fqx PTY-DRAFT-SEAMS-GREEN-57d1 $RUNS_DIR/fixture-contract.out"
+    }
+    judge "MODEL FREE — the scaffold contains no provider payload, provider launch, network, or model call" {
+      exec "grep -Fqx HERMETIC-SKELETON-GREEN-57d1 $RUNS_DIR/fixture-contract.out"
+    }
+  }
+}

cells/ding-adapter-fixture-contract/fixture/ab-cases.tsv

@@ -0,0 +1,10 @@
+case_id	a_delivery	b_delivery	durable_expectation	fifo_expectation	model_calls	b_pty_bytes	b_collision_goal
+idle	aggressive-pty	bounded-idle-wake	eventual-exactly-once	preserve	0	exactly-one	zero
+active-turn	aggressive-pty	next-hook-context	eventual-exactly-once	preserve	0	zero	zero
+long-child	aggressive-pty	hold-until-authorized	eventual-exactly-once	preserve	0	zero	zero
+dnd	dnd-hold-then-deliver	dnd-hold-then-authorize	eventual-exactly-once	preserve	0	zero-while-held	zero
+stale-unknown	aggressive-pty	hold-until-authorized	eventual-exactly-once	preserve	0	zero	zero
+hook-failure	aggressive-pty	durable-hold-until-authorized	eventual-exactly-once	preserve	0	zero	zero
+crash-restart	restart-backlog	durable-hold-until-authorized	eventual-exactly-once	preserve	0	zero-before-authorization	zero
+compaction-clear	aggressive-pty	hold-until-authorized	eventual-exactly-once	preserve	0	zero	zero
+fifo-burst	aggressive-fifo	authorized-fifo	eventual-exactly-once	preserve	0	bounded-by-messages	zero