Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Architecture Decision Records

    1

    Turn hard-to-reverse architecture choices into searchable repo documents that capture options, tradeoffs, rollout plans, and validation criteria.

    Free

    3 usesSecurity scanned
    Architecture Decision Records

    Architecture Decision Records

    Example session with this skill installed

    We're about to switch our campaign tracker from polling the REST API to an event-sourced local store in IndexedDB and the team is split. Write up the decision properly before we touch code. Repo is at /home/dev/campaign-tracker, existing notes live in docs/adr. Hard constraints: offline-first on mobile Safari, no new server dependency, two junior devs maintain it. Compare keeping polling, a service-worker cache, and the event-sourced option, then document tradeoffs, migration phases, rollback plan, and how we'll validate that startup and sync latency don't regress.

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

    Summary

    ER-0007 (docs/adr/er-0007-event-sourced-campaign-store.md, status: Proposed) records the move from REST polling to an offline-first, event-sourced IndexedDB store; keeping polling and a service-worker cache layer were the rejected alternatives.

    AreaExpected effectConfidence
    Startup / load timeDecrease: first paint no longer blocks on /campaignsHigh
    NetworkDecrease ~90%: batched sync replaces 5s pollingMedium
    MemoryIncrease: replay buffer, est. +8 MBLow
    Disk / assetsIncrease: journal needs a compaction jobMedium
    Build timeNeutral: no new dependencyHigh

    Next steps

    • Backend and mobile leads sign off on the event replay contract (section 5)
    • Land the storage abstraction behind feature/er-0007-flag before Phase 2
    • Add the 10k-event replay benchmark and compaction test
    • Mark ER-0002 (polling client) Superseded when Phase 4 deletes the old path

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

    About this skill

    The problem

    Engineering teams often lose the "why" behind critical architectural choices. Months later, developers struggle to understand why a specific database, framework, or communication protocol was chosen, leading to costly relitigation or brittle workarounds.

    What it does

    • Scans the repository to identify existing ADR patterns and determines the next sequence number.
    • Analyzes codebase constraints, including manifests, lockfiles, and CI configurations.
    • Drafts comprehensive Engineering Choice Records covering tradeoffs, rejected alternatives, and rollout plans.
    • Generates actionable validation criteria and migration steps for implementers.
    • Maintains a searchable history of decisions by linking related records and marking superseded notes.

    Why this beats prompting it yourself

    Generic prompts often produce shallow summaries that lack technical rigor or honest tradeoffs. This skill enforces a structured engineering discipline, ensuring every record includes concrete system impact estimates, ASCII diagrams, and specific rollback strategies that a standard LLM would likely omit.

    Use cases

    • Documenting the transition from REST to GraphQL for internal service communication.
    • Justifying the selection of a specific state-management library in a frontend project.
    • Recording the decision to move from local JSON storage to an embedded database like SQLite.
    • Defining service boundaries and ownership during a monolith-to-microservices migration.

    Known limitations

    Requires an existing directory structure or permission to create one. Performance and cost impact estimates are based on provided context and not live telemetry.

    Requires a destination directory for docs. Impact estimates are qualitative unless specific benchmarks are provided. Confined to repository-local context.

    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 any payment.

    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

    3 uses

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 months ago

    • Free with an Agensi account

    Permissions

    No special permissions declared or detected

    Listed2 months ago

    What's inside

    Frequently Asked Questions

    The skill automates the creation of Architecture Decision Records (ADRs) by analyzing your existing codebase constraints and historical decisions to ensure technical rigor and consistent formatting.

    It is designed to work with any agent that has filesystem access; it identifies your repository's structure to determine the correct ADR numbering and placement automatically.

    Beyond just summaries, this skill specifically generates technical tradeoffs, ASCII diagrams for visual clarity, and concrete rollback strategies that standard LLM prompts often miss.

    Yes, the skill is designed to scan your repository history to link related decisions and will automatically mark previous ADRs as 'Superseded' when a new choice replaces an old one.

    Upon purchase, you receive the full skill configuration, documentation on how to trigger ADR generation, and any updates or refinements released for the lifecycle of the product.

    The skill relies on the context provided within your codebase and manifests; it does not access live performance telemetry or real-time infrastructure costs.

    New in Technical Documentation