definition-of-done

name	definition-of-done
description	Applicability-aware Definition of Done checklist. Core checks apply to every task; overlay checks activate only when the applicability manifest says the surface exists. Binary verification — pass or fail. Triggers at Phase 4 completion and Phase 5 entry.

Definition of Done

Overview

Every task carries a DoD core baseline. Overlay checks are selected from the applicability manifest and only evaluated when the task actually touches that surface. A task is not complete until every active check is verified. Each check is binary (pass/fail) with a specified verification method and a tag:

• automatable: must be verified by tool output
• judgement: may call an LLM or human reviewer

When to Use

• Phase 4 completion: Verify all DoD items after execution
• Phase 5 entry: DoD verification is the first step of the quality gate for active overlays
• Manual: Any time task completion needs verification

Applicability Model

• Core baseline checks apply to every task.
• Overlay checks activate only when applicability_manifest marks the relevant surface as active.
• Any suppressed overlay must include a concrete not_applicable reason tied to task class or repo evidence.

Overlay Activation Trigger Table

This table is the authoritative mapping of change_surface flags and task class to DoD overlay check IDs. It is consumed by the deterministic dod_overlays evaluator. Any change here must be mirrored in tests/backtest/evaluators/dod_overlays.sh.

Overlay	Check IDs	Trigger Expression
Security	5, 6, 7, 8, 9	new_attack_surface OR auth_change OR external_integration
Observability	10, 11, 12, 13, 14	runtime_path_change OR user_facing_operation
Integration	18, 19, 20	service_boundary_change OR external_integration OR api_change OR data_format_change
Content	21, 22	task_classification == "docs" or human-facing output is in scope
Implementation Interface / Productionization	23, 24, 25, 26	service_boundary_change OR external_integration OR api_change OR data_format_change OR runtime_path_change OR user_facing_operation when an interface or production claim is active

Overlay checks that are not activated by this table must be reported as not_applicable with a concrete reason in the manifest's section_policy.

Mixed Acceptance Criteria Shapes

Acceptance criteria may arrive as strings or structured Given/When/Then objects.

Extraction rules:

• If an item is an object, read its .then field as the criterion text and preserve any id and measurable_post_condition in the verification record.
• If an item is a string, use the string directly.
• Do not mark a task done if the evaluation path drops an upstream id or measurable_post_condition.

Core Baseline Checks

Code Quality

#	Check	Tag	Scope	Verification	Blocks On
1	All linters pass	automatable	core	Automated: LDD gate	Any violation
2	All tests pass	automatable	core	Automated: TDD suite	Any failure
3	Code complexity within budget	automatable	core	Automated: configured cyclomatic-complexity threshold only	Exceeding configured complexity ceiling
4	No code slop or stub patterns	automatable	core	Automated: anti-slop scanner	TODO, FIXME, PLACEHOLDER, todo!(), unimplemented!(), panic!(\"not implemented\"), NotImplementedError, pass, empty placeholder bodies, commented-out code, fake/mock placeholder logic in shipped code

Reuse & Patterns

#	Check	Tag	Scope	Verification	Blocks On
15	Reuse analysis confirmed	automatable	core	Automated: reuse-analysis output	Reimplementation detected
16	Existing conventions followed	automatable	core	Automated: bin/adlc convention-scan --file <changed-file> --json when target repo conventions are present	Open convention-scan issue
17	Antipattern checklist cleared	automatable	core	Automated: anti-slop plus antipattern scanner outputs	Known antipattern used
28	Repo convention waivers explicit	automatable	core	Convention scan or task output records file, rule, and reason for every waiver	Silent convention waiver, missing reason, or waived rule not tied to a file
29	Honesty contract satisfied	automatable	core	Build Brief schema plus emit-work-items --require-ready; artifact outputs expose no_overclaim and limitations when required	Missing honesty_contract, missing no-external-claims skip reason, missing docs honesty section requirement, or missing artifact no_overclaim / limitations fields
30	Performance envelope satisfied	automatable	core	Build Brief schema plus emit-work-items --require-ready; benchmark-required tasks record benchmark evidence before closeout	Missing performance_envelope, missing no-data-path skip reason, missing input scale or hot-path complexity, missing benchmark spec, or missing benchmark evidence
31	Task sizing satisfied	automatable	core	Build Brief schema plus emit-work-items --require-ready; task_sizing declares one module, one coherent structured file-set, or an atomic cross-module reason	Missing task_sizing, split-required task, broad multi-module surface without coherent file-set or atomic reason, code-module creation without module_plan, or size-only split/pass rationale
32	Ponytail minimality admitted	automatable	core	Automated: bin/adlc ponytail-admit --build-brief <brief> --diff-file <final.diff> --json when a final diff exists	Missing two-field minimality_contract, unapproved dependency diff, or removed validation/error/security/accessibility anatomy without waiver
33	Compatibility evidence satisfied	automatable	core	Automated: bin/adlc contract-surface-inventory plus bin/adlc compatibility-evidence --build-brief <brief> --inventory <inventory> --json when a published surface is touched	Missing named contract surface, missing verification predicate, generic compatibility claim, missing compatibility_evidence_refs, or absent migration/deprecation evidence for a breaking surface change

No DoD check may use file size, line count, or SLOC as a criterion. Split decisions are based on responsibility, side effects, coordinator boundaries, and repo conventions, never size.

Security Overlay

#	Check	Tag	Scope	Verification	Blocks On
5	STRIDE threat model complete	judgement	overlay	Council: Security Auditor	Missing categories
6	STRIDE mitigations implemented	judgement	overlay	Council review with cited implementation evidence	Unimplemented mitigations
7	OWASP Top 10 scan clean	automatable	overlay	Automated: SAST scan	High/Critical findings
8	No hardcoded secrets	automatable	overlay	Automated: pattern scan	Any match
9	Input validation at trust boundaries	judgement	overlay	Council review	Missing validation

Observability Overlay

#	Check	Tag	Scope	Verification	Blocks On
10	Error logging at failure points	automatable	overlay	Automated: AST scan	Missing error logs
11	Audit logging for state changes	automatable	overlay	Automated: AST scan	Missing audit logs
12	Operational logging for health	automatable	overlay	Automated: AST scan	Missing general logs
13	Log format matches convention	automatable	overlay	Automated: format check	Format mismatch
14	Correlation IDs propagate	automatable	overlay	Automated: pattern check	Missing correlation

Integration Overlay

#	Check	Tag	Scope	Verification	Blocks On
18	Integration wiring complete	automatable	overlay	Automated: registration check	Missing wiring, dead routes, unwired providers, unused flags/config, placeholder entry points
19	Scalability concerns documented	judgement	overlay	Council review	Undocumented
20	Degradation strategy defined	judgement	overlay	Council review	Missing strategy

Content Overlay (human-facing output only)

#	Check	Tag	Scope	Verification	Blocks On
21	Stop-slop gate passed	judgement	overlay	Slop judge plus scoring output	Score < 35/50
22	Voice/brand compliance	judgement	overlay	Voice / brand review	Voice mismatch

Implementation Interface / Productionization Overlay

#	Check	Tag	Scope	Verification	Blocks On
23	Implementation Interface preserved	judgement	overlay	Council review plus schema/CLI evidence	Missing or changed implementation_interface_contract, reuse, consumes/emits, minimum fields, invariants, integration points, or validation gates
24	Productionization Gate bounded	automatable	overlay	Build Brief schema plus emit-work-items --require-ready	Missing productionization_gate for active production claim
25	Production-ready evidence present	automatable	overlay	Readiness report	overclaimed_production_ready, missing validation evidence, no-overclaim, reliability failure modes, owner, rollback, runbook/observability, or security/privacy posture
26	No-Overclaim respected	judgement	overlay	Code review and PR closeout evidence	Shipped code, ticket text, or PR summary claims support beyond the Coverage State

Domain Adaptation

SWElfare: Core baseline always applies. Security, observability, integration, and content overlays activate only when the applicability manifest says the surface exists.

Ratatosk: Core baseline applies to every task. Security overlay → trade execution STRIDE. Observability overlay → trade/risk logging. Integration overlay → exchange API wiring. Content overlay → briefing slop gate.

Magnus: Core baseline applies to every task. Security overlay → brand STRIDE. Observability overlay → content performance logging. Integration overlay → platform integration. Content overlay → always active for user-facing copy.

Verification Schema

{
  "task_id": "TASK-001",
  "task_classification": "build_validation",
  "applicability_manifest": {
    "security_overlay": { "status": "not_applicable", "reason": "No new attack surface or trust boundary changes." },
    "observability_overlay": { "status": "not_applicable", "reason": "No new runtime path or user-facing operation." },
    "integration_overlay": { "status": "active", "reason": "Build wiring changed and must be verified." },
    "content_overlay": { "status": "not_applicable", "reason": "No human-facing output changed." }
  },
  "active_checks": ["1", "2", "3", "4", "15", "16", "17", "18"],
  "dod_checks": {
    "ldd_pass": { "status": true, "evidence": "ruff: 0 violations, mypy: 0 errors" },
    "tdd_pass": { "status": true, "evidence": "42 passed, 0 failed", "coverage": 87 },
    "complexity_ok": { "status": true, "evidence": "max configured cyclomatic complexity: 8" },
    "no_code_slop": { "status": true, "evidence": "0 violations: no TODO/FIXME/PLACEHOLDER, no stub bodies, no NotImplementedError, no commented-out code" },
    "stride_complete": { "status": true, "threats": 6, "mitigations": 4 },
    "stride_implemented": { "status": true, "evidence": "4/4 mitigations present" },
    "owasp_clean": { "status": true, "findings": 0 },
    "no_secrets": { "status": true, "evidence": "0 matches" },
    "input_validation": { "status": true, "evidence": "council approved" },
    "error_logging": { "status": true, "error_points": 3 },
    "audit_logging": { "status": true, "audit_points": 2 },
    "general_logging": { "status": true, "general_points": 5 },
    "log_format": { "status": true, "evidence": "structured JSON" },
    "correlation_ids": { "status": true, "evidence": "propagated" },
    "reuse_confirmed": { "status": true, "reimplementations": 0 },
    "conventions_followed": { "status": true, "evidence": "convention-scan: pass for changed Rust files" },
    "repo_convention_waivers_explicit": { "status": true, "evidence": "0 waivers, or each waiver records file/rule/reason" },
    "honesty_contract_satisfied": { "status": true, "evidence": "honesty_contract present; artifact outputs include no_overclaim and limitations when required" },
    "performance_envelope_satisfied": { "status": true, "evidence": "performance_envelope present; data-path tasks define scale and complexity, benchmark_required decisions are explicit, and required benchmark evidence is recorded" },
    "task_sizing_satisfied": { "status": true, "evidence": "task_sizing present; work is one module, a coherent structured file-set, or atomic cross-module work with an explicit reason" },
    "compatibility_evidence_satisfied": { "status": true, "evidence": "contract-surface-inventory plus compatibility-evidence report: pass with a verification predicate and compatibility_evidence_refs" },
    "antipatterns_cleared": { "status": true, "violations": 0 },
    "wiring_complete": { "status": true, "upstream": 2, "downstream": 1 },
    "scalability_documented": { "status": true },
    "degradation_defined": { "status": true },
    "slop_gate": { "status": true, "score": 42 },
    "voice_compliance": { "status": true },
    "implementation_interface_preserved": { "status": true },
    "productionization_gate_bounded": { "status": true },
    "production_ready_evidence_present": { "status": true },
    "no_overclaim_respected": { "status": true }
  },
  "all_passed": true,
  "timestamp": "2026-04-05T12:00:00Z"
}

Enforcement

• DoD verification runs automatically at Phase 4 completion for the core baseline and any active overlays
• Any failed active check blocks pipeline progression to Phase 6
• Failed checks produce specific remediation guidance
• No partial passes — ALL active checks must be green
• Scaffolding artifacts are never considered done. Any remaining stub, placeholder, or partial wiring fails DoD.
• Mixed acceptance-criteria handlers that lose an upstream id or measurable_post_condition fail DoD verification.
• Repo-convention failures block DoD unless every waiver is explicit and present in task output.
• Honesty contracts block DoD unless each executable task either carries does-not-do / limitations / unsafe-claims requirements or explicitly declares no external claims.
• Performance envelopes block DoD unless each executable task either defines expected input scale, hot-path complexity, benchmark requirement, and benchmark evidence when required, or uses structured applicability=not_applicable with a meaningful reason.
• Task sizing blocks DoD unless each executable task declares one module, one coherent structured file-set, or atomic cross-module work with an atomicity reason; code-module creation and explicit module-plan sizing bases require a real module plan, while coherent non-code file sets do not; split-required tasks stay blocked until decomposed.
• File size, line count, and SLOC are not valid DoD criteria.