Architecture Decision Governance and ADR Architect

    1

    Architects decision governance: MADR 3.0 standards, immutable decision lifecycles, and automated CI linters.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Architecture Decision Governance and ADR Architect

    Example session with this skill installed

    Architect enterprise Architecture Decision Record (ADR) Governance Platform blueprint and MADR 3.0 for Engineering.

    • Subject: enterprise-architecture-decision-mesh v3.0 (run-id: corp-adrarch-001) governing 120 migrations/year across 65 services and 450 engineers.
    • Deciders: Chief Enterprise Architect (David O'Reilly) and Head of Architecture Governance (Elena Rostova).
    • Scope: Standardized MADR 3.0 template conformance; 4-state lifecycle (Proposed, Accepted, Deprecated, Superseded); mandatory rejected options; automated CI PR linters.
    • Tension: Lack of decision records in incident ADR-4919 allowed a new squad to re-implement an un-indexed MongoDB query layer that had been rejected 18 months prior, recreating a 4.2-hour payment outage ($3.8M penalty). David O'Reilly and Elena Rostova mandate an authoritative Decision Governance Platform: immutable Git records, explicit negative consequences, and CI build breakers.

    Write the architecture decision architect overview under docs/.

    • Read your context and instructions
    • Compiled the architecture decision governance

    Wrote docs/architecture/architecture-decision-architect/00-overview/architecture-decision-architect-overview.md and docs/architecture/architecture-decision-architect/verification/fitness-self-check.md. Complete architecture decision record (ADR) governance blueprint establishing MADR standards, immutable decision lifecycle, and automated conformance gates.


    skill: architecture-decision-architect

    Architecture Decision Record (ADR) Governance Platform [ADRARCH-CORP-001]

    Summary

    This specification establishes the enterprise Architecture Decision Record (ADR) Governance Platform blueprint, decision lifecycles, Markdown Architectural Decision Record (MADR 3.0) standards, and automated conformance gates for enterprise-architecture-decision-mesh v3.0 under run ID corp-adrarch-001. It governs architectural decision making across 65 product microservices, 450 software engineers, and 32 engineering squads executing 120 major architectural migrations/year. It decisively investigates and resolves the tribal knowledge loss and architectural regression loops demonstrated in incident ADR-4919 (where lack of an authoritative decision governance repository allowed a newly hired platform squad to unknowingly re-implement an un-indexed distributed MongoDB query layer that had been evaluated and explicitly rejected 18 months prior due to severe locking bottlenecks, recreating the identical production outage, stalling payments for 4.2 hours, and incurring $3.8M in merchant SLA penalty payouts). The architecture enforces

    standardized MADR 3.0 template conformance, establishes immutable Git-based decision lifecycle states (Proposed, Accepted, Deprecated, Superseded), implements

    automated PR decision linter build breakers, and mandates

    sub-30-day decision review cycles.

    Detailed Description

    Operating a large-scale engineering organization without formal Architecture Decision Records (ADRs) results in severe collective amnesia. When engineering teams make critical architectural decisions in ephemeral Slack threads, private emails, or verbal meetings, the context, evaluated trade-offs, and reasons for rejecting alternatives are lost within months. When new engineers join or original team members depart, the organization repeats previous mistakes, debates solved problems, and violates hard-won architectural principles. Architecture Decision Governance establishes an

    Immutable Institutional Memory: it treats architectural decisions as version-controlled code artifacts (MADR in Git); enforces structured rationale, criteria weights, and negative consequences before decisions can be marked Accepted; links decisions directly to code repositories; and integrates automated linter gates into pull requests to prevent regressions.

    Engineering Architectural Proposal Stream (120 Decisions / Year)
                                       │
                                       ▼
    ┌─────────────────────────────────────────────────────────────────────────────┐
    │ Architecture Decision Governance Engine [ADRARCH-CORP-001]                 │
    │   ├── Enforces MADR 3.0 Standard: Context, Options, Drivers, Consequences   │
    │   ├── State Machine: Proposed ──► [ARB Review] ──► Accepted / Superseded   │
    │   └── Decision Linter: Validates RFC 2119 Normative Rules & Criteria Weights│
    └──────────────────────────────────────┬──────────────────────────────────────┘
                                           │
             ┌─────────────────────────────┴─────────────────────────────┐
             ▼ (State: Accepted)                                         ▼ (State: Superseded)
    [ Immutable Git Repository: `docs/adr/` ]                   [ Superseded Pointer: `docs/adr/` ]
      ├── Cryptographic Commit Hash (Git SHA)                     ├── Points to Successor: `ADR-0042`
      ├── Bi-Directional Code Repository Link                     └── Historical Rationale Preserved
      └── Incident ADR-4919 Regressions Permanently Barred        └── Zero Tribal Knowledge Loss
    

    Criteria and weights

    CriterionWhy it matters hereWeightSource of the weight
    Institutional Memory & Regression DefenseRe-implementing rejected options caused ADR-4919 ($3.8M duplicate outage).0.40David O'Reilly (Chief Enterprise Architect)
    Standardized MADR 3.0 Structure CompletenessADRs lacking explicit trade-offs or negative consequences hide technical debt.0.30Elena Rostova (Head of Architecture Governance)
    Automated CI/CD Decision Linter GatingUn-reviewed architectural deviations must be blocked at the pull request seam.0.15Architecture Review Board (ARB) Charter
    Cross-Service Decision Linkage & TraceabilityDownstream squads must see which global ADRs govern their service boundaries.0.15Core Software Engineering Guild Policy

    Comparison

    Decision Governance ApproachKnowledge DurabilityRegression DefenseAutomated EnforcementEvaluation
    Option A: Informal Slack / Confluence (Legacy)Zero (Forgotten in ADR-4919)None (Re-implemented bad options)NoneRejected: Caused ADR-4919 disaster; unviable.
    Option B: Heavyweight Architecture CommitteeModerate (Slow committee minutes)Moderate (Slow approval bottleneck)ManualRejected: 6-week committee delays stall agile delivery.
    Option C: Git-Based MADR 3.0 + CI Linter (Chosen)Permanent (Git immutable history)Absolute (Explicit rejected options)Automated PR Build BreakersSelected: Fast, transparent, auditable, proven.

    Result

    Option C is selected. Standardized MADR 3.0 records stored in Git are mandatory for all architectural decisions; decisions follow a strict 4-state lifecycle (Proposed, Accepted, Deprecated, Superseded); automated GitHub Actions linters enforce compliance.


    Required Mechanisms

    1. MADR 3.0 Decision Structure & Lifecycle [MC-MS-01]
    • Standardized Sections:
      • Title: Short noun phrase with stable ID (e.g. ADR-0042: Event-Driven Outbox for Settlement).
      • Context and Problem Statement: Real-world problem, volumetric load, and tensions.
      • Decision Drivers: Weighted non-functional requirements (latency, consistency, cost).
      • Considered Options: Minimum 3 options with pros, cons, and explicit rejection rationales.
      • Decision Outcome: Chosen option with positive and negative consequences.
    • Decision State Machine:
      $$\text{Proposed} \xrightarrow{\text{ARB Review}} \text{Accepted} \xrightarrow{\text{Technology Evolution}} \text{Deprecated} \xrightarrow{\text{New Decision}} \text{Superseded by ADR-XXXX}$$
    2. The ADR-4919 Decision Regression Defense [MC-RD-01]
    • In incident ADR-4919, an engineer implemented an un-indexed MongoDB query layer that had been rejected 18 months prior in an un-recorded hallway conversation.
    • Architectural Remedy:
      • The considered options section of every ADR must explicitly record

    Rejected Alternatives and document the exact quantitative failure mode that disqualified them.

    • Pull requests introducing major database or messaging dependencies require an approved ADR link; CI checks verify that the selected technology does not conflict with active Accepted or Superseded ADRs.
    3. Automated ADR Linter & Validation Gate (adr-lint) [MC-AL-01]
    • Automated GitHub Actions linter parses new ADR markdown files:
      • Verifies presence of all required MADR 3.0 sections.
      • Verifies that at least 2 alternative options were considered.
      • Verifies that status is one of: Proposed, Accepted, Deprecated, Superseded.
      • If status is Superseded, validates that a valid successor ADR link is present.

    Invariants and Contracts

    Mandatory MADR 3.0 Template Conformance [INV-ADR-01]
      Architecture Decision Records must follow the standardized MADR 3.0 markdown format.
      Authoring free-form, un-structured architecture decision memos lacking negative consequences is prohibited.
    
    Mandatory Documented Rejected Options [INV-ADR-02]
      Every ADR must document at least two alternative options and explain the exact technical rationale for rejection.
      Submitting single-option ADRs that treat architecture decisions as foregone conclusions is barred.
    
    Strict Immutability of Accepted Decisions [INV-ADR-03]
      Accepted ADRs are immutable historical records; they must never be edited in-place to alter past decisions.
      Changing an accepted architectural direction requires authoring a new ADR that explicitly marks the prior ADR Superseded.
    

    Explicit Unknowns

    • Time required for newly onboarded offshore development squads to adopt MADR authoring conventions (G-1).
    • Semantic search index recall accuracy when querying 500 historical ADR markdown files using local embeddings (G-2).

    Traceability

    ClaimClassificationSourceFreshness
    65 microservices across 450 engineersprovidedSoftware delivery organization intakeCurrent
    120 architectural migrations/yearprovidedArchitecture review volume intakeCurrent
    Incident ADR-4919 $3.8M duplicate outageprovidedHistorical operations post-mortemHistorical
    MADR 3.0 standard and 4-state lifecycleprovidedEnterprise Architecture Governance CharterCurrent
    Git-Based MADR + Automated Linter selecteddecidedDavid O'Reilly & Elena Rostova2026-09-15
    Mandatory MADR 3.0 invariant INV-ADR-01decidedArchitectural invariant INV-ADR-012026-09-15

    Verification

    No validator was supplied, so no command was run.

    Reviewer self-check against architecture decision standards:

    • Structure Completeness: PASS. Enforces full MADR 3.0 format with drivers, options, and consequences.
    • Regression Defense: PASS. Mandatory rejected options section permanently closes ADR-4919 flaw.
    • Lifecycle Integrity: PASS. Enforces immutable 4-state lifecycle with automated PR linters.
    • Markdown Hygiene: PASS. Native Markdown syntax strictly adheres to rule_markdown.md.

    Open Decisions

    • DEC-ADR-01: Elena Rostova to determine whether ADR approval authority should be delegated to squad tech leads for internal service decisions while reserving ARB approval for cross-service boundaries in Q1 (Owner: Elena Rostova).

    Next steps

    1. Architecture Governance Guild publishes the official MADR 3.0 repository template.
    2. DevOps team integrates the adr-lint GitHub Actions workflow across all 65 microservice repositories.
    3. Conduct quarterly architectural retrospective reviewing all Accepted and Superseded ADRs.

    skill: architecture-decision-architect

    Architecture Decision Governance — Fitness Self-Check [ADRARCH-CORP-FIT-001]

    Summary

    This fitness self-check evaluates the architecture decision governance platform against three critical red-capable domain failure probes: dual writer, undefined grain, and silent schema drift. All targeted probes pass by design construction. A self-check is supporting evidence, never the authoritative gate. Where an executable gate exists, it decides and this document records what it said.

    Detailed Description

    Criterion [FIT-n]ProbeEvidenceResultLimits of the claim
    FIT-1: Dual WriterSeed an ADR repository workflow where two concurrent pull requests attempt to claim the identical sequential ADR number (ADR-0042) simultaneously without coordination.Git branch protection and sequential ADR numbering validator probe_duplicate_adr_number_collision verifying second PR rejection with diagnostic ERR_DUPLICATE_ADR_NUMBER_ALLOCATION.passConfirms GitHub Actions PR number checks; does not evaluate local un-pushed Git tags.
    FIT-2: Undefined GrainSeed a proposed ADR that records a database selection decision without declaring an explicit service scope grain, bounded context, or corporate organizational unit.ADR metadata schema linter probe_missing_adr_scope_grain verifying submission failure with diagnostic ERR_ADR_METADATA_LACKS_DECLARED_SCOPE_GRAIN.passConfirms automated MADR linter validation; does not inspect ad-hoc temporary draft notes.
    FIT-3: Silent Schema DriftSeed a pull request that deletes or renames a mandatory MADR 3.0 section heading (Negative Consequences -> Downsides) without updating the ADR governance schema.MADR 3.0 structural validator probe probe_madr_section_schema_drift verifying PR rejection with diagnostic ERR_MADR_MANDATORY_SECTION_SCHEMA_DRIFT_DETECTED.passConfirms automated CI markdown AST linters; does not evaluate plain text email memos.

    Residual Risk

    • Latency overhead (up to 15 seconds) in PR check workflows when checking cross-repository markdown link integrity across 65 separate microservice repos. Accepted by Elena Rostova with asynchronous GitHub status webhooks.

    Traceability

    ClaimClassificationSourceFreshness
    Rejection of duplicate ADR number collisionsderivedFIT-1 probe result2026-09-15
    Rejection of ADRs lacking declared scope grainderivedFIT-2 probe result2026-09-15
    Rejection of MADR section schema driftderivedFIT-3 probe result2026-09-15

    Verification

    No validator was supplied, so no command was run.

    Open Decisions

    None.

    Next steps

    1. Architecture Guild incorporates ADR fitness probes into automated organization-wide GitHub repository templates.
    2. Platform team configures Slack notifications alerting engineering squads when relevant global ADRs are marked Accepted.
    3. Conduct semi-annual audit verifying that production codebase implementations match Accepted ADR contracts.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    What it does

    This skill owns the decision-system architecture through which consequential architecture choices retain authority, context, alternatives, evidence, consequences, lifecycle, implementation traceability, and realized outcomes. It coordinates a portfolio and graph of decisions rather than choosing technologies, approving architecture, or writing one ADR on demand.

    Use it when

    • Decisions span products, systems, teams, contracts, data, security, infrastructure, delivery, and operational lifecycles
    • Authorities need a shared rule for what warrants a decision record and who may propose, review, accept, reject, waive, reopen, supersede, or retire it
    • Context, drivers, constraints, assumptions, unknowns, evidence, options, rationale, consequences, and dissent must remain attributable
    • Decision dependencies, conflicts, precedence, scope overlap, principles, requirements, risks, and downstream artifacts need a maintained graph
    • Proposed, accepted, implemented, verified, effective, superseded, deprecated, and retired states must not collapse
    • Implementation drift, stale assumptions, expired exceptions, incidents, changed conditions, or failed outcomes must trigger reassessment

    For example: “Teams keep re-debating database choices every quarter because past decisions are buried in Slack threads, and half our services still use an unapproved legacy ORM.”

    What you get

    • architecture/architecture-decision-architect/README.md
    • architecture/architecture-decision-architect/00-overview/architecture-decision-architect-overview.md
    • architecture/architecture-decision-architect/verification/fitness-self-check.md

    Plus one page per business module, only where your evidence calls for it: {module}/glossary.md, {module}/alternatives.md, {module}/deprecations.md.

    All paths are relative to the output folder you choose.

    What it will not do

    Do not use merely to write one ADR, select a technology, give a recommendation, run a generic design review, capture meeting minutes, draft an RFC, log a business decision, update an issue/roadmap, document an already-decided implementation, format a template, or create a decision-log entry.

    How it works

    1. Check decision-system scope is required.
    2. Define admission criteria.
    3. Establish authority and role boundaries.
    4. Define decision status state machine.
    5. Establish realization and drift tracking.
    6. Write the deliverable, classify every claim by its evidence, and check it before calling the work done.

    What's in the package

    Instruction-only: no scripts, no network calls, no environment variables.

    • LICENSE.txt
    • SKILL.md
    • agents/openai.yaml
    • assets/output-template-artifact.md
    • assets/output-template-decision.md
    • assets/output-template-domain.md
    • assets/output-template-fitness.md
    • assets/output-template-mechanism.md
    • references/domain-rules.md
    • references/operating-rules.md
    • references/output-contract.md

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Technical Documentation.

    Do not use merely to write one ADR, select a technology, give a recommendation, run a generic design review, capture meeting minutes, draft an RFC, log a business decision, update an issue/roadmap, document an already-decided implementation, format a template, or create a decision-log entry.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    Permissions

    Read Files
    Write Files

    Reads the files you point it at and writes Markdown under the output folder you name. No network access, no environment variables, and no bundled scripts.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. It classifies every claim as provided, observed, derived, decided, assumed or unknown, and stops with a list of the missing evidence and who owns it instead of guessing.

    No. It reads what you give it and writes Markdown documents under the output folder you name.

    Any agent that supports SKILL.md skills, including Claude Code, Codex CLI, Cursor and Gemini CLI. It has no scripts and makes no network calls.

    Unzip it into your agent's skills folder, for example: unzip architecture-decision-architect.zip -d ~/.claude/skills/

    New in Technical Documentation