By lkzppm
Spec-driven dev ecosystem for Claude Code. Turns CLAUDE.md into a curated index of spec/*.md files, ships a meta-agent that scaffolds per-project specs/agents/hooks tailored to the detected stack, and bundles MCP servers for spec-linting and offline token-counting (gpt-tokenizer, no API key required).
Add a single sub-agent, skill, or hook to .claude/. Prompts interactively if details are missing.
One-time scaffolding — creates spec/, rewrites CLAUDE.md as a brief + spec index, then interactively suggests per-project hooks/skills/sub-agents
Read-only health check of spec/, CLAUDE.md, and .claude/ — reports issues, does not modify files
Add a new spec — either spec/<topic>.md for a concept/subsystem or spec/pipelines/<topic>.html for a workflow walkthrough (with --pipeline)
Refresh specs to match current code — drift detection across all markdown specs, or a single spec when a path is given (markdown or pipeline-spec HTML)
Add a single project-local Claude Code automation — sub-agent, skill, or hook — into .claude/. Use when the user runs /adam:claude-add, /claude-add, or asks to "add a sub-agent that...", "add a skill for X", "add a hook that runs Y on file write", "wire ruff into claude-code", "give me a code-reviewer agent for this project", or "create a new agent/skill/hook for the project". Distinct from /adam:setup (which suggests a batch); this adds exactly ONE artifact at a time, prompting for missing details via AskUserQuestion.
One-time scaffolding of a spec-driven Claude Code workflow in this project. Use when the user runs /adam:setup, /setup, or asks to "set up adam", "scaffold spec/ folder", "init spec-driven docs", "convert CLAUDE.md into a spec index", or "personalize claude code for this repo". Pipeline is strictly ordered: gitnexus analyze → spec scaffolding → per-class AskUserQuestion menus (hooks, subagents, skills) → create only what the user picks → CLAUDE.md → spec-lint verify → final brief.
Lint the spec-driven Claude Code ecosystem in this project — checks spec/, CLAUDE.md, and .claude/ for integration issues without rewriting anything. Use when the user runs /adam:spec-audit, /spec-audit (in this plugin's namespace), or asks to "lint specs", "check spec ecosystem", "verify CLAUDE.md is wired up", "audit my claude code setup", or "is my spec-driven workflow healthy?". Read-only — reports issues but does not modify files. For drift fixing, point users at /adam:spec-update.
Add a single new spec/<topic>.md when a fresh concept, subsystem, or integration enters the project — OR add a pipeline-spec (HTML workflow walkthrough) at spec/pipelines/<topic>.html when the user passes --pipeline or asks for a "flow chart", "pipeline", "workflow diagram". Use when the user runs /adam:spec-create, /spec-create, or asks to "add a spec for X", "document the new <subsystem>", "create a spec covering <topic>", "we just added <X>, write its spec", "draw a pipeline for <flow>", "add a workflow walkthrough for <X>". Will refuse if spec-driven scaffolding (`spec/INDEX.md`, `CLAUDE.md`) is missing — points the user at /adam:setup instead.
Refresh existing specs to match the current code. Use when the user runs /adam:spec-update, /spec-update, or asks to "update spec/", "refresh project docs", "audit specs against code", "check CLAUDE.md against the code", or "sync the spec after my refactor". Detects drift between spec/*.md and the actual working tree, rewrites stale specs, merges duplicates, and refreshes spec/INDEX.md plus the CLAUDE.md spec index table (including token counts). Accepts an optional path argument to update a single spec only.
Admin access level
Server config contains admin-level keywords
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
(Spec-driven + Repo knowledge Graph) package for Claude Code. One /setup and every project ships with a curated CLAUDE.md, a spec/ tree, and a live GitNexus index Claude can query.
adam is a context manager helper for Claude Code. Most coding sessions burn tokens because Claude has to re-derive what your repo is on every turn — grep for files, read whole modules to find one function, scan callers before editing. adam pre-stages that context so Claude doesn't have to:
CLAUDE.md + spec/*.md tree, sized to your project, that frames the codebase in a few thousand cached tokensThe result: Claude spends fewer turns finding code and more turns changing it.
30 paired tasks against honojs/hono (5,854 graph nodes / 8,893 edges), Claude Sonnet, raw JSON in bench/results-hono/.
| Family | n | Δ cost vs vanilla | Quality (tsc-pass) |
|---|---|---|---|
| Orientation — asking about the codebase | 10 | −27.4% | n/a (read-only) |
| Coding — additive (new middleware / helper / Context method) | 10 | −50.4% | 10/10 ↔ 10/10 |
| Coding — multi-file refactor (rename, signature change) | 10 | −30.2% | 10/10 ↔ 10/10 |
| Coding (combined) | 20 | −35.7% | 20/20 ↔ 20/20 |
| Total | 30 | −34.1% | All pass |
Adam saves money on every layer with identical quality. Numbers are reproducible from the scripts in bench/scripts/.
For context: the published AGENTS.md study found developer-provided context files typically raise coding cost by 20–23%. adam moves the needle the other direction.
npm install -g gitnexus # required prereq
/plugin marketplace add lkzppm/adam
/plugin install adam@adam
Then in any project:
/setup
Each slash command is a Claude Code skill pairing deterministic shell scripts with the adam sub-agent. The shell handles what doesn't need an LLM (graph queries, drift detection, stack sniffing, file walking); the model spends tokens only where judgment matters. Natural-language triggers fire the same skills — no need to type the slash form.
npx claudepluginhub lkzppm/adam --plugin adamFeature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Production-grade engineering skills for AI coding agents — covering the full software development lifecycle from spec to ship.
Intelligent draw.io diagramming plugin with AI-powered diagram generation, multi-platform embedding (GitHub, Confluence, Azure DevOps, Notion, Teams, Harness), conditional formatting, live data binding, and MCP server integration for programmatic diagram creation and management.
Complete developer toolkit for Claude Code