The Lead Stack: Agent-Agnostic Workflows

`intent-brief → spec → plan → diff → review-report → release`

A high-performance repository of "Skills" and RTK-powered tools designed for Tech Leads. These workflows are Agent-Agnostic, allowing any LLM agent (Gemini, Claude, GPT) to assist with implementation planning, code review, and automated testing.

Live Web App: https://ai-tech-lead-stack.vercel.app

The overview comes first. It is followed by the web app flow, then the IDE flow, and finally the telemetry both flows share. Colours are the same in every diagram: purple for people, blue for the dashboard, green for the shared core and MCP server, and pink for storage and outside services. Dashed lines are optional or conditional calls. Thick arrows are writes to code or to a Git repo.

flowchart LR
  dev(("Developer<br/>in browser"))
  agent(("IDE coding agent"))
  dash["Dashboard app<br/>apps/dashboard"]

  subgraph core["Shared core · packages/core"]
    mcp["MCP server"]
    libs["Reflexion engine · skills<br/>model resolver · telemetry"]
  end

  repo["Your project repo<br/>on GitHub"]
  code[("Files on your machine<br/>(MCP working directory)")]
  ext[("Postgres · AI providers · Langfuse<br/>E2B · ClickUp · Figma · Discord · Ollama")]

  dev -->|"uses"| dash
  agent -->|"calls MCP tools"| mcp
  agent ==>|"edits"| code
  mcp ==>|"reads · apply_patch writes"| code
  dash -->|"uses"| libs
  mcp -->|"uses"| libs
  dash ==>|"reads · commits to feat/* and discovery/*<br/>branches · triggers Actions"| repo
  dash -->|"app data · models · sandboxes · webhooks"| ext
  libs -->|"app data · models · telemetry"| ext
  mcp -->|"code_search embeddings"| ext

  classDef actor fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
  classDef app fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
  classDef coreNode fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class dev,agent actor
  class dash app
  class mcp,libs coreNode
  class repo,code,ext extNode

Web app flow

Everything the dashboard (apps/dashboard) does. You sign in with GitHub, and routes that touch GitHub use your GitHub token on the project's linked repo. Model calls use the API keys you saved in Settings, and fall back to the server's env keys.

1. Chat and feature discovery

Chat streams replies from your chosen model and saves the conversation. When the project has a linked repo, it reads code from that repo on GitHub. Otherwise its tools read skills and files from the dashboard server's own toolbox checkout. Slash-command workflows always come from the toolbox. The ClickUp and Figma tools only read, and only work when the project has those keys saved. Feature discovery streams a planning conversation; its one tool, write_to_sandbox, runs in your browser (close-up 3). Both routes also read your user and project records.

flowchart LR
  discovery["Feature discovery<br/>api/orchestrator/discovery"]
  chat["Chat<br/>api/chat"]

  db[("Postgres")]
  ai["AI model providers"]
  tools["ClickUp · Figma<br/>read-only"]
  github["Your project repo<br/>on GitHub"]
  toolbox[("Toolbox skills<br/>and workflows")]
  core["Telemetry<br/>(see close-up 9)"]

  discovery -->|"streams replies"| ai
  chat -->|"saves chats and messages"| db
  chat -->|"streams replies"| ai
  chat -.->|"reads tasks and designs"| tools
  chat -.->|"reads code (repo linked)"| github
  chat -->|"slash workflows · skills<br/>when no repo is linked"| toolbox
  chat -->|"records events"| core

  click chat "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/chat/route.ts"
  click discovery "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/orchestrator/discovery/route.ts"

  classDef actor fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
  classDef app fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
  classDef coreNode fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class chat,discovery app
  class core coreNode
  class tools,github,toolbox,ai,db extNode

2. Reflexion from the web app

The Reflexion route reads package.json and tsconfig.json from your linked repo for context, then runs the reflexion engine. The engine drafts a plan, critiques it and revises it, saving progress to the ReflexionRun table after each step. Every step is also recorded as telemetry. You get back a plan and an IDE prompt to apply yourself. The resume route continues a paused run the same way, without reading GitHub.

flowchart LR
  dev(("Developer<br/>in browser"))
  reflexapi["Reflexion route<br/>api/orchestrator/reflexion"]
  github["Your project repo<br/>on GitHub"]
  engine["Reflexion engine<br/>engine.ts"]
  runner["Model runner<br/>providers-user"]
  ai["AI model providers"]
  dbstore["DbStateStore"]
  db[("Postgres<br/>ReflexionRun")]
  telemetry["Telemetry<br/>(see close-up 9)"]

  dev -->|"starts a run"| reflexapi
  reflexapi -.->|"reads package.json<br/>and tsconfig.json"| github
  reflexapi -->|"runs loop"| engine
  engine -->|"drafts and critiques"| runner
  runner -->|"your saved keys,<br/>else env keys"| ai
  engine -->|"saves progress"| dbstore
  dbstore --> db
  reflexapi -->|"records each step"| telemetry
  reflexapi -->|"plan + IDE prompt"| dev

  click reflexapi "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/orchestrator/reflexion/route.ts"
  click engine "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/lib/ai/reflexion/engine.ts"
  click runner "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/lib/ai/reflexion/providers-user.ts"
  click dbstore "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/lib/ai/reflexion/db-state-store.ts"

  classDef actor fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
  classDef app fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
  classDef coreNode fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class dev actor
  class reflexapi app
  class engine,runner,dbstore,telemetry coreNode
  class github,ai,db extNode

3. Sandbox preview

Feature discovery can preview changes in an E2B sandbox, using the E2B key you saved in Settings. Your browser first fetches a zip of the linked repo's default branch, together with the project's decrypted env vars. It then asks the server to boot a sandbox, which uploads the project, installs dependencies and starts a dev server. File writes from the discovery AI go into that sandbox only. Nothing is copied back to GitHub automatically.

flowchart LR
  dev(("Developer<br/>in browser"))
  bundle["Project bundle<br/>api/orchestrator/discovery/project-bundle"]
  repo["Your project repo<br/>on GitHub"]
  boot["Sandbox boot<br/>api/orchestrator/sandbox/boot"]
  fileaction["Sandbox file action<br/>discovery/actions/e2b.ts"]
  e2b["E2B sandbox<br/>(your E2B key)"]

  dev -->|"1 · fetch project"| bundle
  bundle -->|"downloads a zip of<br/>the default branch"| repo
  dev -->|"2 · boot sandbox"| boot
  boot -->|"uploads, installs,<br/>starts dev server"| e2b
  dev -->|"3 · AI file writes"| fileaction
  fileaction -->|"writes files"| e2b

  click bundle "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/orchestrator/discovery/project-bundle/route.ts"
  click boot "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/orchestrator/sandbox/boot/route.ts"
  click fileaction "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/feature-development/discovery/actions/e2b.ts"

  classDef actor fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
  classDef app fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class dev actor
  class bundle,boot,fileaction app
  class repo,e2b extNode

4. Writes to your GitHub repo

These routes change your linked repo on GitHub, using your GitHub token. Feedback commits a screenshot to .github/assets/feedback/ on a branch, comments on that branch's pull request and posts to Discord. Sync commits whatever files it is sent; nothing in the dashboard calls it yet. Both refuse any branch that isn't feat/* or discovery/*, and always block main, master, prod, production and staging. Generate and Audit trigger the ai-prompt-agent.yml and audit.yml GitHub Actions on your repo's main branch. Those workflow files live in your repo, so what they change is defined there, not here.

flowchart LR
  feedback["Feedback<br/>api/feature-development/feedback"]
  sync["Sync<br/>api/feature-development/sync"]
  actions["Generate · Audit<br/>api/orchestrator/generate · audit"]
  repo["Your project repo<br/>on GitHub"]
  discord["Discord"]

  feedback ==>|"commits snapshot PNG<br/>· comments on PR"| repo
  feedback -->|"notifies"| discord
  sync ==>|"commits files to<br/>feat/* or discovery/*"| repo
  actions ==>|"triggers ai-prompt-agent.yml<br/>or audit.yml"| repo

  click feedback "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/feature-development/feedback/route.ts"
  click sync "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/feature-development/sync/route.ts"
  click actions "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/orchestrator/generate/route.ts"

  classDef app fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class feedback,sync,actions app
  class repo,discord extNode

5. Settings, design review and the skill editor

Projects and Settings store app data in Postgres: projects, who can access them, your API keys (encrypted), and per-project model routing. Design review keeps review sessions in Postgres, reads Figma files through a proxy with the project's Figma key, and posts to the project's Discord webhook when a review is ready for the designer. Notify devs posts to a separate Discord webhook. The skill editor drafts new skills with your model. When you submit one, it opens a draft pull request on this toolbox's own repo, not on your project.

flowchart LR
  settings["Projects · Settings<br/>api/projects · api/settings/*"]
  design["Design review<br/>api/design-review"]
  notify["Notify devs<br/>api/notify/devs"]
  skilled["Skill editor<br/>api/skills/chat · skills/actions.ts"]
  db[("Postgres")]
  figma["Figma"]
  discord["Discord"]
  ai["AI model providers"]
  tbrepo["Toolbox repo<br/>on GitHub"]

  settings -->|"projects, access, encrypted<br/>API keys, model routing"| db
  design -->|"review sessions"| db
  design -.->|"figma-proxy reads files"| figma
  design -->|"ready for designer"| discord
  notify -->|"webhook"| discord
  skilled -->|"drafts skills"| ai
  skilled ==>|"opens a draft PR<br/>with the new skill"| tbrepo

  click design "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/design-review/route.ts"
  click notify "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/notify/devs/route.ts"
  click skilled "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/api/skills/actions.ts"

  classDef app fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class settings,design,notify,skilled app
  class db,figma,discord,ai,tbrepo extNode

IDE flow

Everything your coding agent does through the MCP server, which runs on your machine. Every project path the server uses is relative to the directory it was started in. The default install starts it with npm --prefix <toolbox>, so that directory is the toolbox checkout unless your editor or a proxy starts it somewhere else.

6. Skills, knowledge items and pipelines

Skills, workflows, policies and the skill graph are read from the toolbox checkout. If the server finds a client project above its working directory, it also reads that project's .ai overrides. Before any get_* call, the server checks the hook rules in .ai/hooks and can refuse the call. Knowledge Items (KIs) are stored in your home directory, in ~/.gemini/antigravity/knowledge. verify_mission_alignment writes .ai/.mission-alignment.json into the toolbox checkout. get_* and plan_pipeline calls are recorded as telemetry.

flowchart LR
  agent(("IDE coding agent"))
  mcp["MCP server<br/>mcp-server/index.ts"]
  hooks[(".ai/hooks<br/>(working directory)")]
  toolbox[("Toolbox files<br/>.ai/skills · workflows · policies")]
  project[("Project .ai overrides<br/>skills.graph.json · policies")]
  kis[("Knowledge items<br/>~/.gemini/antigravity/knowledge")]
  telemetry["Telemetry<br/>(see close-up 9)"]

  agent -->|"skill, pipeline,<br/>KI and alignment tools"| mcp
  mcp -->|"get_* calls checked first,<br/>can refuse"| hooks
  mcp -->|"reads skills, graph, policies<br/>· writes .mission-alignment.json"| toolbox
  mcp -.->|"reads, if a client project is found"| project
  mcp -->|"reads and writes KIs"| kis
  mcp -->|"get_* and plan_pipeline<br/>record events"| telemetry

  click mcp "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/mcp-server/index.ts"

  classDef actor fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
  classDef coreNode fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class agent actor
  class mcp,telemetry coreNode
  class hooks,toolbox,project,kis extNode

7. Code tools

repo_map and code_search read a prebuilt index in .tls-index, which you create with npm run index:build in packages/core. code_search also calls a local Ollama (OLLAMA_URL, default localhost:11434) to embed your query. read_region reads a line range from a file. apply_patch writes a file after checking .ai/hooks for protected paths. None of these tools restrict paths to the working directory.

flowchart LR
  agent(("IDE coding agent"))
  mcp["MCP server<br/>handlers/codebase.ts"]
  index[(".tls-index<br/>built by index:build")]
  ollama["Ollama<br/>localhost:11434"]
  hooks[(".ai/hooks<br/>protected paths")]
  files[("Files in the MCP<br/>working directory")]

  agent -->|"repo_map · code_search<br/>read_region · apply_patch"| mcp
  mcp -->|"repo_map, code_search read"| index
  mcp -->|"code_search embeds query"| ollama
  mcp -->|"apply_patch checks first"| hooks
  mcp -->|"read_region reads"| files
  mcp ==>|"apply_patch writes"| files
  agent ==>|"edits"| files

  click mcp "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/mcp-server/handlers/codebase.ts"

  classDef actor fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
  classDef coreNode fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class agent actor
  class mcp coreNode
  class index,ollama,hooks,files extNode

8. Reflexion from the IDE

reflexion_loop starts a run in the background and returns a run id straight away. Its _sub_max and _sub_pro variants only change the tier check. Before starting, it looks you up in Postgres by your GitHub CLI or git email, along with the project, so it can use the API keys and model routing you saved in the web app. Env keys are the fallback, and env model variables such as MODEL_PLANNER override saved routing. Progress is saved to .reflexion-out/state.json, or to your OS temp directory when the server runs inside another project. Poll reflexion_status for the result, and use reflexion_resume to continue a paused run; resume skips the Postgres lookup and uses env keys only. Each step is recorded as telemetry, and your coding agent applies the final plan.

flowchart LR
  agent(("IDE coding agent"))
  mcp["MCP server<br/>handlers/reflexion.ts"]
  db[("Postgres<br/>saved keys · model routing")]
  engine["Reflexion engine<br/>engine.ts"]
  runner["Model runner<br/>providers-env"]
  ai["AI model providers"]
  filestore["FileStateStore"]
  files[(".reflexion-out/state.json<br/>or OS temp dir")]
  telemetry["Telemetry<br/>(see close-up 9)"]

  agent -->|"reflexion_loop · status · resume"| mcp
  mcp -.->|"reflexion_loop only: looks up<br/>you (gh/git email) and project"| db
  mcp -->|"runs loop in background"| engine
  engine -->|"drafts and critiques"| runner
  runner -->|"saved keys, else env keys<br/>(resume: env keys only)"| ai
  engine -->|"saves progress"| filestore
  filestore --> files
  mcp -->|"records each step"| telemetry
  mcp -->|"run id, then status and plan"| agent

  click mcp "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/mcp-server/handlers/reflexion.ts"
  click engine "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/lib/ai/reflexion/engine.ts"
  click runner "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/lib/ai/reflexion/providers-env.ts"
  click filestore "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/lib/ai/reflexion/state-store.ts"

  classDef actor fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
  classDef coreNode fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class agent actor
  class mcp,engine,runner,filestore,telemetry coreNode
  class db,ai,files extNode

Shared by both flows

9. Where telemetry goes

Web chat, the skill editor, web reflexion and several MCP tools record events through one telemetry service. It writes each event to the AnalyticsEvent table first. Only if that write succeeds, and TLS_LANGFUSE_* keys are set, does it forward the event to Langfuse. The Metrics dashboard reads those events and reflexion runs. It also pulls traces back from Langfuse into the same table, as does /api/admin/sync.

flowchart LR
  webchat["Web chat · skill editor<br/>api/chat · api/skills/chat"]
  webreflex["Web reflexion<br/>and resume routes"]
  mcpcalls["MCP get_* · plan_pipeline<br/>reflexion_loop · reflexion_resume"]
  telemetry["Telemetry service<br/>telemetry-service.ts"]
  db[("Postgres")]
  langfuse["Langfuse"]
  metrics["Metrics dashboard<br/>dashboard/page.tsx"]

  webchat -->|"records events"| telemetry
  webreflex -->|"records events"| telemetry
  mcpcalls -->|"records events"| telemetry
  telemetry -->|"1 · writes AnalyticsEvent"| db
  telemetry -.->|"2 · then forwards<br/>(TLS_LANGFUSE_* keys)"| langfuse
  metrics -->|"reads AnalyticsEvent<br/>and ReflexionRun"| db
  metrics -.->|"pulls traces back<br/>into AnalyticsEvent"| langfuse

  click telemetry "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/packages/core/src/lib/telemetry-service.ts"
  click metrics "https://github.com/bronz3beard/ai.tech-lead-stack/blob/main/apps/dashboard/src/app/dashboard/page.tsx"

  classDef app fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
  classDef coreNode fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
  classDef extNode fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
  class webchat,webreflex,metrics app
  class mcpcalls,telemetry coreNode
  class db,langfuse extNode

[!NOTE] Related project — SML Gate (small-language-model-gate, CLI slm-gate) — a local AI routing and pre-processing layer that uses a small, free local model via Ollama to intercept, compress, and answer easy or repetitive prompts before they reach your paid subscription or API cloud model, cutting token spend and protecting your monthly quota. Its mcp-gate layer can sit in front of this stack's MCP server (TLS_ADAPTER=on + DOWNSTREAM_MCP pointing at dist/mcp-server.mjs) to condense tool and skill payloads before they hit your editor's context window.

Explore SML Gate on GitHub →

Commands Quick Reference

What you're doing	Call this	Key principle
Leading a multi-agent team	`/dev-team`	Orchestrates sub-agents safely in parallel.
Deep architecture planning	`/plan`	Full codebase audit, solid vertical slices.
Fast lean tasks	`/plan-quick`	High velocity for smaller changes.
Breaking down tickets	`/vertical-slice`	Creates ClickUp-ready tasks (<= 2d).
Local pre-commit check	`/code-review`	4 gates (Spec, SOLID, A11y, Evidence).
Visual testing	`/verify-changes`	Playwright-powered before/after screenshots.
Fixing QA/Regression feedback	`/regression-bug-fix`	Maps impact and remediates safely.
Merging to main	`/pr-automator`	Synthesized diffs with visual proof.
Full feature loop (Sandbox)	`/feature-orchestrator`	End-to-end implementation from idea.
Asking codebase questions	`/ask`	High-density technical advice.

Which tier am I on?

Your plan	Loop to call	Dev-team to call	Capabilities & Isolation
API keys (Gemini+Anthropic)	`reflexion-loop`	`dev-team-orchestrator`	Dual-model SDK enforcement (`validateDistinctModels`), 3+ parallel lanes, uncapped.
$100-a-month subscription	`reflexion-loop-sub-max`	`dev-team-sub-max`	Max 2 parallel lanes, git worktrees, L0–L3 cross-vendor verify, 60 turn budget.
$20-a-month subscription	`reflexion-loop-sub-pro`	`dev-team-sub-pro`	Single-lane pair (no worktrees), L0–L3 cross-vendor verify, 20 turn budget, capped at M size.

How to decide, the current platform facts, and what the L0–L3 isolation levels mean: Choosing a tier.

Requirements

RTK (Runtime Toolkit): curl -fsSL https://raw.githubusercontent.com/rtk-ai/rtk/refs/heads/master/install.sh | sh
GitHub CLI (gh): Required for automated PR management.
Browsers (Playwright): npx playwright install chromium
Python Deps: pip install python-dotenv playwright
System: Access to your local Chrome User Data Directory.

Firecrawl API: (Optional) For the planning-expert to read external links.

🚀 Quick Start

1. Installation

Clone this repo and link it globally for easy access:

# Recommended: link the repo's own commands once, no hardcoded paths.
# Run this inside the tech-lead-stack checkout:
#   npm link
# That gives you `lead-init`, `lead-clean` and `lead-run` everywhere.

# Alternative: shell aliases, if you would rather not link globally.
# Add these to your ~/.zshrc, replacing the path with your checkout.
alias lead-init='bash /path/to/tech-lead-stack/install.sh --link .'

# Cursor: register skills globally (~/.cursor/skills/) without touching your app repo
alias lead-init-cursor='bash /path/to/tech-lead-stack/install.sh --link . --ide cursor'

# Continue: register skills and MCP globally (~/.continue/config.yaml) without touching your app repo
alias lead-init-continue='bash /path/to/tech-lead-stack/install.sh --link . --ide continue'

# Claude Code: generate /tls:<name> slash commands + user-scope MCP, globally
alias lead-init-claude='bash /path/to/tech-lead-stack/install.sh --link . --ide claude-code'

# Add an IDE to a project you already linked, without re-running the full install
alias lead-ide-only='bash /path/to/tech-lead-stack/install.sh --link . --ide-only --ide'

2. Initialize a Project

Navigate to any repository you want to automate and run the new alias:

lead-init

Next steps

Set up your editor: see Supported Editors & Agents below.
Connect the MCP server directly, through install.sh, or behind SLM Gate: Running the MCP server.
Choose models per role, or run fully offline: Configuration.
Every install flag, and how to remove everything again: Install, Link & Uninstall.

Supported Editors & Agents

Client	Installer flag	Verified	Notes
Antigravity	manual registration	✅ Tested	Workflows registered through Agent Manager. Protobuf state, so not automatable.
Claude Code	`--ide claude-code`	✅ Tested	Generates `/tls:<name>` slash commands plus user-scope MCP. See Claude Code Setup.
Cline	`--ide cline`	✅ Tested	MCP-only. Also reads `AGENTS.md`. See Cline Setup.
Claude Desktop	auto-detected	⚠️ Configured, unverified	MCP-only. Configured when its config file is found.
Continue	`--ide continue`	⚠️ Configured, unverified	Writes `~/.continue/config.yaml` and prompts; not yet smoke-tested end to end.
Cursor	`--ide cursor`	⚠️ Configured, unverified	Writes `~/.cursor/skills/` and `~/.cursor/mcp.json`; not yet smoke-tested end to end.
Gemini CLI / Desktop	`--ide gemini`	⚠️ Configured, unverified	MCP-only. Merges into `~/.gemini/settings.json`. See Gemini Setup.

Every row above is removed by the same command: lead-clean --global --apply.

"Verified" means a real session invoked a skill through that client and the MCP get_skill call succeeded. Anything marked unverified is wired up and expected to work, but has not been confirmed by hand. Reports welcome.

🧹 Resetting a Project

Full detail lives in Install, Link & Uninstall. This is the short version.

Unlink one project. The default. Removes the .ai, .agents and AGENTS.md symlinks and the two copied GitHub files. Your editor setup keeps working for every other project you have linked.

lead-clean                    # the current directory
lead-clean ../other-project   # somewhere else
lead-clean --dry-run          # preview, delete nothing

Remove the editor setup from this machine. Opt-in, and previews unless you add --apply. Covers every platform at once: Claude Code, Cursor, Continue, Cline, Gemini and Claude Desktop.

lead-clean --global           # show exactly what would go
lead-clean --global --apply   # remove it

It deletes only what points at your checkout. An unrelated MCP server in the same config file survives, as do your account and session state. Every JSON file it edits is backed up to <file>.bak first.

Safety. Cleanup refuses to run against your home directory, the filesystem root, or the tech-lead-stack repository itself. A copied file you have since edited, such as a customised pull request template, is kept and reported rather than deleted.

Without npm link, call the script directly:

bash /path/to/tech-lead-stack/scripts/cleanup.sh .

Documentation

Getting set up

Document	What it covers
Choosing a tier	The tier decision guide, current platform facts, and the L0–L3 model isolation levels.
Running the MCP server	The three ways to connect: direct, `install.sh`, or behind SLM Gate. Building the standalone artifact.
Configuration	Model routing per role, and the fully offline local execution tier.
Install, Link & Uninstall	Every installer flag, what each platform gets, and how to remove it all.
Using the stack in other projects	Context injection for web agents, and the symlink route for IDE agents.

Editor setup: Antigravity · Claude Code · Cline · Continue · Cursor · Gemini CLI & Desktop

Reference

Document	What it covers
Available skills	Every skill by lifecycle phase, with its estimated context footprint. Generated from the skill files.
Workflow catalogue	All engineering, product management and HR workflows.
The web app	The hosted dashboard and its routes.
Methodology	The four pillars, skill handoffs, policies, execution targets, analytics and the Reflexion loop.
Architecture	How RTK and the MCP server fit together, and how skills are discovered.
CI/CD	What CI validates, and a fix for the "profile locked" browser error.
Resources	Methodology sources, tooling and editor documentation.

Guides, designs and decisions

Document	Purpose
`docs/IMPLEMENTATION_PLAYBOOK.md`	The definitive guide on implementation.
`docs/using-the-dev-team.md`	Guide to operating the dev-team orchestrator.
`docs/skill-readiness.md`	Status of skill readiness.
`docs/mcp-proxy-setup.md`	Running the stack behind an upstream MCP proxy (slm-gate or any other).
`docs/reflexion-issue-runner.md`	Running reflexion as a GitHub issue loop.
`docs/designs/2026-07-08-agentic-dev-team-design.md`	Design doc for the dev team orchestrator.
`docs/designs/2026-07-08-reflexion-loop-v2-interview-gate.md`	Design doc for the reflexion loop.
`docs/decisions/0001-packaging.md`	ADR 0001: Packaging and Dependencies.
`docs/decisions/0002-lifecycle-paradigm.md`	ADR 0002: 9-Phase Lifecycle Paradigm.
`docs/decisions/0003-execution-targets.md`	ADR 0003: Agent Execution Targets.

Peripherals & Sibling Apps

Voice Relay Service: A local node service that parses spoken transcripts and executes them via keyless agent CLIs (agy, claude, codex, cursor-agent).
Voice Assistant App: A mobile client (iOS/Android) that acts as a hands-free voice interface for the Tech Lead Stack. It connects to the local voice-relay peripheral to execute codebase changes via voice commands.

Branching Strategy

This repository enforces Trunk Based Development with a rebase-first workflow and squash-and-merge PRs.

For detailed day-to-day workflow examples and guidelines for both developers and AI agents, please refer to the Branch Management Strategy document.

Questions or feature requests? Open an issue or join the discussion on GitHub.

Name		Name	Last commit message	Last commit date
.agents		.agents
.ai		.ai
.github		.github
.husky		.husky
.rtk		.rtk
.vscode		.vscode
apps/dashboard		apps/dashboard
artifacts		artifacts
defect-library		defect-library
docs		docs
packages/core		packages/core
public		public
scripts		scripts
templates		templates
tests/fixtures		tests/fixtures
.gitignore		.gitignore
.markdownlint.json		.markdownlint.json
.prettierignore		.prettierignore
.prettierrc		.prettierrc
AGENTS.md		AGENTS.md
BRANCH_MANAGEMENT.md		BRANCH_MANAGEMENT.md
CHANGELOG.md		CHANGELOG.md
CLAUDE.md		CLAUDE.md
CONTRIBUTING.md		CONTRIBUTING.md
LICENSE		LICENSE
ONBOARDING.md		ONBOARDING.md
README.md		README.md
SECURITY.md		SECURITY.md
docker-compose.yml		docker-compose.yml
eslint.config.mjs		eslint.config.mjs
install.sh		install.sh
jest.config.cjs		jest.config.cjs
package.json		package.json
pnpm-lock.yaml		pnpm-lock.yaml
pnpm-workspace.yaml		pnpm-workspace.yaml
requirements.txt		requirements.txt
tsconfig.json		tsconfig.json

Navigation Menu

Repository files navigation

The Lead Stack: Agent-Agnostic Workflows

`intent-brief → spec → plan → diff → review-report → release`

Contents

How it fits together

Web app flow

1. Chat and feature discovery

2. Reflexion from the web app

3. Sandbox preview

4. Writes to your GitHub repo

5. Settings, design review and the skill editor

IDE flow

6. Skills, knowledge items and pipelines

7. Code tools

8. Reflexion from the IDE

Shared by both flows

9. Where telemetry goes

Commands Quick Reference

Which tier am I on?

Requirements

🚀 Quick Start

1. Installation

2. Initialize a Project

Next steps

Supported Editors & Agents

🧹 Resetting a Project

Documentation

Peripherals & Sibling Apps

Branching Strategy

About

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

The Lead Stack: Agent-Agnostic Workflows

intent-brief → spec → plan → diff → review-report → release

Contents

How it fits together

Web app flow

1. Chat and feature discovery

2. Reflexion from the web app

3. Sandbox preview

4. Writes to your GitHub repo

5. Settings, design review and the skill editor

IDE flow

6. Skills, knowledge items and pipelines

7. Code tools

8. Reflexion from the IDE

Shared by both flows

9. Where telemetry goes

Commands Quick Reference

Which tier am I on?

Requirements

🚀 Quick Start

1. Installation

2. Initialize a Project

Next steps

Supported Editors & Agents

🧹 Resetting a Project

Documentation

Peripherals & Sibling Apps

Branching Strategy

About

Topics

Resources

Contributing

Security policy

Stars

Watchers

Forks

Contributors

Languages

`intent-brief → spec → plan → diff → review-report → release`