claude

AIDLC Core

AIDLC (AI Development Life Cycle) orchestrator: reads work items (Jira/ADO/markdown), routes them through requirements, planning, implementation, review, QA and PR — across one repo or a multi-repo workspace, with or without a git remote, end to end. Stack-agnostic by design; language and framework expertise comes from stack packs.

0 on GitHubby Sourabh Sudheerv0.71.1MITupdated 15 days agodevelopment
/

Plugin aidlc from marketplace aidlc in Sourabh1095/AIDLC. Commands read from the repo's manifest — 13 days ago.

This plugin runs code on your machine — its hooks fire automatically on events like tool use and it starts local MCP servers. The exact commands are listed below; read them before installing.

What it installs

Skills (52)

  • adopt-adr

    Write retroactive ADRs for the decisions the code already embodies, one at a time behind approval, citing the code and leaving the rationale for a person. Use after /aidlc:adopt when docs/adr/ is empty.

  • adopt-apply

    Apply an approved adoption profile — writes aidlc.config.json, the CLAUDE.md facts and the git-workflow rule behind a full diff and your approval, merging with what people wrote. Use after /aidlc:adopt, or to re-apply an updated profile.

  • adopt-backlog

    File the debt an adoption scan found (missing gates, untested critical paths, end-of-life dependencies, TODO clusters) as deduplicated items with acceptance criteria, each shown before it is created. Use after /aidlc:adopt.

  • adopt

    Read-only scan of an existing workspace (one repo or several) into an evidence-backed profile — topology, stack, real gates, CI, git conventions, runtime constraints, undocumented decisions and debt. Re-run it to report drift. Use when adopting AIDLC on code that already exists.

  • agent-contract

    What every AIDLC subagent owes its orchestrator — the finish contract, the report-back shape and the tool-restriction policy. Load when writing, reviewing or scaffolding an agent.

  • architecture

    ADR triggers and format, and the explore-before-deciding checklist. Load when planning a medium-or-larger work item or making a hard-to-reverse technical decision.

  • bootstrap

    Plan a batch of requirements into the tracker in one reviewed pass — a whole backlog from a requirements document, or a document of new features for a product that already ships. Epics to tasks with acceptance criteria, team assignment and sprints. Use instead of /aidlc:intake for more than one requ

  • brain

    The project brain (.aidlc/brain/) — a sourced page per module, plus user types, flows and vocabulary. Load before explaining, planning or changing a module, when taking over or handing over one, or after learning how part of the product works.

  • ceremony

    The four process tiers (answer, direct, tracked, full) and how to pick one. Load before deciding whether a request needs a work item, a branch or a PR.

  • ci-cd

    CI authoring and red-check diagnosis for GitHub Actions and Azure Pipelines, any stack. Load when changing workflow YAML or diagnosing failing checks; on a Node repo, together with aidlc-stack-web:ci-web.

  • code-review

    The review checklist, severity taxonomy and finding format for the verify phase. Load when reviewing a pipeline diff against acceptance criteria and standards.

  • debugging

    The bug protocol — reproduce, failing test, root cause, minimal fix, regression sweep. Load for any bug item or when diagnosing a failure during a run.

  • do

    AIDLC's front door for any prompt — it grounds in the project (config, runs, backlog, ADRs), then answers or acts at the lightest ceremony that fits. Use for fit and opinion questions ("should we use X?"), investigations, small fixes (a typo, a rename) and plain "build this / fix this" requests that

  • docs-writing

    What to update per change type (README, CHANGELOG, API docs) and how to write it. Load when updating project documentation for a work item.

  • doctor

    Read-only diagnosis of an AIDLC workspace (config, plugins, permissions, hooks, run files, tracker and CLI auth), each problem with its fix. Use when /aidlc:* commands are missing or misbehave, a run blocks on permissions, or after installing or updating the plugins.

  • dogfood

    Log friction with the AIDLC plugin itself (wrong or missing guidance, workarounds, skill or template bugs, not project bugs) to a local inbox for its maintainer. Use when the plugin is the problem, or on /aidlc:dogfood.

  • facts

    Project facts (.aidlc/facts.md) — dated operational truths the pipeline keeps relearning. Read before diagnosing an environment failure or briefing an implementer; add one whenever a run learns something the hard way.

  • git-workflow

    Branch naming, conventional commits, push and PR creation (GitHub via gh, Azure Repos via az). Load when branching, committing or opening a pull request for a work item.

  • groom

    Backlog refinement — fix weak acceptance criteria, size items, flag blockers and stale items, propose splits and priority changes. Use when asked to groom, refine or clean up the backlog.

  • init

    Scaffold the AIDLC project template into the current repository — CLAUDE.md, permissions, aidlc.config.json, rules, backlog structure, run-state folders. Use when adopting the AIDLC framework in a new or existing project.

  • insights

    Read-only outcome metrics for this workspace's pipeline — time to done, fix cycles, findings by source, month by month. Use when asked how the pipeline is doing, whether a change helped, where time goes, or what keeps coming back in review.

  • intake

    Turn a plain-language requirement into backlog items with acceptance criteria in the active tracker, deduplicated against the board. Use when the user describes something to build or fix that is not yet a work item.

  • journal

    The workspace journal (.aidlc/journal.md), one line per event, read back at session start. Load when finishing a command that changed something, or to see what has been happening here.

  • learn

    Build or refresh the project brain — map the modules from the code, read one deeply, or refresh pages the code has moved past. Use when taking over a module, starting AIDLC on an existing project, or when brain pages are behind the code.

  • maintenance

    Dependency updates, deprecation sweeps and safe refactors. Load for maintenance items, dependency updates or scheduled maintenance.

  • next

    Pick the highest-priority ready item and run it through the pipeline. Use when asked to work on the next item, pick up work or continue with the backlog.

  • planning

    Decomposition, sizing and dependency detection for AIDLC plans. Load when writing an implementation plan or sizing and sequencing backlog items.

  • promote

    Promote a project-local skill or agent into the shared AIDLC marketplace plugin — validate, generalize, package and open a PR against the marketplace repo. Use when a local extension has proven reusable (reuseCount >= 2) and should become available to all projects.

  • release

    Cut a release — semver from conventional commits, changelog, tag and notes for GitHub or Azure DevOps. Use for release items or when asked to cut, prepare or tag a release.

  • remove

    Remove AIDLC from this project — delete its own files, revert only what it merged, keep everything the team wrote. Shows the plan first and verifies the project's files are back to how they were. The plugin itself goes with /plugin uninstall.

  • replan

    Re-sequence not-yet-started work into waves when priorities change ("checkout before search", "all backend first", a revised spec), saved as the plan /aidlc:next and /aidlc:sprint follow. Use when asked to re-plan, re-prioritize, re-sequence or re-phase.

  • repo

    Declare and bootstrap a repo in a poly workspace — add it to repos[], git-init it and make a base commit. Use when work references a repo that is not declared yet.

  • requirements

    The acceptance-criteria bar, INVEST checklist and assumption-logging protocol. Load when validating, refining or writing acceptance criteria.

  • research

    The spike protocol — sharpen the question, gather cited evidence, verify the load-bearing claims, deliver a decision. Load for a spike item or a technical choice mid-run.

  • reuse

    Use BEFORE writing or planning new code (a feature, screen, component, hook, helper, client, type or test) — find what already exists and extend it instead of writing a second one, and catch what a branch duplicated. Load for any implement, plan or review task, including plain requests like "add X t

  • review-feedback

    Work a reviewer's PR comments like pipeline findings — fetch the unresolved threads (GitHub or Azure Repos), fix them through the fix cycle, push and reply on each. Use when a PR has review comments to address.

  • run-state

    The run-file format (.aidlc/runs/<ID>.md) and checkpoint/resume protocol. Load when creating, updating, resuming or archiving a pipeline run.

  • run

    Run one work item (epic, story, task, bug or spike) through the full AIDLC pipeline — fetch, requirements, plan, implement, review + QA with fix cycles, PR, tracker update. Resumable. Use when asked to work on, implement, fix or deliver a work item by ID.

  • scaffold-agent

    Create a project-local subagent from the AIDLC template. Use only when a role needs its own context, tool surface or independent judgment, and no existing agent or skill covers it.

  • scaffold-skill

    Create a reusable project-local skill from the AIDLC template. Use when a needed procedure, convention or integration exists in no installed plugin or local skill.

  • security

    The security review procedure — OWASP-oriented diff review, dependency audit, secret patterns, authz reasoning. Load for a security pass on a work item or a dependency audit.

  • sprint

    Run several independent work items in parallel, each as a Claude Code background session — routed per repo from the control plane in poly, or in isolated git worktrees in mono — and aggregate progress into one board that says which run needs you. Use when asked to work multiple backlog items at once

  • status

    The AIDLC dashboard — active runs with phase, branch and PR, plus ready backlog items. Use when asked about AIDLC progress, active runs or what's next.

  • study

    Write the business study of the product from the project brain — users, modules, flows, data, decisions, intended versus built, open questions, every claim sourced. Use when a project ends, for a handover, or when asked for a document that explains the whole product.

  • sync

    Reconcile project-local AIDLC extensions against the installed plugin versions — detect promoted skills now shadowed by the plugin, remove the local fork, and report drift. Use after plugin updates or when local and plugin skills may overlap.

  • testing

    Test policy — the pyramid, what to test per change type, coverage and flaky tests. Load when writing or evaluating tests in the verify phase.

  • undo

    Reverse an AIDLC run — close its PR and delete its branch, or revert it through a new PR if it merged — and put the board back. Never rewrites history; shows the plan first. Use when a run should not have happened.

  • upgrade

    Catch a project's AIDLC files up with the installed plugin — config and permission-rule migrations, rules and CLAUDE.md drift, every change shown first. Use after /plugin marketplace update, or when /aidlc:doctor says the config is old.

  • wi-ado

    Work-item adapter for Azure DevOps Boards (MCP server, az CLI fallback). Load when workItems.source is "ado".

  • wi-jira

    Work-item adapter for Jira (Atlassian MCP server). Load when workItems.source is "jira".

  • wi-markdown

    Work-item adapter for the markdown backlog (backlog/). Load when workItems.source is "markdown".

  • work-items

    The canonical WorkItem schema, adapter contract and tracker routing. Load whenever reading or updating epics, stories, tasks, bugs or spikes in any tracker.

Subagents (9)

  • aidlc-analyst

    AIDLC business analyst — validates and refines acceptance criteria, sizes items, splits epics, logs assumptions. Dispatched by /aidlc:run (requirements) and /aidlc:groom.

  • aidlc-architect

    AIDLC architect — explores the codebase and writes the plan for medium-and-larger items, plus ADRs for hard-to-reverse decisions. Dispatched by /aidlc:run.

  • aidlc-devops

    AIDLC DevOps — Dockerfiles, compose, CI pipelines, release mechanics and failing-check diagnosis. Dispatched by /aidlc:run for infra, CI and release items and red PR checks.

  • aidlc-docwriter

    AIDLC documentation writer — keeps README, API docs and CHANGELOG true to an item's changes; polishes ADRs. Dispatched by /aidlc:run in the docs phase.

  • aidlc-implementer

    AIDLC implementer — writes production code for a work item from an approved plan, tests green at every commit. Dispatched by /aidlc:run for the implement phase and fix cycles.

  • aidlc-qa

    AIDLC QA — proves the implementation works, runs the suite, adds missing tests, and writes the failing repro test before a bug fix. Dispatched by /aidlc:run.

  • aidlc-researcher

    AIDLC technical researcher — answers open technical questions with cited evidence (library choice, feasibility, unknown tech); not design research. Dispatched by /aidlc:run for spikes and mid-run unknowns.

  • aidlc-reviewer

    AIDLC adversarial code reviewer — reviews an item's diff against acceptance criteria and standards, never sharing the implementer's context. Dispatched by /aidlc:run in the verify phase.

  • aidlc-security

    AIDLC security reviewer — OWASP review, dependency audit, secret scan and authz reasoning on an item's diff. Dispatched by /aidlc:run for security paths, dependency changes or security items.

MCP servers (1)

  • context7

    runs: npx -y @upstash/context7-mcp

Hooks (5)

  • PreToolUse

    node "${CLAUDE_PLUGIN_ROOT}/hooks/scripts/guard.mjs"

  • PreToolUse

    node "${CLAUDE_PLUGIN_ROOT}/hooks/scripts/dep-vet.mjs"

  • PreToolUse

    node "${CLAUDE_PLUGIN_ROOT}/hooks/scripts/protect-paths.mjs"

  • PreToolUse

    node "${CLAUDE_PLUGIN_ROOT}/hooks/scripts/env-guard.mjs"

  • SessionStart

    node "${CLAUDE_PLUGIN_ROOT}/hooks/scripts/session-context.mjs"

View source on GitHub

More from Sourabh1095/AIDLC

All 6 plugins →

UI/UX pod that starts from whatever already exists. Screens in Figma → it ships the Figma MCP, extracts the spec and variables, builds to the design and gates on fidelity, with the jury offered rather than imposed. A brand design system in Figma → its variables and components become the project's tokens and vocabulary (workspace-wide), and the pod designs the screens within them, jury still gating. Neither → UX-narrative writer, Awwwards inspiration researcher, design-system owner, motion specialist, and a strict unbiased jury that renders the UI and gates on ≥9/10. New or existing projects, page or whole-app scope, honors brand references (logo/colors/fonts). Bundles the Figma and Playwright MCP servers it needs, so a backend-only workspace installs neither. Enabled by default; it only activates on UI items, so backend-only projects are unaffected.

8 skills · 7 agents · 2 MCP servers

0

Default web stack expertise for the AIDLC pipeline: TypeScript standards + strict tooling baseline (ESLint/Prettier/tsconfig/.gitattributes + optional husky/lint-staged) + enterprise project structure (scaffold + dependency-cruiser boundary gate) + the Node/TS CI half (templates, the npm gate, isolated-checkout dependency resolution, cross-platform lockfiles) + the Prettier auto-format hook, Next.js, NestJS, Postgres, MongoDB, DB migrations, Docker, API design.

10 skills · hooks

0

The Azure DevOps MCP server for AIDLC's ADO adapter. Optional — the adapter's `az boards`/`az rest` tier covers every operation and is what headless runs already use, so ADO works without this. Install it for the richer interactive tier; needs ADO_MCP_ORG in the launching shell.

1 MCP server

0

The Atlassian MCP server for AIDLC's Jira adapter. Required when workItems.source is "jira" — that adapter has no CLI fallback. Install only on Jira projects; core ships no tracker server, so a markdown or ADO workspace starts neither.

1 MCP server

0

A mod (Claude Code 2.1.287+) that shows AIDLC beside the chat: a line above the prompt with the runs in flight and anything waiting on you, a board pane for runs and the sprint, /aidlc-board with no model turn, a desktop notification when a sprint run needs you, the module you're changing and what uses it, this session's files and checks, each epic as a tree of its stories, tasks and bugs, and what each work item cost by phase (feeds /aidlc:insights). It draws what core's scripts report, calls no model and gates nothing; it writes only its git-ignored cost ledger and board copy. Optional: AIDLC works the same without it, which matters where an organization allows only its own mods.

0

Similar plugins

superpowers

claudecodex

Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques

15 skills · hooks

obra/superpowers 316,836

context7

claudecodex

Up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your coding agent's context.

1 MCP server

upstash/context7 64,709

Build powerful, stateful CLI interfaces for any GUI application using the cli-anything harness methodology.

5 commands

HKUDS/CLI-Anything 58,665

Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer

8 skills · 1 MCP server

ChromeDevTools/chrome-devtools-mcp 55,181

Automates browser interactions for web testing, form filling, screenshots, and data extraction

12 skills

vercel-labs/agent-browser 50,373

Use Alibaba Open Code Review from local Codex.

2 skills

alibaba/open-code-review 50,068

Maintain AIDLC Core?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex plugins6 plugins