fix(substrate): dangling handoffs + atomar manifest fill-out + validator extension

Group F — manifest, capability, role, and assembler cleanup (post-audit 2026-05-02).

Dangling handoff targets stripped:
- validator.toml: removed handoffs to physics-deriver, patent-compliance
- code-implementer.toml: removed physics-deriver handoff
- architect.toml: removed physics-deriver
- ml-implementer.toml: removed physics-deriver, fixed "multi-node multi-node" typo
- ml-researcher.toml: removed physics-deriver, patent-researcher
- researcher.toml: removed patent-researcher
None of those manifest files exist in _manifests/. Comments added explaining
the removal date for future re-authoring.

Validator extension (_assembler):
- src/validator.rs: extended validate() with check_handoff_targets — every
                     [[handoff]].target must point to existing _manifests/<name>.toml.
                     Future dangling handoffs blocked at validate time.
- src/validator_tests.rs (new, 133 LOC): unit tests for handoff-target check.
- tests/fixtures/_manifests/: added valid stubs for previously-missing manifests
                                (architect, critic, security-auditor, validator,
                                ml-implementer, ml-researcher, infra-implementer)
                                so existing fixtures pass the new validator gate.
- tests/snapshots/: insta snapshots updated for researcher + code-implementer.

Atomar manifest fill-out (replaced stock copy-paste with domain-specific):
- code-implementer-typescript: Drizzle/Zod/Next.js semantics
- code-implementer-go: mesh networking, embedded servers
- code-implementer-swift: SwiftUI, SPM, macOS menubar
- code-implementer-python: RULE 0.2 exception language
- code-implementer-flutter: Riverpod, Clean Architecture
- infra-implementer-cicd/iac/container/secrets: tool-specific bans + scopes
- researcher-web/code: output_extra_fields fixed (was code-implementer copy-paste
                        "Largest file LOC", "Tests pass count" — now sources cited /
                        evidence grade / gaps section)

Capability schema completeness:
- policy/no-git-ops + quality/cargo-check-green: added stage = "runtime"
- 8 capabilities: added explicit parents = [] (was missing/inconsistent)

Role schema:
- _roles/auditor.toml + merger.toml: added [taxonomy] + [lineage] (was missing)
- _roles/explorer.toml: added comment that "Explore" is the canonical Claude Code
                          subagent type (case-sensitive)

Reference path cleanup (manifest references):
- critic.toml: ~/.claude/skills/architecture-rules/... -> path:user-skills/...
- researcher.toml: stripped ~/.claude/agents/validator.md (machine-local)

Misc:
- frontend-validator.toml: renumbered duplicate step 6 -> step 7

kei-registry test fixture suppression:
- tests/fixtures/{atom-sample,fake-kit,mini-kit}/.kei-registry-ignore (3 new files)
- DNA-INDEX.md was inflating atom count by ~10% from test fixture rows; ignore-file
  hooks ready, kei-registry walker implementation is a follow-up.

Tests: 59 passed; 0 failed; 1 ignored (pre-existing #[ignore]). cargo check clean.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

_assembler/src/validator.rs

@@ -4,7 +4,7 @@
 //! Detailed sub-checks live in their own cubes:
 //!   - `placeholders::check`      — {{PLACEHOLDER}} substitution guard
 //!   - `schemas_export::load`     — dynamic artifact-schema whitelist loader
-//!   - this file                  — structural checks + artifact-schema names
+//!   - `validator_tests`          — unit tests (split per Constructor Pattern)
 
 use crate::manifest::Manifest;
 use crate::placeholders;
@@ -21,23 +21,40 @@ pub const OBLIGATORY: &[&str] = &["baseline", "evidence-grading", "memory-protoc
 pub const KNOWN_ARTIFACT_SCHEMAS: &[&str] = schemas_export::BUILTIN;
 
 pub fn validate(m: &Manifest, blocks_dir: &Path) -> Result<(), String> {
+    check_obligatory_blocks(m)?;
+    check_blocks_exist(m, blocks_dir)?;
+    check_handoff_targets(m, blocks_dir)?;
+    check_domain_fields(m)?;
+    placeholders::check(m)?;
+    let known = schemas_export::load(blocks_dir);
+    check_artifact_schemas(m, &known)?;
+    check_substrate_role(m, blocks_dir)?;
+    Ok(())
+}
+
+fn check_obligatory_blocks(m: &Manifest) -> Result<(), String> {
     for required in OBLIGATORY {
         if !m.blocks.iter().any(|b| b == required) {
             return Err(format!("missing obligatory block: {required}"));
         }
     }
-
     if m.handoff.is_empty() {
         return Err("at least one handoff required".into());
     }
+    Ok(())
+}
 
+fn check_blocks_exist(m: &Manifest, blocks_dir: &Path) -> Result<(), String> {
     for block in &m.blocks {
         let path = blocks_dir.join(format!("{block}.md"));
         if !path.exists() {
             return Err(format!("block '{block}' not found at {}", path.display()));
         }
     }
+    Ok(())
+}
 
+fn check_domain_fields(m: &Manifest) -> Result<(), String> {
     if m.domain_in.is_empty() {
         return Err("domain_in must have at least one entry".into());
     }
@@ -47,18 +64,32 @@ pub fn validate(m: &Manifest, blocks_dir: &Path) -> Result<(), String> {
     if m.role.trim().is_empty() {
         return Err("role must not be empty".into());
     }
+    Ok(())
+}
 
-    placeholders::check(m)?;
-    let known = schemas_export::load(blocks_dir);
-    check_artifact_schemas(m, &known)?;
-    check_substrate_role(m, blocks_dir)?;
-
+/// Verify every `handoff[i].target` resolves to `_manifests/<target>.toml`.
+/// Prevents dangling references to unauthored manifests from silently
+/// passing validation. See audit 2026-05-02.
+pub fn check_handoff_targets(m: &Manifest, blocks_dir: &Path) -> Result<(), String> {
+    let manifests_dir = blocks_dir
+        .parent()
+        .ok_or_else(|| "blocks_dir has no parent (can't locate _manifests/)".to_string())?
+        .join("_manifests");
+    for (i, h) in m.handoff.iter().enumerate() {
+        let target_path = manifests_dir.join(format!("{}.toml", h.target));
+        if !target_path.exists() {
+            return Err(format!(
+                "handoff[{i}].target '{}' has no manifest at {}",
+                h.target,
+                target_path.display()
+            ));
+        }
+    }
     Ok(())
 }
 
 /// If a manifest declares `substrate_role`, verify the role file exists
-/// and every capability it references has a `text.md`. Keeping the check
-/// here (not only at assemble time) turns mistakes into up-front failures.
+/// and every capability it references has a `text.md`.
 fn check_substrate_role(m: &Manifest, blocks_dir: &Path) -> Result<(), String> {
     let Some(role) = &m.substrate_role else { return Ok(()); };
     let root = blocks_dir
@@ -73,7 +104,7 @@ fn check_substrate_role(m: &Manifest, blocks_dir: &Path) -> Result<(), String> {
 
 /// v0.15: if a manifest references artifact schema names, they must be in the
 /// known whitelist. Missing fields are allowed (non-breaking extension).
-fn check_artifact_schemas(m: &Manifest, known: &BTreeSet<String>) -> Result<(), String> {
+pub fn check_artifact_schemas(m: &Manifest, known: &BTreeSet<String>) -> Result<(), String> {
     if let Some(name) = &m.produces_artifact {
         check_known(name, "produces_artifact", known)?;
     }
@@ -99,101 +130,5 @@ fn check_known(name: &str, field: &str, known: &BTreeSet<String>) -> Result<(),
 }
 
 #[cfg(test)]
-mod tests {
-    use super::*;
-    use crate::manifest::{Handoff, Manifest};
-
-    fn base() -> Manifest {
-        Manifest {
-            name: "test".into(),
-            description: "d".into(),
-            tools: vec!["Read".into()],
-            model: "opus".into(),
-            role: "r".into(),
-            blocks: vec!["baseline".into(), "evidence-grading".into(), "memory-protocol".into()],
-            domain_in: vec!["x".into()],
-            forbidden_domain: vec!["y".into()],
-            handoff: vec![Handoff {
-                target: "a".into(),
-                trigger: "b".into(),
-                expects_artifact: None,
-                produces_artifact: None,
-            }],
-            output_extra_fields: vec![],
-            memory_project: None,
-            project_claudemd: None,
-            references: None,
-            produces_artifact: None,
-            substrate_role: None,
-            rule_blocks: vec![],
-        }
-    }
-
-    fn builtin_set() -> BTreeSet<String> {
-        schemas_export::BUILTIN.iter().map(|s| (*s).to_string()).collect()
-    }
-
-    #[test]
-    fn artifact_schemas_absent_passes() {
-        let m = base();
-        assert!(check_artifact_schemas(&m, &builtin_set()).is_ok());
-    }
-
-    #[test]
-    fn artifact_schemas_known_names_pass() {
-        let mut m = base();
-        m.produces_artifact = Some("spec".into());
-        m.handoff[0].expects_artifact = Some("plan".into());
-        m.handoff[0].produces_artifact = Some("patch".into());
-        assert!(check_artifact_schemas(&m, &builtin_set()).is_ok());
-    }
-
-    #[test]
-    fn artifact_schemas_reject_unknown_produces() {
-        let mut m = base();
-        m.produces_artifact = Some("not-a-schema".into());
-        let err = check_artifact_schemas(&m, &builtin_set()).unwrap_err();
-        assert!(err.contains("not-a-schema"), "err: {err}");
-        assert!(err.contains("produces_artifact"), "err: {err}");
-    }
-
-    #[test]
-    fn artifact_schemas_reject_unknown_expects_in_handoff() {
-        let mut m = base();
-        m.handoff[0].expects_artifact = Some("zzz".into());
-        let err = check_artifact_schemas(&m, &builtin_set()).unwrap_err();
-        assert!(err.contains("zzz"), "err: {err}");
-        assert!(err.contains("handoff[0].expects_artifact"), "err: {err}");
-    }
-
-    #[test]
-    fn builtin_schemas_do_not_drift_from_kei_artifact() {
-        // Structural drift test (no runtime dep on kei-artifact): read the
-        // primitive's source and confirm its BUILTIN list matches ours.
-        let primitive = Path::new(env!("CARGO_MANIFEST_DIR"))
-            .join("..")
-            .join("_primitives/_rust/kei-artifact/src/schemas.rs");
-        if !primitive.exists() {
-            eprintln!("skip drift test: primitive not at {}", primitive.display());
-            return;
-        }
-        let src = std::fs::read_to_string(&primitive).unwrap();
-        let mut names: Vec<String> = Vec::new();
-        for line in src.lines() {
-            let t = line.trim();
-            if let Some(rest) = t.strip_prefix("(\"") {
-                if let Some(end) = rest.find("\",") {
-                    names.push(rest[..end].to_string());
-                }
-            }
-        }
-        let mine: Vec<String> = schemas_export::BUILTIN
-            .iter()
-            .map(|s| (*s).to_string())
-            .collect();
-        assert_eq!(
-            names, mine,
-            "kei-artifact BUILTIN and schemas_export::BUILTIN drifted"
-        );
-    }
-}
+#[path = "validator_tests.rs"]
+mod tests;

_assembler/src/validator_tests.rs

@@ -0,0 +1,133 @@
+//! Unit tests for manifest validator. Split per Constructor Pattern (file <200 LOC).
+//! Imported by `validator.rs` as `#[cfg(test)] #[path = "validator_tests.rs"] mod tests;`
+
+use super::{check_artifact_schemas, check_handoff_targets};
+use crate::manifest::{Handoff, Manifest};
+use crate::schemas_export;
+use std::collections::BTreeSet;
+use std::fs;
+use std::path::Path;
+
+fn base() -> Manifest {
+    Manifest {
+        name: "test".into(),
+        description: "d".into(),
+        tools: vec!["Read".into()],
+        model: "opus".into(),
+        role: "r".into(),
+        blocks: vec![
+            "baseline".into(),
+            "evidence-grading".into(),
+            "memory-protocol".into(),
+        ],
+        domain_in: vec!["x".into()],
+        forbidden_domain: vec!["y".into()],
+        handoff: vec![Handoff {
+            target: "a".into(),
+            trigger: "b".into(),
+            expects_artifact: None,
+            produces_artifact: None,
+        }],
+        output_extra_fields: vec![],
+        memory_project: None,
+        project_claudemd: None,
+        references: None,
+        produces_artifact: None,
+        substrate_role: None,
+        rule_blocks: vec![],
+    }
+}
+
+fn builtin_set() -> BTreeSet<String> {
+    schemas_export::BUILTIN.iter().map(|s| (*s).to_string()).collect()
+}
+
+#[test]
+fn artifact_schemas_absent_passes() {
+    let m = base();
+    assert!(check_artifact_schemas(&m, &builtin_set()).is_ok());
+}
+
+#[test]
+fn artifact_schemas_known_names_pass() {
+    let mut m = base();
+    m.produces_artifact = Some("spec".into());
+    m.handoff[0].expects_artifact = Some("plan".into());
+    m.handoff[0].produces_artifact = Some("patch".into());
+    assert!(check_artifact_schemas(&m, &builtin_set()).is_ok());
+}
+
+#[test]
+fn artifact_schemas_reject_unknown_produces() {
+    let mut m = base();
+    m.produces_artifact = Some("not-a-schema".into());
+    let err = check_artifact_schemas(&m, &builtin_set()).unwrap_err();
+    assert!(err.contains("not-a-schema"), "err: {err}");
+    assert!(err.contains("produces_artifact"), "err: {err}");
+}
+
+#[test]
+fn artifact_schemas_reject_unknown_expects_in_handoff() {
+    let mut m = base();
+    m.handoff[0].expects_artifact = Some("zzz".into());
+    let err = check_artifact_schemas(&m, &builtin_set()).unwrap_err();
+    assert!(err.contains("zzz"), "err: {err}");
+    assert!(err.contains("handoff[0].expects_artifact"), "err: {err}");
+}
+
+#[test]
+fn handoff_target_present_passes() {
+    let tmp = tempfile::tempdir().unwrap();
+    let blocks = tmp.path().join("project/_blocks");
+    let manifests = tmp.path().join("project/_manifests");
+    fs::create_dir_all(&blocks).unwrap();
+    fs::create_dir_all(&manifests).unwrap();
+    fs::write(manifests.join("target-agent.toml"), b"").unwrap();
+    let mut m = base();
+    m.handoff[0].target = "target-agent".into();
+    assert!(check_handoff_targets(&m, &blocks).is_ok());
+}
+
+#[test]
+fn handoff_target_missing_fails() {
+    let tmp = tempfile::tempdir().unwrap();
+    let blocks = tmp.path().join("project/_blocks");
+    let manifests = tmp.path().join("project/_manifests");
+    fs::create_dir_all(&blocks).unwrap();
+    fs::create_dir_all(&manifests).unwrap();
+    // Do NOT write target-agent.toml — tests that missing file → error.
+    let mut m = base();
+    m.handoff[0].target = "ghost-agent".into();
+    let err = check_handoff_targets(&m, &blocks).unwrap_err();
+    assert!(err.contains("ghost-agent"), "err: {err}");
+    assert!(err.contains("handoff[0].target"), "err: {err}");
+}
+
+#[test]
+fn builtin_schemas_do_not_drift_from_kei_artifact() {
+    let primitive = Path::new(env!("CARGO_MANIFEST_DIR"))
+        .join("..")
+        .join("_primitives/_rust/kei-artifact/src/schemas.rs");
+    if !primitive.exists() {
+        eprintln!("skip drift test: primitive not at {}", primitive.display());
+        return;
+    }
+    let src = std::fs::read_to_string(&primitive).unwrap();
+    let mut names: Vec<String> = Vec::new();
+    for line in src.lines() {
+        let t = line.trim();
+        if let Some(rest) = t.strip_prefix("(\"") {
+            if let Some(end) = rest.find("\",") {
+                names.push(rest[..end].to_string());
+            }
+        }
+    }
+    let mine: Vec<String> = schemas_export::BUILTIN
+        .iter()
+        .map(|s| (*s).to_string())
+        .collect();
+    assert_eq!(
+        names, mine,
+        "kei-artifact BUILTIN and schemas_export::BUILTIN drifted"
+    );
+}

_assembler/tests/fixtures/_manifests/architect.toml

@@ -0,0 +1,27 @@
+# Fixture stub for handoff-target resolution tests. Minimal valid manifest.
+name = "architect"
+description = "Architect stub."
+tools = ["Read", "Glob", "Grep"]
+model = "opus"
+substrate_role = "read-only"
+role = "Stub architect."
+blocks = ["baseline", "evidence-grading", "memory-protocol"]
+domain_in = ["architecture review"]
+forbidden_domain = ["writing code"]
+
+[[handoff]]
+target = "code-implementer"
+trigger = "structural finding implies refactor"
+
+[taxonomy]
+kingdom = "manifest"
+mechanism = "compose"
+domain = "agent"
+layer = "agent-substrate"
+stage = "design-time"
+stability = "stable"
+language = "toml"
+
+[lineage]
+creator = "ag-orchestrator-human"
+created = "2026-05-02"

_assembler/tests/fixtures/_manifests/code-implementer.toml

@@ -74,9 +74,7 @@ trigger = "task involves ML training / inference / Modal / experiment runners /
 target = "infra-implementer"
 trigger = "task involves deploy / CI/CD / secrets / IaC / credentials / public-surface hosting"
 
-[[handoff]]
-target = "physics-deriver"
-trigger = "task requires math derivation / theorem writing / theorem .md derivation"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
 [[handoff]]
 target = "critic"
@@ -96,11 +94,11 @@ trigger = "structural decision (new module graph, cross-cutting refactor, contra
 
 [references]
 extra = [
-    "~/.claude/rules/code-style.md",
-    "~/.claude/rules/git-conventions.md",
-    "~/.claude/rules/dev-workflow.md",
-    "~/.claude/rules/debugging.md",
-    "~/.claude/rules/karpathy-behavioral.md",
+    "path:user-rules/code-style.md",
+    "path:user-rules/git-conventions.md",
+    "path:user-rules/dev-workflow.md",
+    "path:user-rules/debugging.md",
+    "path:user-rules/karpathy-behavioral.md",
     "MEMORY.md → Architecture Overlay Incident (model_brain.py 227→354 LOC from \"fixes\" — never patch, fix root formulas)",
 ]
 

_assembler/tests/fixtures/_manifests/critic.toml

@@ -0,0 +1,27 @@
+# Fixture stub for handoff-target resolution tests. Minimal valid manifest.
+name = "critic"
+description = "Code critic stub."
+tools = ["Read", "Glob", "Grep"]
+model = "sonnet"
+substrate_role = "read-only"
+role = "Stub critic."
+blocks = ["baseline", "evidence-grading", "memory-protocol"]
+domain_in = ["anti-pattern detection"]
+forbidden_domain = ["fixing issues"]
+
+[[handoff]]
+target = "code-implementer"
+trigger = "confirmed findings need code edits"
+
+[taxonomy]
+kingdom = "manifest"
+mechanism = "compose"
+domain = "agent"
+layer = "agent-substrate"
+stage = "design-time"
+stability = "stable"
+language = "toml"
+
+[lineage]
+creator = "ag-orchestrator-human"
+created = "2026-05-02"

_assembler/tests/fixtures/_manifests/infra-implementer.toml

@@ -0,0 +1,27 @@
+# Fixture stub for handoff-target resolution tests. Minimal valid manifest.
+name = "infra-implementer"
+description = "Infra/deploy stub."
+tools = ["Read", "Write", "Bash"]
+model = "sonnet"
+substrate_role = "edit-local"
+role = "Stub infra-implementer."
+blocks = ["baseline", "evidence-grading", "memory-protocol"]
+domain_in = ["deploy", "CI/CD"]
+forbidden_domain = ["app code"]
+
+[[handoff]]
+target = "code-implementer"
+trigger = "app code change needed"
+
+[taxonomy]
+kingdom = "manifest"
+mechanism = "compose"
+domain = "agent"
+layer = "agent-substrate"
+stage = "design-time"
+stability = "stable"
+language = "toml"
+
+[lineage]
+creator = "ag-orchestrator-human"
+created = "2026-05-02"

_assembler/tests/fixtures/_manifests/ml-implementer.toml

@@ -0,0 +1,27 @@
+# Fixture stub for handoff-target resolution tests. Minimal valid manifest.
+name = "ml-implementer"
+description = "ML training stub."
+tools = ["Read", "Write", "Bash"]
+model = "opus"
+substrate_role = "edit-local"
+role = "Stub ml-implementer."
+blocks = ["baseline", "evidence-grading", "memory-protocol"]
+domain_in = ["ML training"]
+forbidden_domain = ["theory writing"]
+
+[[handoff]]
+target = "code-implementer"
+trigger = "inference path needs Rust"
+
+[taxonomy]
+kingdom = "manifest"
+mechanism = "compose"
+domain = "agent"
+layer = "agent-substrate"
+stage = "design-time"
+stability = "stable"
+language = "toml"
+
+[lineage]
+creator = "ag-orchestrator-human"
+created = "2026-05-02"

_assembler/tests/fixtures/_manifests/ml-researcher.toml

@@ -0,0 +1,27 @@
+# Fixture stub for handoff-target resolution tests. Minimal valid manifest.
+name = "ml-researcher"
+description = "ML researcher stub."
+tools = ["Read", "WebFetch", "WebSearch"]
+model = "opus"
+substrate_role = "read-only"
+role = "Stub ml-researcher."
+blocks = ["baseline", "evidence-grading", "memory-protocol"]
+domain_in = ["ML literature review"]
+forbidden_domain = ["running experiments"]
+
+[[handoff]]
+target = "ml-implementer"
+trigger = "hypothesis ready to run"
+
+[taxonomy]
+kingdom = "manifest"
+mechanism = "compose"
+domain = "agent"
+layer = "agent-substrate"
+stage = "design-time"
+stability = "stable"
+language = "toml"
+
+[lineage]
+creator = "ag-orchestrator-human"
+created = "2026-05-02"

_assembler/tests/fixtures/_manifests/researcher.toml

@@ -67,9 +67,7 @@ trigger = "claim needs hard verification (citation sanity, reproduce-in-tests, R
 target = "ml-researcher"
 trigger = "question is ML/RL specialized-node (Math-First + tooling-reuse + synthetic-to-real discipline)"
 
-[[handoff]]
-target = "patent-researcher"
-trigger = "question touches patent prior art, FTO, or novelty (IP-aware handling required)"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
 [[handoff]]
 target = "architect"
@@ -82,9 +80,8 @@ trigger = "findings suggest anti-pattern sweep or Constructor-Pattern violation
 # References (extra files beyond auto-included baseline/memory/project)
 [references]
 extra = [
-    "~/.claude/rules/debugging.md",
-    "~/.claude/rules/no-downgrade-constructive.md",
-    "~/.claude/agents/validator.md",
+    "path:user-rules/debugging.md",
+    "path:user-rules/no-downgrade-constructive.md",
 ]
 
 [taxonomy]

_assembler/tests/fixtures/_manifests/security-auditor.toml

@@ -0,0 +1,27 @@
+# Fixture stub for handoff-target resolution tests. Minimal valid manifest.
+name = "security-auditor"
+description = "Security auditor stub."
+tools = ["Read", "Glob", "Grep"]
+model = "opus"
+substrate_role = "read-only"
+role = "Stub security-auditor."
+blocks = ["baseline", "evidence-grading", "memory-protocol"]
+domain_in = ["security review"]
+forbidden_domain = ["fixing code"]
+
+[[handoff]]
+target = "code-implementer"
+trigger = "confirmed vulnerability needs fix"
+
+[taxonomy]
+kingdom = "manifest"
+mechanism = "compose"
+domain = "agent"
+layer = "agent-substrate"
+stage = "design-time"
+stability = "stable"
+language = "toml"
+
+[lineage]
+creator = "ag-orchestrator-human"
+created = "2026-05-02"

_assembler/tests/fixtures/_manifests/validator.toml

@@ -0,0 +1,27 @@
+# Fixture stub for handoff-target resolution tests. Minimal valid manifest.
+name = "validator"
+description = "Fact-checker stub."
+tools = ["Read"]
+model = "sonnet"
+substrate_role = "read-only"
+role = "Stub validator."
+blocks = ["baseline", "evidence-grading", "memory-protocol"]
+domain_in = ["claim verification"]
+forbidden_domain = ["fixing issues"]
+
+[[handoff]]
+target = "code-implementer"
+trigger = "FALSE API claim needs fix"
+
+[taxonomy]
+kingdom = "manifest"
+mechanism = "compose"
+domain = "agent"
+layer = "agent-substrate"
+stage = "design-time"
+stability = "stable"
+language = "toml"
+
+[lineage]
+creator = "ag-orchestrator-human"
+created = "2026-05-02"

_assembler/tests/snapshots/code-implementer.snap

@@ -1,5 +1,6 @@
 ---
 source: tests/golden.rs
+assertion_line: 47
 expression: out
 ---
 ---
@@ -358,7 +359,6 @@ Counter: each FAILED attempt on the SAME problem = +1. Success = reset.
 **Out (hand off):**
 - `ml-implementer` — task involves ML training / inference / Modal / experiment runners / Math-First paradigm
 - `infra-implementer` — task involves deploy / CI/CD / secrets / IaC / credentials / public-surface hosting
-- `physics-deriver` — task requires math derivation / theorem writing / theorem .md derivation
 - `critic` — anti-pattern sweep / code smell review on large diff (>500 LOC) or long function chains
 - `security-auditor` — code touches auth, crypto, network protocol, deserialization, FFI, or any HIGH-risk surface (see debugging.md Security Review)
 - `validator` — pre-commit citation or RULE 0.4 check on docs written alongside code
@@ -368,7 +368,6 @@ Counter: each FAILED attempt on the SAME problem = +1. Success = reset.
 
 - **ml-implementer** — task involves ML training / inference / Modal / experiment runners / Math-First paradigm
 - **infra-implementer** — task involves deploy / CI/CD / secrets / IaC / credentials / public-surface hosting
-- **physics-deriver** — task requires math derivation / theorem writing / theorem .md derivation
 - **critic** — anti-pattern sweep / code smell review on large diff (>500 LOC) or long function chains
 - **security-auditor** — code touches auth, crypto, network protocol, deserialization, FFI, or any HIGH-risk surface (see debugging.md Security Review)
 - **validator** — pre-commit citation or RULE 0.4 check on docs written alongside code
@@ -415,9 +414,9 @@ Blockers / next: <list>
 
 - `~/.claude/CLAUDE.md` — baseline umbrella
 - `~/.claude/memory/MEMORY.md` — memory index (adjust if your Claude Code user-slug path differs)
-- `~/.claude/rules/code-style.md`
-- `~/.claude/rules/git-conventions.md`
-- `~/.claude/rules/dev-workflow.md`
-- `~/.claude/rules/debugging.md`
-- `~/.claude/rules/karpathy-behavioral.md`
+- `path:user-rules/code-style.md`
+- `path:user-rules/git-conventions.md`
+- `path:user-rules/dev-workflow.md`
+- `path:user-rules/debugging.md`
+- `path:user-rules/karpathy-behavioral.md`
 - `MEMORY.md → Architecture Overlay Incident (model_brain.py 227→354 LOC from "fixes" — never patch, fix root formulas)`

_assembler/tests/snapshots/researcher.snap

@@ -1,5 +1,6 @@
 ---
 source: tests/golden.rs
+assertion_line: 33
 expression: out
 ---
 ---
@@ -188,7 +189,6 @@ Rules: architectural decision → E1-E2. Financial (compute) → ONLY E1. Data >
 **Out (hand off):**
 - `validator` — claim needs hard verification (citation sanity, reproduce-in-tests, RULE 0.4 gate before commit)
 - `ml-researcher` — question is ML/RL specialized-node (Math-First + tooling-reuse + synthetic-to-real discipline)
-- `patent-researcher` — question touches patent prior art, FTO, or novelty (IP-aware handling required)
 - `architect` — question is structural/architectural — dependency graph, pattern inventory, module boundaries
 - `critic` — findings suggest anti-pattern sweep or Constructor-Pattern violation review
 
@@ -196,7 +196,6 @@ Rules: architectural decision → E1-E2. Financial (compute) → ONLY E1. Data >
 
 - **validator** — claim needs hard verification (citation sanity, reproduce-in-tests, RULE 0.4 gate before commit)
 - **ml-researcher** — question is ML/RL specialized-node (Math-First + tooling-reuse + synthetic-to-real discipline)
-- **patent-researcher** — question touches patent prior art, FTO, or novelty (IP-aware handling required)
 - **architect** — question is structural/architectural — dependency graph, pattern inventory, module boundaries
 - **critic** — findings suggest anti-pattern sweep or Constructor-Pattern violation review
 
@@ -239,6 +238,5 @@ Blockers / next: <list>
 
 - `~/.claude/CLAUDE.md` — baseline umbrella
 - `~/.claude/memory/MEMORY.md` — memory index (adjust if your Claude Code user-slug path differs)
-- `~/.claude/rules/debugging.md`
-- `~/.claude/rules/no-downgrade-constructive.md`
-- `~/.claude/agents/validator.md`
+- `path:user-rules/debugging.md`
+- `path:user-rules/no-downgrade-constructive.md`

_capabilities/output/report-format/capability.toml

@@ -30,5 +30,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_capabilities/output/severity-grade/capability.toml

@@ -30,5 +30,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_capabilities/policy/no-git-ops/capability.toml

@@ -30,6 +30,7 @@ kingdom = "capability"
 mechanism = "gate"
 domain = "policy"
 layer = "agent-substrate"
+stage = "runtime"
 stability = "stable"
 language = "rust"
 

_capabilities/quality/cargo-check-green/capability.toml

@@ -25,6 +25,7 @@ kingdom = "capability"
 mechanism = "verify"
 domain = "quality"
 layer = "agent-substrate"
+stage = "runtime"
 stability = "stable"
 language = "rust"
 

_capabilities/quality/constructor-pattern/capability.toml

@@ -30,5 +30,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_capabilities/quality/tests-green/capability.toml

@@ -30,5 +30,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_capabilities/safety/no-dep-bump/capability.toml

@@ -35,5 +35,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_capabilities/scope/files-denylist/capability.toml

@@ -35,5 +35,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_capabilities/scope/files-whitelist/capability.toml

@@ -35,5 +35,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_capabilities/tools/deny-tools/capability.toml

@@ -30,5 +30,6 @@ stability = "stable"
 language = "rust"
 
 [lineage]
+parents = []
 creator = "ag-orchestrator-human"
 created = "2026-04-23"

_manifests/architect.toml

@@ -86,9 +86,7 @@ trigger = "system is ML/specialized-node-class and structural review must apply
 target = "validator"
 trigger = "architectural claim needs hard reproduction (build graph, import graph, coupling metric)"
 
-[[handoff]]
-target = "physics-deriver"
-trigger = "structural review asks how a new theorem family fits the existing T1-T68 proof graph"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
 # References (extra files beyond auto-included baseline/memory/project)
 [references]

_manifests/code-implementer-flutter.toml

@@ -21,9 +21,28 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "Flutter multi-platform apps — iOS, Android, Web, macOS, Windows (RULE 0.2 exception #2 — existing Flutter app)",
+    "Riverpod state management — NotifierProvider, AsyncNotifierProvider, @riverpod codegen",
+    "Clean Architecture + Feature-First directory layout",
+    "freezed value types + json_serializable for data models",
+    "Supabase client integration, REST/Realtime subscriptions",
+    "Constructor Pattern enforcement: file <200 LOC, function <30 LOC",
+]
+forbidden_domain = [
+    "Writing Rust / Swift / Python / Go / TypeScript — hand off to matching sibling",
+    "Hardcoded secrets (RULE 0.8) — every token by ENV var name, never inline",
+    "setState in complex state — use Riverpod providers",
+    "Business logic in Widget build() — extract to Notifier / service layer",
+    "Writing tests later — tests land with the change or before it (Test-First)",
+    "Skipping `flutter analyze` + `flutter test` before declaring done",
+]
+output_extra_fields = [
+    "Language: Flutter/Dart (RULE 0.2 exception #2 — existing Flutter app)",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Constructor Pattern compliance: largest file <N LOC / limit 200>, largest function <M LOC / limit 30>",
+    "Tests: <name> — <pass/fail> — `flutter test`",
+]
 
 [[handoff]]
 target = "validator"

_manifests/code-implementer-go.toml

@@ -21,9 +21,27 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "Mesh networking nodes — TCP/UDP listeners, peer discovery, connection pools",
+    "Lightweight CLI tools with cobra or stdlib flag (RULE 0.2 exception #2 — existing Go service)",
+    "Embedded HTTP servers for sidecar services and health endpoints",
+    "Go stdlib-first: prefer net/http, sync, encoding/json over heavy frameworks",
+    "Constructor Pattern enforcement: file <200 LOC, function <30 LOC",
+]
+forbidden_domain = [
+    "Writing Rust / Swift / Python / Flutter / TypeScript — hand off to matching sibling",
+    "Hardcoded secrets (RULE 0.8) — every token by ENV var name, never inline",
+    "Starting a new Go project without citing RULE 0.2 exception number",
+    "Global mutable state — pass deps explicitly via constructor injection",
+    "Goroutine leak — every goroutine must have a cancel path",
+    "Writing tests later — tests land with the change or before it (Test-First)",
+]
+output_extra_fields = [
+    "Language: Go (RULE 0.2 exception #2 — existing Go service / mesh)",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Constructor Pattern compliance: largest file <N LOC / limit 200>, largest function <M LOC / limit 30>",
+    "Tests: <name> — <pass/fail> — `go test ./...`",
+]
 
 [[handoff]]
 target = "validator"

_manifests/code-implementer-python.toml

@@ -21,9 +21,26 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "ML training scripts for models >10M params (RULE 0.2 exception #1 — PyTorch/JAX; inference stays Rust)",
+    "Extending existing language-locked Python projects (RULE 0.2 exception #2)",
+    "One-shot scripts <50 lines that will be deleted within 24h (RULE 0.2 exception #5)",
+    "External-binding wrappers where Python-only SDK exists (RULE 0.2 exception #6)",
+    "Constructor Pattern enforcement: file <200 LOC, function <30 LOC",
+]
+forbidden_domain = [
+    "Writing Python without stating a RULE 0.2 exception number (1-7) — refuse and delegate to code-implementer-rust",
+    "Starting new long-lived Python services — Rust is default for inference, servers, and tools",
+    "Hardcoded secrets (RULE 0.8) — every token by ENV var name, never inline",
+    "Global mutable state / singleton patterns outside of ML framework idioms",
+    "Writing tests later — tests land with the change or before it (Test-First)",
+]
+output_extra_fields = [
+    "Language: Python (RULE 0.2 exception #<N> — <stated reason>)",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Constructor Pattern compliance: largest file <N LOC / limit 200>, largest function <M LOC / limit 30>",
+    "Tests: <name> — <pass/fail> — `pytest`",
+]
 
 [[handoff]]
 target = "validator"

_manifests/code-implementer-swift.toml

@@ -21,9 +21,27 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "macOS menubar apps — NSStatusItem + popover, switch `.regular` → `.accessory` pattern",
+    "SwiftUI views + AppKit interop for macOS (RULE 0.2 exception #3 — platform-native UI)",
+    "iOS apps — SwiftUI lifecycle, NavigationStack, @Observable, SwiftData",
+    "SPM executables with `-Xlinker -sectcreate` info_plist embed + `codesign --force --sign -`",
+    "Avoid MenuBarExtra with SPM executables — use NSStatusItem directly (platform known issue)",
+    "Constructor Pattern enforcement: file <200 LOC, function <30 LOC",
+]
+forbidden_domain = [
+    "Writing Rust / Python / Go / Flutter / TypeScript — hand off to matching sibling",
+    "Hardcoded secrets (RULE 0.8) — every token by ENV var name, never inline",
+    "Force-unwrapping optionals without fallback or guard",
+    "MenuBarExtra with SPM executables (broken — use NSStatusItem)",
+    "Writing tests later — tests land with the change or before it (Test-First)",
+]
+output_extra_fields = [
+    "Language: Swift (RULE 0.2 exception #3 — platform-native macOS/iOS UI)",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Constructor Pattern compliance: largest file <N LOC / limit 200>, largest function <M LOC / limit 30>",
+    "Tests: <name> — <pass/fail> — `swift test`",
+]
 
 [[handoff]]
 target = "validator"

_manifests/code-implementer-typescript.toml

@@ -21,9 +21,28 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "Next.js 16 app-router pages, API routes, server actions",
+    "Node services with Express / Hono — type-safe with Zod schemas",
+    "Browser-DOM TypeScript — event handlers, fetch wrappers, Web APIs",
+    "Drizzle ORM schemas + queries (PostgreSQL / SQLite / MySQL)",
+    "Zod validation schemas for request/response contracts",
+    "Constructor Pattern enforcement: file <200 LOC, function <30 LOC",
+]
+forbidden_domain = [
+    "Writing Rust / Swift / Python / Go / Flutter — hand off to matching sibling",
+    "Hardcoded secrets (RULE 0.8) — every token by ENV var name, never inline",
+    "any=true / ts-ignore / @ts-expect-error without filed issue comment",
+    "Raw SQL string-building — use Drizzle query builder or parameterised statements",
+    "Writing tests later — tests land with the change or before it (Test-First)",
+]
+output_extra_fields = [
+    "Language: TypeScript (RULE 0.2 exception #4 — browser/DOM)",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Constructor Pattern compliance: largest file <N LOC / limit 200>, largest function <M LOC / limit 30>",
+    "Tests: <name> — <pass/fail> — <command to reproduce>",
+    "cargo-check: N/A (TypeScript) | tsc --noEmit: <pass/fail>",
+]
 
 [[handoff]]
 target = "validator"

_manifests/code-implementer.toml

@@ -74,9 +74,7 @@ trigger = "task involves ML training / inference / Modal / experiment runners /
 target = "infra-implementer"
 trigger = "task involves deploy / CI/CD / secrets / IaC / credentials / public-surface hosting"
 
-[[handoff]]
-target = "physics-deriver"
-trigger = "task requires math derivation / theorem writing / theorem .md derivation"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
 [[handoff]]
 target = "critic"

_manifests/critic.toml

@@ -31,7 +31,7 @@ domain_in = [
     "Performance — N+1 queries, missing indexes, memory leaks, blocking I/O, hot-path allocations",
     "Tech debt — duplicated logic, inconsistent naming, missing tests, outdated deps",
     "Constructor-Pattern violations — files >200 LOC, functions >30 LOC, mixed responsibilities",
-    "Cross-ref rules DB: `~/.claude/skills/architecture-rules/references/antipatterns.md`, `duplication.md`, `stack-compat.md`",
+    "Cross-ref rules DB: `path:user-skills/architecture-rules/references/antipatterns.md`, `duplication.md`, `stack-compat.md`",
 ]
 
 forbidden_domain = [
@@ -71,13 +71,14 @@ target = "architect"
 trigger = "anti-pattern is structural (new family, needs design review)"
 
 # References (extra files beyond auto-included baseline/memory/project)
+# ~/.claude/ absolute paths stripped — not portable across installs. See audit 2026-05-02.
 [references]
 extra = [
     "path:user-rules/debugging.md",
     "path:user-rules/code-style.md",
-    "~/.claude/skills/architecture-rules/references/antipatterns.md",
-    "~/.claude/skills/architecture-rules/references/duplication.md",
-    "~/.claude/skills/architecture-rules/references/stack-compat.md",
+    "path:user-skills/architecture-rules/references/antipatterns.md",
+    "path:user-skills/architecture-rules/references/duplication.md",
+    "path:user-skills/architecture-rules/references/stack-compat.md",
 ]
 
 [taxonomy]

_manifests/frontend-validator.toml

@@ -28,7 +28,7 @@ Your steps in order, each emitting a section of the final report:
 
 6. **A11y quick** — if `package.json` has `a11y-check` script, invoke. Else skip. Severity: WARN.
 
-6. **Verdict block** — summary table: each check, status (PASS / WARN / FAIL), brief evidence pointer.
+7. **Verdict block** — summary table: each check, status (PASS / WARN / FAIL), brief evidence pointer.
 
 You do NOT autofix. You do NOT spawn other agents. You do NOT commit. You report.
 """

_manifests/infra-implementer-cicd.toml

@@ -21,9 +21,27 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "GitHub Actions workflows — build matrices, caching, artifact upload/download",
+    "GitLab CI pipelines — stages, rules, variables, Docker-in-Docker runners",
+    "Forgejo / Gitea workflows (subset of GitHub Actions syntax)",
+    "Build-and-deploy scripts — cargo build --release, docker build+push, npm run build",
+    "Secret-injection from vault / GitHub Secrets / GitLab CI variables — ENV only, never inline",
+    "Staged deploys — canary, blue/green, rollback triggers",
+]
+forbidden_domain = [
+    "IaC (Terraform/Pulumi/CDK) — hand off to infra-implementer-iac",
+    "Dockerfiles or OCI images — hand off to infra-implementer-container",
+    "Secrets management (Vault, sops, age) — hand off to infra-implementer-secrets",
+    "Hardcoded secrets in workflow YAML (RULE 0.8) — use `${{ secrets.NAME }}` / ENV refs",
+    "Skipping build-cache steps — always cache cargo registry + target, node_modules, pip cache",
+]
+output_extra_fields = [
+    "CI platform: GitHub Actions | GitLab CI | Forgejo",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Jobs defined: <list with trigger events>",
+    "Secrets used: <list of secret names — verify RULE 0.8 compliance>",
+]
 
 [[handoff]]
 target = "validator"

_manifests/infra-implementer-container.toml

@@ -21,9 +21,28 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "Dockerfiles — multi-stage builds with cargo-chef (Rust), distroless final image",
+    "OCI image labelling — `org.opencontainers.image.*` standard labels",
+    "docker-compose for local dev + integration tests",
+    "Helm chart templates for Kubernetes deployments",
+    "Non-root user (`USER nonroot:nonroot`), layer-pinning with digest locks",
+    "Build-cache optimisation: COPY Cargo.toml first, then src — avoid rebuilding deps on code changes",
+]
+forbidden_domain = [
+    "App code changes — hand off to matching code-implementer sibling",
+    "Kubernetes manifests outside Helm chart templates — hand off to infra-implementer-iac",
+    "Hardcoded secrets in Dockerfile / compose (RULE 0.8) — use build-args + secret mounts only",
+    "Running as root in final image",
+    "FROM latest — always pin to a specific digest or version tag",
+]
+output_extra_fields = [
+    "Base image: <registry/image:tag@digest>",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Stages: <list of stage names + purpose>",
+    "Final image size estimate: <MB>",
+    "Non-root user: <yes | no>",
+]
 
 [[handoff]]
 target = "validator"

_manifests/infra-implementer-iac.toml

@@ -21,9 +21,27 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "Terraform HCL modules — resource-per-file, remote backend (S3/GCS/Terraform Cloud), lock file",
+    "Pulumi (TypeScript or Go) — stack-per-env, explicit outputs, state in Pulumi Cloud or S3",
+    "OpenTofu (Terraform OSS fork) — drop-in compatible HCL",
+    "AWS CDK (TypeScript) — construct libraries, stack separation by lifecycle",
+    "Module-per-resource-type layout: one directory per resource family",
+    "State-management discipline: remote backend required, local state forbidden in production",
+]
+forbidden_domain = [
+    "CI/CD yaml pipelines — hand off to infra-implementer-cicd",
+    "Dockerfiles or OCI images — hand off to infra-implementer-container",
+    "Hardcoded secrets in .tf / .ts / pulumi config (RULE 0.8) — use var + SecretManager refs",
+    "Monolithic single-file IaC (>200 LOC) — decompose into modules",
+    "Committing `terraform.tfstate` or `.tfstate.backup` to git — use remote backend only",
+]
+output_extra_fields = [
+    "IaC tool: Terraform | Pulumi | OpenTofu | CDK",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Resources defined: <list with provider + resource type>",
+    "State backend: <remote backend details>",
+]
 
 [[handoff]]
 target = "validator"

_manifests/infra-implementer-secrets.toml

@@ -21,9 +21,27 @@ blocks = [
     "rule-double-audit",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "sops encryption — `.sops.yaml` key rules, `sops --encrypt` / `--decrypt` workflow",
+    "age key generation + encryption for file-at-rest secrets",
+    "HashiCorp Vault — dynamic secrets, AppRole auth, KV v2, lease renewal",
+    "Cloudflare Secret / AWS Secrets Manager / GCP Secret Manager integration patterns",
+    "ENV-var injection patterns: `.env` files gitignored, referenced via `$VAR` only (RULE 0.8)",
+    "Rotation playbooks — how to rotate a token without downtime",
+]
+forbidden_domain = [
+    "App code changes — hand off to matching code-implementer sibling",
+    "CI/CD pipeline YAML — hand off to infra-implementer-cicd",
+    "Hardcoded secret values anywhere (RULE 0.8 hard ban)",
+    "Committing `.env` files with real values to git",
+    "Storing secrets in IaC state files — use Secrets Manager + data source",
+]
+output_extra_fields = [
+    "Secrets tool: sops | age | Vault | CF Secret | AWS SM | GCP SM",
+    "Plan-Mode used: <yes | no + trivial-edit exemption reason>",
+    "Secrets catalogued: <count + names (not values)>",
+    "RULE 0.8 compliance: <all tokens by ENV var name — yes | violations listed>",
+]
 
 [[handoff]]
 target = "validator"

_manifests/ml-implementer.toml

@@ -78,9 +78,7 @@ output_extra_fields = [
 ]
 
 # Handoffs MUST come after all top-level keys (TOML array-of-tables scope rule)
-[[handoff]]
-target = "physics-deriver"
-trigger = "numerical result implies a new theorem / refutation / observable classification (write to `theory/**/*.md`)"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
 [[handoff]]
 target = "ml-researcher"
@@ -104,7 +102,7 @@ trigger = "anti-pattern sweep on training script (coefficient creep, E1-E11 chec
 
 [[handoff]]
 target = "architect"
-trigger = "multi-node multi-node composition design, experiment matrix layout, benchmark/baseline integration"
+trigger = "multi-node composition design, experiment matrix layout, benchmark/baseline integration"
 
 [references]
 extra = [
@@ -114,7 +112,7 @@ extra = [
     "path:user-rules/observable-classification.md",
     "path:user-rules/manifold-tangent-sanity.md",
     "path:user-rules/no-downgrade-constructive.md",
-    "path:user-memory/wrong-paths-specialized-ml.md",
+    "path:user-memory/wrong-paths-specialized-ml.md",  # TODO verify path:user-memory exists in assembler resolver
     "MEMORY.md → Compute Cost Incident (2026-02-26): promised $27, spent $98.78 on Modal. NEVER AGAIN.",
     "MEMORY.md → Architecture Overlay Incident: model_brain.py 227→354 LOC from audit fixes. No Patching.",
 ]

_manifests/ml-researcher.toml

@@ -73,9 +73,7 @@ output_extra_fields = [
 target = "ml-implementer"
 trigger = "hypothesis is formulated and experiment must be run (train, benchmark, ablate, Monte Carlo)"
 
-[[handoff]]
-target = "physics-deriver"
-trigger = "literature finding feeds a theorem / derivation in `~/your-project/theory/`"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
 [[handoff]]
 target = "validator"
@@ -85,10 +83,6 @@ trigger = "citation sanity before commit (RULE 0.4 gate) or reproducibility clai
 target = "researcher"
 trigger = "non-ML sub-question surfaces (general library / API / pricing / doc lookup)"
 
-[[handoff]]
-target = "patent-researcher"
-trigger = "ML finding is patent-relevant (prior art, FTO, novelty for a filable claim)"
-
 [[handoff]]
 target = "architect"
 trigger = "question is about ML-system architecture (node graph, data-flow, module boundaries) not algorithm"
@@ -101,7 +95,7 @@ extra = [
     "path:user-rules/observable-classification.md",
     "path:user-rules/api-cost-guard.md",
     "path:user-rules/no-downgrade-constructive.md",
-    "path:user-memory/wrong-paths-specialized-ml.md",
+    "path:user-memory/wrong-paths-specialized-ml.md",  # TODO verify path:user-memory exists in assembler resolver
 ]
 
 [taxonomy]

_manifests/researcher-code.toml

@@ -17,9 +17,24 @@ blocks = [
     "memory-protocol",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "Symbol discovery — Grep for function/type/constant definitions across the repo",
+    "Dependency tracing — who imports X, what does X import (caller/callee graph via Grep)",
+    "Ownership mapping — identify the canonical SSoT file for a type or route",
+    "Analogue search — find existing implementations before proposing new code",
+    "Constructor-Pattern audit — identify files >200 LOC or functions >30 LOC",
+]
+forbidden_domain = [
+    "Web research — hand off to researcher-web",
+    "Editing or writing code (read-only agent — no Edit / Write)",
+    "Whole-file dumps when Grep + targeted Read suffices",
+    "Fabricating file paths, function names, or line numbers",
+]
+output_extra_fields = [
+    "Sources cited (≥2 for load-bearing)",
+    "Evidence grade per claim",
+    "Gaps section present",
+]
 
 [[handoff]]
 target = "validator"

_manifests/researcher-web.toml

@@ -17,9 +17,25 @@ blocks = [
     "memory-protocol",
 ]
 
-domain_in = ["task scope (verbatim user prompt)", "target paths / files"]
-forbidden_domain = ["hardcoded secrets (RULE 0.8)", "cross-language drift (use the matching sibling)"]
-output_extra_fields = ["Largest file LOC", "Tests pass count"]
+domain_in = [
+    "Official documentation — vendor docs, RFC, spec, changelog, pricing page",
+    "GitHub repositories — README, CHANGELOG, open issues, PR descriptions",
+    "Academic and technical papers — arXiv, ACM DL, IEEE (cite with [VERIFIED: url] per RULE 0.4)",
+    "Pricing and SLA pages — E1 only (primary source from vendor, not third-party comparison)",
+    "Blog posts and tutorials — max E4; flag recency and single-source limit",
+]
+forbidden_domain = [
+    "Local filesystem reads — hand off to researcher-code",
+    "Hardcoded or paraphrased secret values anywhere",
+    "Fabricating URLs, DOIs, authors, or version numbers (RULE 0.4 hard ban)",
+    "Quoting Stack Overflow / Reddit / random blogs above E4",
+    "Pricing claims from anything other than the vendor's own pricing page",
+]
+output_extra_fields = [
+    "Sources cited (≥2 for load-bearing)",
+    "Evidence grade per claim",
+    "Gaps section present",
+]
 
 [[handoff]]
 target = "validator"

_manifests/researcher.toml

@@ -67,9 +67,7 @@ trigger = "claim needs hard verification (citation sanity, reproduce-in-tests, R
 target = "ml-researcher"
 trigger = "question is ML/RL specialized-node (Math-First + tooling-reuse + synthetic-to-real discipline)"
 
-[[handoff]]
-target = "patent-researcher"
-trigger = "question touches patent prior art, FTO, or novelty (IP-aware handling required)"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
 [[handoff]]
 target = "architect"
@@ -80,11 +78,11 @@ target = "critic"
 trigger = "findings suggest anti-pattern sweep or Constructor-Pattern violation review"
 
 # References (extra files beyond auto-included baseline/memory/project)
+# ~/.claude/agents/validator.md stripped — machine-local file, not portable. See audit 2026-05-02.
 [references]
 extra = [
     "path:user-rules/debugging.md",
     "path:user-rules/no-downgrade-constructive.md",
-    "~/.claude/agents/validator.md",
 ]
 
 [taxonomy]

_manifests/validator.toml

@@ -56,19 +56,13 @@ output_extra_fields = [
     "Overall verdict: ALL VERIFIED | PARTIAL (fix list) | BLOCK (FALSE findings present)",
 ]
 
-# Handoffs MUST come after all top-level keys (TOML array-of-tables scope rule)
-[[handoff]]
-target = "physics-deriver"
-trigger = "theory doc has FALSE or UNVERIFIED citation — rewrite before commit"
+# physics-deriver / patent-compliance / patent-researcher manifests not yet authored — handoffs removed 2026-05-02 per audit
 
+# Handoffs MUST come after all top-level keys (TOML array-of-tables scope rule)
 [[handoff]]
 target = "ml-researcher"
 trigger = "claim needs literature/arXiv deep-search to resolve (returns `[VERIFIED: url]`)"
 
-[[handoff]]
-target = "patent-compliance"
-trigger = "FALSE claim is in patent draft — pre-filing block"
-
 [[handoff]]
 target = "code-implementer"
 trigger = "FALSE API/version claim is in code — needs fix before ship"

_primitives/_rust/kei-registry/tests/fixtures/atom-sample/.kei-registry-ignore

@@ -0,0 +1 @@
+# DNA-INDEX scanner: skip this directory (test fixture)

_primitives/_rust/kei-registry/tests/fixtures/fake-kit/.kei-registry-ignore

@@ -0,0 +1 @@
+# DNA-INDEX scanner: skip this directory (test fixture)

_primitives/_rust/kei-registry/tests/fixtures/mini-kit/.kei-registry-ignore

@@ -0,0 +1 @@
+# DNA-INDEX scanner: skip this directory (test fixture)

_roles/auditor.toml

@@ -26,3 +26,17 @@ language = "toml"
 
 [pipeline]
 handoff = ["merger"]
+
+[taxonomy]
+kingdom = "role"
+mechanism = "audit"
+domain = "agent"
+layer = "agent-substrate"
+stage = "runtime"
+stability = "stable"
+language = "toml"
+
+[lineage]
+parents = []
+creator = "ag-orchestrator-human"
+created = "2026-04-23"

_roles/explorer.toml

@@ -3,6 +3,7 @@ name = "explorer"
 display-name = "explorer + cargo-check (read-only analyst with build probe)"
 description = "Read-only analyst that may run cargo-family commands for build/test introspection. No edits, no git, no non-cargo shell."
 spawnable = true
+# claude-subagent-type values are case-sensitive; "Explore" is the Claude Code built-in
 claude-subagent-type = "Explore"
 
 [capabilities]

_roles/merger.toml

@@ -26,3 +26,17 @@ language = "toml"
 
 [pipeline]
 handoff = []
+
+[taxonomy]
+kingdom = "role"
+mechanism = "merge"
+domain = "agent"
+layer = "agent-substrate"
+stage = "runtime"
+stability = "stable"
+language = "toml"
+
+[lineage]
+parents = []
+creator = "ag-orchestrator-human"
+created = "2026-04-23"