Estate audit — Wave 4: DYADT, post-action agent-claim verification (rein in LLMs)

.github/workflows/dyadt-verify.yml

@@ -0,0 +1,44 @@
+# SPDX-License-Identifier: MPL-2.0
+# dyadt-verify — DYADT dogfood. Verify this repo's CLAIMS.a2ml against primary
+# evidence, and run the DYADT conformance vector suite. If a claim this change
+# makes is false, the verifier REFUTES it and this job fails loudly.
+name: DYADT Verify
+
+on:
+  push:
+    branches: [ main, master ]
+  pull_request:
+    branches: [ main, master ]
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+jobs:
+  verify-claims:
+    name: Verify CLAIMS.a2ml + conformance
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          fetch-depth: 0  # need origin/main as the DYADT diff base
+
+      - name: DYADT conformance vectors
+        run: |
+          bash did-you-actually-do-that/spec/conformance/run-conformance.sh
+
+      - name: Verify this change's own claims (dogfood)
+        env:
+          DYADT_BASE: origin/${{ github.base_ref || 'main' }}
+        run: |
+          if [ -f CLAIMS.a2ml ]; then
+            bash scripts/verify-claims.sh CLAIMS.a2ml
+          else
+            echo "No CLAIMS.a2ml at root — nothing to verify."
+          fi

.machine_readable/REGISTRY.a2ml

@@ -20,7 +20,7 @@
 version = "1.0.0"
 generator = "scripts/build-registry.sh"
 hash_algorithm = "sha256(git ls-files -s <home>)  # local; external: recorded pin"
-entry_count = 31
+entry_count = 32
 
 [registry.streams]
 foundation  = "A2ML format family + K9 + contractiles (Stream 1)"
@@ -228,6 +228,15 @@ canonical_doc = "session-management-standards/README.adoc"
 source_hash = "sha256:4b1a97d2ef91578b2c262f1af5b93577cb36d53d234c330947538f4eafc01a6c"
 route = "continuity / verify / handover protocols"
 
+[[spec]]
+id = "did-you-actually-do-that"
+name = "DYADT — Did-You-Actually-Do-That"
+stream = "governance"
+home = "did-you-actually-do-that/"
+canonical_doc = "did-you-actually-do-that/README.adoc"
+source_hash = "sha256:445359ddcc92b56dfc8e8a3bdc16062439f1236b5fd0f42099113e7afa86d2e0"
+route = "post-action agent-claim verification (Tier 4 accountability)"
+
 [[spec]]
 id = "ensaid-config"
 name = "ENSAID Config"

.machine_readable/scorecards/did-you-actually-do-that.scorecard.a2ml

@@ -0,0 +1,87 @@
+# SPDX-License-Identifier: CC-BY-SA-4.0
+# did-you-actually-do-that.scorecard.a2ml
+# Hand-authored source. Regenerate the dashboard with: just scorecards
+# Schema: .machine_readable/scorecards/scorecard.schema.json
+
+[scorecard]
+spec_id = "did-you-actually-do-that"
+version = "1.0.0"
+assessed_date = "2026-07-03"
+assessor = "estate-audit"
+
+[[must]]
+id = "M1"
+text = "DYADT MUST define a typed, machine-checkable claim format."
+system = "did-you-actually-do-that/spec/CLAIM-FORMAT.adoc (normative claim classes + required fields)"
+status = "pass"
+evidence = "spec/CLAIM-FORMAT.adoc present; 7 claim classes and required-field table defined; example CLAIMS.a2ml shipped."
+effects = "Without a fixed claim format the parallel production verifier has no contract to build against."
+
+[[must]]
+id = "M2"
+text = "DYADT MUST ship a verifier that re-derives outcomes from primary evidence and can REFUTE a false claim."
+system = "scripts/verify-claims.sh, exercised by scripts/tests/wave4-dyadt-test.sh"
+status = "pass"
+evidence = "wave4-dyadt-test.sh (7/7) asserts a false command is REFUTED despite an honest-sounding statement; verifier never reads back the agent's evidence field."
+effects = "A verifier that trusted the statement would be theatre — the whole point is mechanical refutation."
+
+[[must]]
+id = "M3"
+text = "Verdicts MUST be exactly confirmed|refuted|unverifiable, and unverifiable MUST be loud (non-passing)."
+system = "scripts/verify-claims.sh (exit 1 on any refuted or, by default, unverifiable)"
+status = "pass"
+evidence = "wave4-dyadt-test.sh asserts refuted -> non-zero exit; all-confirmed -> exit 0; VERIFICATION-PROTOCOL.adoc fixes the three-verdict contract."
+effects = "Silent 'assumed pass' would reintroduce the false-green disease DYADT exists to cure."
+
+[[must]]
+id = "M4"
+text = "A conformance vector suite MUST exist so any verifier can be validated against shared ground truth."
+system = "did-you-actually-do-that/spec/conformance/run-conformance.sh (+ 6 vector pairs)"
+status = "pass"
+evidence = "run-conformance.sh passes 6/6 vectors covering confirmed/refuted/unverifiable/incompatible-verifier/manual-only."
+effects = "Without shared vectors the reference and production verifiers could silently diverge."
+
+[[must]]
+id = "M5"
+text = "The change introducing DYADT MUST dogfood it: a CLAIMS.a2ml verified in CI."
+system = ".github/workflows/dyadt-verify.yml runs scripts/verify-claims.sh CLAIMS.a2ml"
+status = "pass"
+evidence = "Root CLAIMS.a2ml ships 7 claims about this change; dyadt-verify.yml runs the verifier + conformance suite on push/PR."
+effects = "If a claim here were false, CI is refuted and fails loudly — the spec proves itself on itself."
+
+[[should]]
+id = "S1"
+text = "DYADT SHOULD specify an append-only, dual-signed consequence ledger and a per-actor confirmation rate."
+system = "did-you-actually-do-that/spec/CONSEQUENCE-LEDGER.adoc + .machine_readable/ledger/ format"
+status = "pass"
+evidence = "CONSEQUENCE-LEDGER.adoc defines the entry format, the confirmation-rate formula, and how Tier 3 MAY gate on it."
+effects = "Verification without memory cannot escalate on a repeatedly-over-claiming actor."
+
+[[should]]
+id = "S2"
+text = "The PLASMA naming collision SHOULD be resolved and recorded in canon."
+system = "did-you-actually-do-that/docs/NAMING-RESOLUTION.adoc + CANONICAL-NAMES.adoc entry"
+status = "pass"
+evidence = "NAMING-RESOLUTION.adoc splits DYADT (claim verification) from PLASMA (licence/exactness); a DYADT/PLASMA entry is added to CANONICAL-NAMES.adoc."
+effects = "Ambiguous 'grounded by PLASMA' references would keep entangling licence exactness with claim verification."
+
+[[should]]
+id = "S3"
+text = "The verifier SHOULD implement the network verifiers (ci-run, issue-state, pr-state) against real forge/CI APIs."
+system = "none — the reference verifier returns 'unverifiable: needs-network-verifier' for network classes"
+status = "fail"
+effects = "ci-green / issue-closed / pr-merged claims cannot be confirmed by the reference impl; the production verifier in hyperpolymath/did-you-actually-do-that must implement them (chartered)."
+
+[[could]]
+id = "C1"
+text = "Tier-3 contractiles COULD block pre_merge on an actor's DYADT confirmation rate."
+system = "none (ledger-gate.sh is illustrative in CONSEQUENCE-LEDGER.adoc, not implemented)"
+status = "aspirational"
+effects = "A live consequence gate is the end state; it depends on the production ledger writer existing first."
+
+[[could]]
+id = "C2"
+text = "The ledger COULD be a central estate-wide store rather than per-repo."
+system = "none"
+status = "aspirational"
+effects = "Cross-repo actor reputation needs a central ledger; a reach target, not a current gap."

CANONICAL-NAMES.adoc

@@ -62,6 +62,23 @@ DEPRECATED NAMES — DO NOT REINTRODUCE (estate mandate 2026-06-30):
   mismatch of broken parts adding to the pain of the agentic experience. It is
   not an institution.
 
+| **PLASMA**
+| Palimpsest licence / exactness tooling (only)
+| "the claim-grounder", "the DYADT engine", "agent-claim verifier"
+| PLASMA (`palimpsest-plasma`) is licence & exactness tooling *only* — the
+  authority for licence identity, invariants, and equivalence. It is NOT an
+  agent-claim verifier. Post-action verification of an agent's asserted
+  outcomes is *DYADT* (see below), which subsumes the former "claim-grounder"
+  concept and drops the PLASMA name. See
+  `did-you-actually-do-that/docs/NAMING-RESOLUTION.adoc`.
+
+| **DYADT**
+| Did-You-Actually-Do-That
+| "PLASMA", "the grounder"
+| Post-action agent-claim verification (Tier 4 of the accountability pipeline):
+  takes an agent's asserted outcomes and mechanically confirms/refutes each
+  against primary evidence. Home: `did-you-actually-do-that/`.
+
 |===
 
 == Deprecated names — do not reintroduce (MANDATE, 2026-06-30)

CLAIMS.a2ml

@@ -0,0 +1,68 @@
+# SPDX-License-Identifier: CC-BY-SA-4.0
+# CLAIMS.a2ml — DYADT dogfood: what this change claims to have done.
+# Verified in CI by scripts/verify-claims.sh (did-you-actually-do-that/).
+# Every claim below is re-derived from primary evidence, not from this text.
+# Claims use content markers (not "created") so they stay true after the change
+# merges into the base branch — a claim must describe a durable fact.
+
+[claims]
+schema = "dyadt/claim@1"
+session = "session_0114ps6mY5jAH4Sz"
+actor = "claude-opus-4-8"
+emitted_commit = "HEAD"
+
+[[claim]]
+id = "C1"
+claim_class = "file-changed"
+statement = "the DYADT front door defines the Tier-4 accountability pipeline"
+target = "did-you-actually-do-that/README.adoc"
+expect = "contains:Post-Action Claim Verification"
+verifier = "git-diff"
+
+[[claim]]
+id = "C2"
+claim_class = "file-changed"
+statement = "the claim format spec defines typed claim classes"
+target = "did-you-actually-do-that/spec/CLAIM-FORMAT.adoc"
+expect = "contains:claim_class"
+verifier = "git-diff"
+
+[[claim]]
+id = "C3"
+claim_class = "file-changed"
+statement = "the verification protocol fixes the three-verdict contract"
+target = "did-you-actually-do-that/spec/VERIFICATION-PROTOCOL.adoc"
+expect = "contains:unverifiable"
+verifier = "git-diff"
+
+[[claim]]
+id = "C4"
+claim_class = "file-changed"
+statement = "the consequence ledger defines a per-actor confirmation rate"
+target = "did-you-actually-do-that/spec/CONSEQUENCE-LEDGER.adoc"
+expect = "contains:confirmation rate"
+verifier = "git-diff"
+
+[[claim]]
+id = "C5"
+claim_class = "file-changed"
+statement = "the reference verifier can emit an unverifiable verdict"
+target = "scripts/verify-claims.sh"
+expect = "contains:unverifiable"
+verifier = "git-diff"
+
+[[claim]]
+id = "C6"
+claim_class = "command-ran"
+statement = "the DYADT conformance vector suite passes"
+target = "bash did-you-actually-do-that/spec/conformance/run-conformance.sh"
+expect = "exit==0"
+verifier = "command-transcript"
+
+[[claim]]
+id = "C7"
+claim_class = "test-passed"
+statement = "the verifier refutes a false claim and confirms a true one (self-check)"
+target = "bash scripts/tests/wave4-dyadt-test.sh"
+expect = "exit==0"
+verifier = "command-transcript"

COMPLIANCE-DASHBOARD.md

@@ -42,6 +42,7 @@
 | `toolchain-readiness-grades` | ❌ gap | 1/5 | 2/4 | 0/3 | 83% | 2026-07-03 |
 | `rhodium-standard-repositories` | ❌ gap | 2/3 | 1/2 | 0/1 | 50% | 2026-07-03 |
 | `session-management-standards` | ❌ gap | 1/5 | 1/4 | 0/3 | 41% | 2026-07-03 |
+| `did-you-actually-do-that` | ✅ met | 5/5 | 2/3 | 0/2 | 90% | 2026-07-03 |
 | `ensaid-config` | ❌ gap | 0/5 | 0/3 | 0/3 | 90% | 2026-07-03 |
 | `accessibility` | ❌ gap | 2/5 | 0/5 | 0/3 | 100% | 2026-07-03 |
 | `publication-pre-flight` | ❌ gap | 0/5 | 0/4 | 0/2 | 36% | 2026-07-03 |
@@ -51,10 +52,10 @@
 
 ## Estate rollup
 
-- **Specs registered (local):** 28
-- **Specs with a scorecard:** 28 / 28
-- **MUST requirements:** 41 passing / 138 total (74 failing)
-- **Estate systems coverage:** 66% of 328 graded requirements have a mechanical check
+- **Specs registered (local):** 29
+- **Specs with a scorecard:** 29 / 29
+- **MUST requirements:** 46 passing / 143 total (74 failing)
+- **Estate systems coverage:** 67% of 338 graded requirements have a mechanical check
 
 ## How this dashboard stays honest
 

Justfile

@@ -73,6 +73,18 @@ scorecards-check:
 scorecards-check-strict:
     @bash scripts/build-scorecards.sh --check --strict
 
+# DYADT: verify a CLAIMS.a2ml against primary evidence (default: root CLAIMS.a2ml)
+verify-claims path="CLAIMS.a2ml":
+    @bash scripts/verify-claims.sh "{{path}}"
+
+# DYADT: run the conformance vector suite
+dyadt-conformance:
+    @bash did-you-actually-do-that/spec/conformance/run-conformance.sh
+
+# DYADT regression test (confirm/refute/unverifiable + guards)
+dyadt-test:
+    @bash scripts/tests/wave4-dyadt-test.sh
+
 # Aggregate compliance gate: registry drift is the HARD gate (registry-check,
 # a hard dep). The RSR self-audit is INFORMATIONAL — a monorepo is not expected
 # to score Gold — but a *broken* audit (exit 4 / unexpected) must fail loudly

TOPOLOGY.md

@@ -9,7 +9,7 @@
 > It cannot freeze: every regeneration re-reads ground truth. Do not edit by hand.
 
 - **Phase:** active   |  **Maturity:** experimental   |  **STATE last-updated:** 2026-06-03T00:00:00Z
-- **Registry entries:** 31 specs across 6 streams
+- **Registry entries:** 32 specs across 6 streams
 - **Front door:** human → [README.adoc](README.adoc); machine → [0-AI-MANIFEST.a2ml](0-AI-MANIFEST.a2ml)
 - **Registry:** [.machine_readable/REGISTRY.a2ml](.machine_readable/REGISTRY.a2ml) (index + source hashes) · prose: [REGISTRY.adoc](REGISTRY.adoc)
 
@@ -55,6 +55,7 @@
 |---|---|---|
 | RSR — Rhodium Standard Repositories | [`rhodium-standard-repositories/`](rhodium-standard-repositories/) | the repository-compliance standard every repo is graded against |
 | Session Management Standards | [`session-management-standards/`](session-management-standards/) | continuity / verify / handover protocols |
+| DYADT — Did-You-Actually-Do-That | [`did-you-actually-do-that/`](did-you-actually-do-that/) | post-action agent-claim verification (Tier 4 accountability) |
 | ENSAID Config | [`ensaid-config/`](ensaid-config/) | the ensaid configuration standard |
 | Accessibility Standard | [`accessibility/`](accessibility/) | estate accessibility requirements |
 | Publication Pre-Flight | [`publication-pre-flight/`](publication-pre-flight/) | submission gate (HOL + Zenodo checklists) |