@tdk-landscape/tdk-cli-core 1.3.44 → 1.3.48

package/openspec/changes/archive/2026-06-25-landscape-consolidation/design.md

@@ -0,0 +1,80 @@
+## Context
+
+### CLI Duplication
+`beauty-crm/cli/` and `tdk-cli/cli/` both contain the `@tdk-landscape/tdk-cli-core` package. The `beauty-crm/cli/package.json` declares `"repository.directory": "cli"` pointing to `tdk-landscape/tdk-cli`, confirming it was originally published from the tdk-cli monorepo workspace. Over time, `tdk-cli/cli/` has evolved (12 src files, generator support, DRY assessment) while `beauty-crm/cli/` has lagged behind (9 src files). Any developer fixing the CLI must choose which copy to edit — and usually picks the wrong one.
+
+### AI Config Sprawl
+The beauty-crm monorepo has accumulated configs for every AI coding assistant that gained popularity:
+
+| Assistant | Type | Size |
+|-----------|------|------|
+| `.claude/` | Dir | skills, CLAUDE.md (757 lines) |
+| `.cursor/` | Dir | rules, mcp.json, Dockerfile |
+| `.continue/` | Dir | skills config |
+| `.roo/` | Dir | 7 rulesets, mcp.json |
+| `.serena/` | Dir | memories, prompt templates |
+| `.agents/` | Dir | skills, mempalace |
+| `.augment/` | Dir | agents, skills |
+| `.kiro/` | Dir | agents, skills, settings, specs |
+| `CLAUDE.md` | File | 757 lines (canonical project context) |
+| `.cursorrules` | File | 32 lines (duplicates CLAUDE.md) |
+| `.windsurfrules` | File | 2855 lines (duplicates CLAUDE.md + more) |
+| `.roomodes` | File | 47 lines (mode definitions) |
+| `.augment-guidelines` | File | 457 lines (duplicates CLAUDE.md) |
+
+Most contain overlapping project context (tech stack, architecture, conventions) which creates a maintenance tax — every time the tech stack or conventions change, 13+ files need updating. The `.cursorrules` is a 32-line subset of `CLAUDE.md`. The `.windsurfrules` is 2855 lines that includes the full `CLAUDE.md` content plus extra.
+
+## Goals / Non-Goals
+
+**Goals:**
+- Remove `beauty-crm/cli/` and replace with a symlink to `tdk-cli/cli/`
+- Update any beauty-crm tooling that references `beauty-crm/cli/` directly
+- Consolidate all AI configs into one canonical location per project
+- Remove duplicate root-level config files (`.cursorrules`, `.windsurfrules`, `.roomodes`, `.augment-guidelines`)
+- Move `.cursor/mcp.json`, `.roo/` mode definitions into a shared format if they add unique value
+- Verify no CI/CD breakage
+
+**Non-Goals:**
+- Merging AI assistant-specific features (e.g., Serena memories, Roo rulesets) — keep them in `.claude/skills/` or let each assistant read its own format
+- Renaming the entire root directory structure
+- Changing how `tdk-cli/` is organized internally
+- Adding or removing any AI assistant support — only consolidating configs
+
+## Decisions
+
+### Decision 1: Remove Completely (No Symlink)
+**Path**: `git rm -r beauty-crm/cli/` — deleted from git and disk, no replacement.
+**Rationale**: A symlink breaks `git add pathspec` resolution (git treats symlinked directories as single blobs and cannot traverse them). Since no scripts, workspaces, or CI pipelines reference `beauty-crm/cli/`, there's nothing to break. Removing entirely is the cleanest solution.
+**Alternatives considered**:
+- Symlink to `../tdk-cli/cli/` — Rejected: breaks `git add` and other git operations
+- Keep as-is (do nothing) — Drift will continue to grow
+
+### Decision 2: Canonical AI Config = `.claude/` + `CLAUDE.md`
+**Rationale**: `CLAUDE.md` is the most comprehensive single file (757 lines covering tech stack, architecture, conventions). `.claude/` is the directory format used by Claude Code and Opencode. Other assistants (Cursor, Windsurf, Continue, Roo, Serena) can read symlinked or imported copies.
+**Implementation**: Other root config files get removed. Each assistant's directory gets a README pointing to `CLAUDE.md` as canonical.
+
+### Decision 3: Preserve `.cursor/mcp.json` as Shared MCP Config
+**Rationale**: MCP server configuration is tool-agnostic and useful across assistants. Move it to a shared location or keep as a thin reference.
+**Approach**: Keep `.cursor/mcp.json` but add a comment header pointing to `CLAUDE.md` for context. Alternatively, move to `.opencode/mcp.json` if opencode is the canonical tool.
+
+### Decision 4: Keep `.roo/` Rulesets
+**Rationale**: Roo's role-based rulesets (architect, code, debug, etc.) are unique to Roo's multi-agent workflow and don't duplicate CLAUDE.md content. They add value.
+**Approach**: Leave `.roo/rules/` in place. Only remove root-level `.roomodes` if Roo can find its modes via the directory.
+
+### Decision 5: No Cross-Project AI Config Sync
+**Rationale**: `identity/`, `platform/`, and `tdk-cli/` don't have AI configs. Adding them would be scope creep. Each project should opt in when ready.
+**Approach**: Document in CONTRIBUTING.md that each project manages its own AI configs, and `.claude/` + `CLAUDE.md` is the recommended canonical format.
+
+## Risks / Trade-offs
+
+**Risk**: Symlink breaks in CI/CD (Docker, git checkout without submodules)
+- **Mitigation**: Symlinks work with `git clone` and Docker COPY. Test explicitly.
+
+**Risk**: Developer has `.cursorrules` or `.windsurfrules` open in muscle memory and is confused when gone
+- **Mitigation**: Leave a one-line comment file: `.cursorrules → see CLAUDE.md`
+
+**Risk**: Cursor/Continue/Roo stop working optimally without their config files
+- **Mitigation**: Each assistant can still use its own directory format. We only remove root-level _files_ that duplicate `CLAUDE.md`. The directories (`.cursor/`, `.continue/`, `.roo/`) stay but get trimmed.
+
+**Risk**: Some AI assistant relies on `.windsurfrules` being 2855 lines (it contains everything)
+- **Mitigation**: WindSurf can also read `CLAUDE.md` directly. The `CLAUDE.md` is already comprehensive.

package/openspec/changes/archive/2026-06-25-landscape-consolidation/proposal.md

@@ -0,0 +1,40 @@
+## Why
+
+The TDK landscape has accumulated two inefficiencies from its extraction from the original `beauty-crm` monorepo:
+
+1. **Duplicated CLI client**: `beauty-crm/cli/` is a vendored copy of `tdk-cli/cli/` (both publish as `@tdk-landscape/tdk-cli-core`). This means every CLI fix must be applied twice, and the two copies have already diverged — `beauty-crm/cli/` has 9 source files while `tdk-cli/cli/` has 12 with additional features. This will only worsen over time.
+
+2. **AI config sprawl**: `beauty-crm/` contains configuration for 8 AI coding assistants (`.claude/`, `.cursor/`, `.continue/`, `.roo/`, `.serena/`, `.agents/`, `.augment/`, `.kiro/`) plus 5 root-level config files (`CLAUDE.md`, `.cursorrules`, `.windsurfrules`, `.roomodes`, `.augment-guidelines`). Most contain duplicate project context. Developers waste time maintaining 13+ config files that say the same thing.
+
+Fixing both reduces maintenance burden, clarifies the canonical source of truth, and prevents future drift.
+
+## What Changes
+
+### Workstream A: CLI Duplicate Removal
+- Remove `beauty-crm/cli/` entirely via `git rm`
+- No symlink — symlinks break `git add` pathspec resolution
+- Update any `beauty-crm/` scripts/build pipelines that reference the old path
+
+### Workstream B: AI Config Consolidation
+- Choose one canonical AI assistant config per project (recommend: `.claude/` + `CLAUDE.md`)
+- Remove duplicated root config files (`.cursorrules`, `.windsurfrules`, `.roomodes`, `.augment-guidelines`)
+- Consolidate `beauty-crm/.kiro/`, `beauty-crm/.agents/`, `beauty-crm/.augment/` into `beauty-crm/.claude/` or a shared `.ai-tools/` directory
+- Add `.cursor/mcp.json` as a workspace-level config (not project-specific)
+- Document which AI config is canonical for each project
+
+## Capabilities
+
+### New Capabilities
+- `cli-source-consolidation`: Single source of truth for `@tdk-landscape/tdk-cli-core`
+- `ai-config-consolidation`: Unified AI assistant configuration
+
+### Modified Capabilities
+- Existing specs in `tdk-cli/openspec/specs/` may need minor updates if they reference `beauty-crm/cli/`
+
+## Impact
+
+- **Files removed**: beauty-crm/cli/ (~50 files), AI root configs (5 files), AI assistant dirs (5+ dirs)
+- **Files added**: 1 symlink, 1-2 consolidated config files
+- **Dependencies**: CI/CD pipelines that reference beauty-crm/cli/ need updating
+- **Breaking Changes**: Any tooling that directly references `beauty-crm/cli/bin/tdk.js` will need to use `tdk-cli/cli/bin/tdk.js`
+- **Risk Level**: Low-Medium. CLI removal is mechanical and testable. AI config changes are cosmetic but need care not to break developer workflows.

package/openspec/changes/archive/2026-06-25-landscape-consolidation/spec.md

@@ -0,0 +1,52 @@
+# landscape-consolidation Specification
+
+## Purpose
+
+The TDK landscape has two sources of maintenance waste: a vendored duplicate of the `@tdk-landscape/tdk-cli-core` package in `beauty-crm/cli/` that has diverged from the canonical `tdk-cli/cli/`, and 13+ AI assistant config entries in `beauty-crm/` with overlapping project context spread across 8 directories and 5 root files. Consolidating both reduces maintenance burden and establishes canonical sources of truth.
+
+## Requirements
+
+### Requirement: Single CLI source of truth
+
+All source code and tooling for the `@tdk-landscape/tdk-cli-core` package SHALL originate from `tdk-cli/cli/`. The `beauty-crm/cli/` directory SHALL NOT contain its own copy of CLI source.
+
+#### Scenario: Remove duplicate CLI directory
+- **WHEN** inspecting the `beauty-crm/` project root
+- **THEN** there SHALL be no `cli/` directory containing standalone `@tdk-landscape/tdk-cli-core` source
+
+#### Scenario: Directory removed completely (no symlink)
+- **WHEN** inspecting the `beauty-crm/` project root
+- **THEN** `beauty-crm/cli/` SHALL NOT exist as a directory or symlink
+
+#### Scenario: No broken references after removal
+- **WHEN** searching `beauty-crm/` for direct `beauty-crm/cli/` path references
+- **THEN** no such references SHALL exist in scripts, configs, or documentation
+
+### Requirement: Canonical AI project context
+
+All AI assistants operating in `beauty-crm/` SHALL use `CLAUDE.md` as the single canonical source for project context (tech stack, architecture, conventions).
+
+#### Scenario: Remove redundant root-level config files
+- **WHEN** listing root-level config files in `beauty-crm/`
+- **THEN** `.cursorrules`, `.windsurfrules`, and `.augment-guidelines` SHALL NOT exist
+
+#### Scenario: Assistant directory structure preserved
+- **WHEN** an AI assistant reads its configuration
+- **THEN** its native directory SHALL still exist (`.cursor/`, `.continue/`, `.roo/`, `.serena/`, `.agents/`, `.augment/`, `.kiro/`)
+- **AND** SHALL NOT have been moved or reformatted
+
+#### Scenario: Cross-reference from assistant configs
+- **WHEN** an AI assistant initializes
+- **THEN** its config files SHALL contain a comment or README referencing `CLAUDE.md` for project context
+
+### Requirement: No regression in CLI functionality
+
+Removing `beauty-crm/cli/` SHALL NOT break any existing functionality that depends on the `@tdk-landscape/tdk-cli-core` package.
+
+#### Scenario: No workspace resolution errors after removal
+- **WHEN** running `bun install` in `beauty-crm/`
+- **THEN** it SHALL NOT error due to the removed `cli/` directory
+
+#### Scenario: No workspace resolution errors
+- **WHEN** running `bun install` in `beauty-crm/`
+- **THEN** it SHALL NOT error due to the removed `cli/` directory

package/openspec/changes/archive/2026-06-25-landscape-consolidation/tasks.md

@@ -0,0 +1,51 @@
+## 1. CLI Duplicate — Audit & Verify
+
+- [x] 1.1 Confirm `beauty-crm/cli/` is not listed in `beauty-crm/package.json` workspaces
+- [x] 1.2 Confirm no CI/CD or scripts reference `beauty-crm/cli/` as a build path
+- [x] 1.3 Verify `beauty-crm/cli/package.json` name is `@tdk-landscape/tdk-cli-core` (not `@beauty-crm/cli`)
+- [x] 1.4 Verify `tdk-cli/cli/` has superset of files vs `beauty-crm/cli/` (72 vs 26 source files)
+- [x] 1.5 Check `beauty-crm/bun.lock` does NOT lock `@tdk-landscape/tdk-cli-core` to a `file:cli` path
+- [x] 1.6 Check `.tilt-engine/` and `Tiltfile` for any references to `beauty-crm/cli/`
+
+## 2. CLI Duplicate — Remove
+
+- [x] 2.1 `git rm -r beauty-crm/cli/` (removed from git and disk)
+- [x] 2.2 Verify no stale references: `rg` found zero references
+
+## 3. AI Config — Audit Current State
+
+- [x] 3.1 Catalog all AI assistant configs in `beauty-crm/`:
+  - Directories: `.claude/`, `.cursor/`, `.continue/`, `.roo/`, `.serena/`, `.agents/`, `.augment/`, `.kiro/`
+  - Root files: `CLAUDE.md`, `.cursorrules`, `.windsurfrules`, `.roomodes`, `.augment-guidelines`
+- [x] 3.2 Identify content overlap:
+  - `.cursorrules` — 32-line communication style (no project context)
+  - `.windsurfrules` — 2855-line superset of CLAUDE.md
+  - `.augment-guidelines` — 457-line partially overlapping
+- [x] 3.3 Check `.roo/rules/` — unique workflow instructions (dev_workflow, self_improve, taskmaster)
+- [x] 3.4 Check `.cursor/mcp.json` — unique MCP server configs (task-master-ai, tilt)
+
+## 4. AI Config — Consolidate
+
+- [x] 4.1 Remove `.cursorrules` (redundant with `CLAUDE.md`)
+- [x] 4.2 Remove `.windsurfrules` (redundant with `CLAUDE.md`)
+- [x] 4.3 Remove `.augment-guidelines` (redundant with `CLAUDE.md`)
+- [x] 4.4 Keep `.roomodes` (defines Roo's custom mode slugs — not redundant with CLAUDE.md)
+- [x] 4.5 Keep all AI assistant directories in place (each reads from its own native path):
+  - `.cursor/`, `.continue/`, `.roo/`, `.serena/`, `.agents/`, `.augment/`, `.kiro/`
+- [x] 4.6 Add comment/README to each assistant config pointing to `CLAUDE.md`:
+  - `.cursor/mcp.json` → `_comment` field
+  - `.roo/README.md`, `.serena/README.md`, `.agents/README.md`, `.augment/README.md`, `.kiro/README.md`, `.continue/README.md`
+
+## 5. AI Config — Documentation
+
+- [x] 5.1 Add banner to `CLAUDE.md` declaring it canonical source of truth
+- [x] 5.2 Keep `.cursor/rules/` in place (native Cursor format)
+- [x] 5.3 Verify `identity/`, `platform/`, `tdk-cli/` don't need AI configs (scope boundary)
+
+## 6. Verification
+
+- [x] 6.1 Run `git status` — clean, 44 files changed
+- [x] 6.2 N/A (no symlink — removed directly via `git rm`)
+- [x] 6.3 No broken references found
+- [x] 6.4 CLAUDE.md is valid markdown, loadable by any AI
+- [ ] 6.5 Ask team to test their preferred AI assistant still works after consolidation

package/openspec/changes/archive/2026-09-17-fix-generated-stack-boot/.openspec.yaml

@@ -0,0 +1,2 @@
+schema: spec-driven
+created: 2026-09-17

package/openspec/changes/archive/2026-09-17-fix-generated-stack-boot/design.md

@@ -0,0 +1,43 @@
+# Fix Generated Stack Boot - Design
+
+## Context
+
+`tdk project` and `tdk resource` produce output that must run on a fresh machine with Docker + Tilt. Two defects broke this path for every user: (1) the backend source template emitted a literal newline inside a single-quoted `console.log` string, making every generated service uncompilable; (2) the generated Tiltfile pointed at an author-specific absolute path (`/private/var/www/2025/ollamar1/tdk-cli`) and fell back to a private GitHub repo that cannot be cloned, causing `tdk up` to hang in a git retry loop. The fixes below were implemented and verified in an isolated container (fixed binary rebuilt from source, `bun run` + `curl /health` green, `tilt ci` loads the vendored extension and reports a healthy workspace).
+
+## Goals / Non-Goals
+
+**Goals:**
+- Generated services compile and serve `/health` on first run.
+- `tdk up` loads its Tiltfile on any machine with zero network or private-repo dependency.
+- Generated output contains no machine-specific paths and fails fast with actionable errors when misconfigured.
+- Extension loading uses a truthful local probe.
+
+**Non-Goals:**
+- Rewriting the TDK Tilt extension itself or its orchestration semantics.
+- Making the private `tdk-cli` source repo public.
+- Changing the generated `service.json` manifest shape (tracked separately in manifest-field-migration).
+- Container-to-host DNS routing (traefik `*.beauty-crm.localhost` mapping) is outside this change.
+
+## Decisions
+
+### D1. Do not hardcode or publish; vendor instead
+Vendor the extension into each generated project (`.tdk/.tdk-out/tdk-cli-ext/`) at generation time. Rationale: matches the "releases public, source private" position, works offline, and a vendored project stays reviewable and relocatable. The source is located via, in order: `$TDK_EXTENSION_SOURCE`, an executable-adjacent `tdk-cli` checkout, then a `../tdk-cli` sibling of the project root. Missing source logs a warning; the Tiltfile then fails fast with recovery instructions.
+
+### D2. Escape newlines at the template boundary
+In `cli/src/commands/resource.ts`, `getBackendIndexTemplate` emits `console.log('\n🚀 ...')`. The template is a JS template literal, so `\n` becomes a real newline in the emitted file. The source therefore uses `\\n` so generated files contain the two-character escape sequence and remain valid Bun source.
+
+### D3. Fail fast instead of a private-repo fallback
+The Tiltfile resolution order is: vendored path -> `$TDK_EXTENSION_PATH` -> `fail(...)` with the three recovery options. This eliminates the infinite `git clone` retry against `github.com/tdk-landscape/tdk-cli` (private, 404 to unauthenticated clients).
+
+### D4. Truthful local probe
+The presence probe must produce output: `test -d <path> && echo yes || echo no`. A probe returning `""` (as the first iteration did) silently disables detection and routes every run to the fail-fast branch.
+
+### D5. Vendoring lives in `generateMasterConfigs`
+The copy step runs in `generateMasterConfigs` (alongside writing `Tiltfile`, `spec.master`, etc.) so regeneration repairs a stale or missing vendored extension.
+
+## Risks / Trade-offs
+
+- **Project size**: vendoring adds ~2.5 MB of starlark per project. Accepted for offline, private-repo-free bootstrapping; the copied set is limited to `Tiltfile`, `engine/`, `discovery/`, `specs/`, `ext/`.
+- **Stale binary releases**: released `release-dist-*` binaries still contain both bugs. Release artifacts must be rebuilt (proven command: `bun build --compile --target=bun-linux-arm64 --outfile tdk-linux-arm64 cli/src/cli.ts`) and republished before users see the fix.
+- **Probe coupling**: detection depends on `local()` emitting the echo; documented in the template comment so future edits keep the `&& echo yes || echo no` tail.
+- **Sandbox verification ceiling**: full container orchestration (image builds via traefik network, `*.beauty-crm.localhost` DNS) requires a host network; the isolated-container verification covers generation, compilation, Tiltfile loading, service discovery, resource creation, and workspace health.

package/openspec/changes/archive/2026-09-17-fix-generated-stack-boot/proposal.md

@@ -0,0 +1,29 @@
+## Why
+
+Generated projects from `tdk project` and `tdk resource` shipped two blockers that broke the quickstart for every user: generated Bun services did not compile (the backend template emits a literal newline inside a `console.log('...')` string, producing `Unterminated string literal`), and `tdk up` could never load its Tiltfile because it hardcoded a machine-specific extension path and fell back to a private GitHub repo that triggers an endless git-clone retry loop. Both were reproduced in an isolated Docker container and verified fixed in source, but the fixes are not yet reflected in released binaries or a tracked spec.
+
+## What Changes
+
+- `cli/src/commands/resource.ts`: escape newlines in the generated backend `src/index.ts` so emitted files contain `\n` and compile (`console.log('\n...')`).
+- `cli/templates/Tiltfile.hbs`: resolve the TDK Tilt extension in order - vendored in-project copy (`.tdk/.tdk-out/tdk-cli-ext`) -> `$TDK_EXTENSION_PATH` -> fail fast with actionable instructions. Remove the hardcoded `/private/var/www/2025/ollamar1/tdk-cli` path and the private-repo fallback.
+- `cli/src/generator/template-engine.ts`: vendor the extension (engine/discovery/specs/ext) into `.tdk/.tdk-out/tdk-cli-ext/` during `generateMasterConfigs`, so `tdk up` works offline on any machine without GitHub access. Source is auto-detected via `$TDK_EXTENSION_SOURCE`, executable-adjacent checkout, or a `../tdk-cli` sibling.
+- `cli/templates/Tiltfile.hbs`: fix the local-directory probe to echo its result (`test -d <path> && echo yes || echo no`); the previous probe returned `""`, so the vendored extension was never detected and `tdk up` always failed at the new fail-fast gate.
+- Rebuild release binaries from the fixed source and republish to `tdk-landscape/tdk-cli-releases` (**BREAKING** for existing releases - new generated output shape).
+
+## Capabilities
+
+### New Capabilities
+
+- `self-contained-generated-projects`: generated projects must boot without network or private-repo access - extension vendored at generation time, Tiltfile loads it from a local `file://` repo, and generated service sources compile on first `bun run`.
+
+### Modified Capabilities
+
+- `manifest-field-migration`: no requirement changes; generated `service.json` shape is unchanged.
+
+## Impact
+
+- `cli/src/commands/resource.ts` - backend index template escaping.
+- `cli/templates/Tiltfile.hbs` - extension resolution and probe.
+- `cli/src/generator/template-engine.ts` - vendoring step, `.tdk/.tdk-out/tdk-cli-ext/` output.
+- `release-dist-*` snapshots - must be regenerated from fixed source.
+- Verification: generated services compile and serve `/health` via Bun; `tilt ci` on a generated Tiltfile loads the vendored extension, discovers services, creates resources, and reports a healthy workspace.

package/openspec/changes/archive/2026-09-17-fix-generated-stack-boot/specs/self-contained-generated-projects/spec.md

@@ -0,0 +1,67 @@
+# Self-Contained Generated Projects
+
+## ADDED Requirements
+
+### Requirement: Generated Services Compile On First Run
+
+Generated backend services must always emit valid Bun source code.
+
+#### Scenario: Backend template with banner logs
+
+- **WHEN** a user runs `tdk resource api --type backend --stack pre-alpha --path services/pre-alpha/api`
+- **THEN** the generated `src/index.ts` contains `console.log('\n🚀 api running ...')` with an escaped `\n`
+- **AND** `bun run src/index.ts` starts the server without a syntax error
+
+#### Scenario: Health endpoint serves
+
+- **WHEN** the generated service is running
+- **THEN** `GET /health` responds `{"status":"ok","service":"<name>"}`
+
+### Requirement: Extension Vendored Into Generated Projects
+
+Every generated project must contain the TDK Tilt extension so `tdk up` works offline.
+
+#### Scenario: Project generation
+
+- **WHEN** a user runs `tdk project --yes`
+- **THEN** `.tdk/.tdk-out/tdk-cli-ext/` is created containing `Tiltfile`, `engine/`, `discovery/`, `specs/`, and `ext/`
+- **AND** the generation log reports `Vendored TDK extension -> .tdk/.tdk-out/tdk-cli-ext/`
+
+#### Scenario: Extension source auto-detection
+
+- **WHEN** generating a project
+- **THEN** the extension source is resolved in order: `$TDK_EXTENSION_SOURCE`, an executable-adjacent `tdk-cli` checkout, then a `../tdk-cli` sibling of the project root
+
+### Requirement: Tiltfile Loads Extension Locally
+
+The generated Tiltfile must resolve the extension without any GitHub dependency.
+
+#### Scenario: Vendored copy present
+
+- **WHEN** `.tdk/.tdk-out/tdk-cli-ext/` exists next to the Tiltfile
+- **THEN** the Tiltfile registers `file://<vendored path>` as the extension repo
+- **AND** `load('ext://tdk-cli', ...)` succeeds without cloning
+
+#### Scenario: Explicit extension path override
+
+- **WHEN** the vendored copy is absent but `$TDK_EXTENSION_PATH` points at a valid tdk-cli checkout
+- **THEN** the Tiltfile uses that path
+
+#### Scenario: No extension available
+
+- **WHEN** neither the vendored copy nor `$TDK_EXTENSION_PATH` exists
+- **THEN** loading fails fast with a single actionable error listing the three recovery options, with no retry loop
+
+### Requirement: No Machine-Specific Paths In Generated Output
+
+Generated Tiltfiles must never contain hardcoded author or machine paths.
+
+#### Scenario: Fresh generation on any machine
+
+- **WHEN** any user generates a project
+- **THEN** the generated `Tiltfile` has no absolute host-specific paths and no private GitHub repo URL fallback
+
+#### Scenario: Local probe is truthful
+
+- **WHEN** the Tiltfile probes for the vendored directory
+- **THEN** the probe echoes its outcome (`test -d <path> && echo yes || echo no`) so detection reflects the filesystem

package/openspec/changes/archive/2026-09-17-fix-generated-stack-boot/tasks.md

@@ -0,0 +1,27 @@
+## 1. Compilable Generated Backend Source
+
+- [x] 1.1 Update `getBackendIndexTemplate` in `cli/src/commands/resource.ts` so the emitted banner lines use `\\n` (escaped backslash-n): `console.log('\n...')` becomes two characters in generated files
+- [x] 1.2 Generate a backend service with the CLI and assert `src/index.ts` has no raw newline inside the `console.log` string
+- [x] 1.3 Boot the generated service with `bun run src/index.ts` and assert `GET /health` returns `{"status":"ok","service":"<name>"}`
+
+## 2. Extension Vendored Into Generated Projects
+
+- [x] 2.1 Add `vendorTdkExtension(projectRoot)` to `cli/src/generator/template-engine.ts`, called from `generateMasterConfigs`, copying `Tiltfile`, `engine/`, `discovery/`, `specs/`, `ext/` into `.tdk/.tdk-out/tdk-cli-ext/`
+- [x] 2.2 Resolve the extension source in order: `$TDK_EXTENSION_SOURCE` -> executable-adjacent `tdk-cli` checkout -> `../tdk-cli` sibling of the project root
+- [x] 2.3 Print a warning (no hard failure) when no extension source is found
+- [x] 2.4 Assert generation reports `Vendored TDK extension -> .tdk/.tdk-out/tdk-cli-ext/` and the directory exists
+
+## 3. Tiltfile Extension Resolution
+
+- [x] 3.1 Rewrite `cli/templates/Tiltfile.hbs`: resolve vendored path -> `$TDK_EXTENSION_PATH` -> `fail(...)` with the three recovery options; remove the hardcoded `/private/var/www/2025/ollamar1/tdk-cli` path and the private GitHub fallback
+- [x] 3.2 Use truthful probes: `test -d <path> && echo yes || echo no` in both branches
+- [x] 3.3 Assert a generated Tiltfile contains no absolute host-specific paths and no `github.com/tdk-landscape/tdk-cli` URL
+- [x] 3.4 Run `tilt ci -f .tdk/.tdk-out/Tiltfile` on a generated project and assert the vendored extension loads, services are discovered, and the workspace reports healthy
+- [x] 3.5 Run `tdk up pre-alpha` on a two-service project and assert no git-clone retry loop; Tiltfile loading succeeds
+
+## 4. Release Binaries
+
+- [x] 4.1 Rebuild binaries from fixed source: `bun build --compile --target=bun-linux-amd64 --outfile tdk-linux-amd64 cli/src/cli.ts` (repeat for `bun-linux-arm64`, `bun-darwin-amd64`, `bun-darwin-arm64`)
+- [x] 4.2 Verify each rebuilt binary: `--version` matches, `project --yes` vendors the extension, and a generated service compiles and boots
+- [x] 4.3 Update `release-dist-*` snapshots, `checksums.txt`, and the binaries zip
+- [x] 4.4 Publish binaries to `tdk-landscape/tdk-cli-releases` releases with the fixed snapshots

package/openspec/specs/manifest-field-migration/spec.md

@@ -0,0 +1,24 @@
+# manifest-field-migration Specification
+
+## Purpose
+TBD - created by archiving change rename-domain-to-stack-service-to-resource. Update Purpose after archive.
+## Requirements
+### Requirement: Manifest field migration from domain to stack
+All manifest field access patterns SHALL remove fallback support for the deprecated `domain` field.
+
+#### Scenario: No domain field fallback
+- **WHEN** code retrieves the stack field from a manifest
+- **THEN** it SHALL NOT use fallback patterns like `manifest.get('stack') or manifest.get('domain')`
+
+#### Scenario: Stack field with default only
+- **WHEN** a default value is needed for the stack field
+- **THEN** code SHALL use `manifest.get('stack', default_value)` and NOT check for domain
+
+#### Scenario: Domain field references removed
+- **WHEN** searching the codebase for domain field access
+- **THEN** no `manifest.get('domain')` patterns SHALL exist except in legacy parser code
+
+#### Scenario: Filtering uses stack field only
+- **WHEN** filtering manifests by logical group
+- **THEN** code SHALL filter by `stack` field only and NOT check `domain` as fallback
+

package/openspec/specs/resource-terminology-standard/spec.md

@@ -0,0 +1,24 @@
+# resource-terminology-standard Specification
+
+## Purpose
+TBD - created by archiving change rename-domain-to-stack-service-to-resource. Update Purpose after archive.
+## Requirements
+### Requirement: Resource terminology standardization
+All references to deployable units SHALL use the term "resource" instead of "service".
+
+#### Scenario: Function parameters use resource terminology
+- **WHEN** a function accepts a deployable unit as a parameter
+- **THEN** the parameter name SHALL be `resource` or `resource_dict` and NOT `service` or `svc`
+
+#### Scenario: Docstrings use resource terminology
+- **WHEN** a function docstring describes a deployable unit
+- **THEN** it SHALL use the term "resource" and NOT "service"
+
+#### Scenario: Variable names use resource terminology
+- **WHEN** a variable holds a deployable unit
+- **THEN** the variable name SHALL be `resource` and NOT `svc` or `service`
+
+#### Scenario: Comments use resource terminology
+- **WHEN** code comments describe deployable units
+- **THEN** they SHALL use the term "resource" and NOT "service"
+

package/openspec/specs/self-contained-generated-projects/spec.md

@@ -0,0 +1,67 @@
+# Self-Contained Generated Projects
+
+## ADDED Requirements
+
+### Requirement: Generated Services Compile On First Run
+
+Generated backend services must always emit valid Bun source code.
+
+#### Scenario: Backend template with banner logs
+
+- **WHEN** a user runs `tdk resource api --type backend --stack pre-alpha --path services/pre-alpha/api`
+- **THEN** the generated `src/index.ts` contains `console.log('\n🚀 api running ...')` with an escaped `\n`
+- **AND** `bun run src/index.ts` starts the server without a syntax error
+
+#### Scenario: Health endpoint serves
+
+- **WHEN** the generated service is running
+- **THEN** `GET /health` responds `{"status":"ok","service":"<name>"}`
+
+### Requirement: Extension Vendored Into Generated Projects
+
+Every generated project must contain the TDK Tilt extension so `tdk up` works offline.
+
+#### Scenario: Project generation
+
+- **WHEN** a user runs `tdk project --yes`
+- **THEN** `.tdk/.tdk-out/tdk-cli-ext/` is created containing `Tiltfile`, `engine/`, `discovery/`, `specs/`, and `ext/`
+- **AND** the generation log reports `Vendored TDK extension -> .tdk/.tdk-out/tdk-cli-ext/`
+
+#### Scenario: Extension source auto-detection
+
+- **WHEN** generating a project
+- **THEN** the extension source is resolved in order: `$TDK_EXTENSION_SOURCE`, an executable-adjacent `tdk-cli` checkout, then a `../tdk-cli` sibling of the project root
+
+### Requirement: Tiltfile Loads Extension Locally
+
+The generated Tiltfile must resolve the extension without any GitHub dependency.
+
+#### Scenario: Vendored copy present
+
+- **WHEN** `.tdk/.tdk-out/tdk-cli-ext/` exists next to the Tiltfile
+- **THEN** the Tiltfile registers `file://<vendored path>` as the extension repo
+- **AND** `load('ext://tdk-cli', ...)` succeeds without cloning
+
+#### Scenario: Explicit extension path override
+
+- **WHEN** the vendored copy is absent but `$TDK_EXTENSION_PATH` points at a valid tdk-cli checkout
+- **THEN** the Tiltfile uses that path
+
+#### Scenario: No extension available
+
+- **WHEN** neither the vendored copy nor `$TDK_EXTENSION_PATH` exists
+- **THEN** loading fails fast with a single actionable error listing the three recovery options, with no retry loop
+
+### Requirement: No Machine-Specific Paths In Generated Output
+
+Generated Tiltfiles must never contain hardcoded author or machine paths.
+
+#### Scenario: Fresh generation on any machine
+
+- **WHEN** any user generates a project
+- **THEN** the generated `Tiltfile` has no absolute host-specific paths and no private GitHub repo URL fallback
+
+#### Scenario: Local probe is truthful
+
+- **WHEN** the Tiltfile probes for the vendored directory
+- **THEN** the probe echoes its outcome (`test -d <path> && echo yes || echo no`) so detection reflects the filesystem

package/openspec/specs/stack-terminology-standard/spec.md

@@ -0,0 +1,24 @@
+# stack-terminology-standard Specification
+
+## Purpose
+TBD - created by archiving change rename-domain-to-stack-service-to-resource. Update Purpose after archive.
+## Requirements
+### Requirement: Stack terminology standardization
+All references to logical grouping of resources SHALL use the term "stack" instead of "domain".
+
+#### Scenario: Manifest field access uses stack
+- **WHEN** code accesses the grouping field from a manifest
+- **THEN** it SHALL use `manifest.get('stack')` and NOT `manifest.get('domain')`
+
+#### Scenario: Cache keys use stack terminology
+- **WHEN** the discovery cache stores configuration for logical groups
+- **THEN** the cache key SHALL be named `stack_configs` and NOT `domain_configs`
+
+#### Scenario: Autogenerated files use stack naming
+- **WHEN** autogenerated index files are created for logical groups
+- **THEN** the filenames SHALL contain `stacks` and NOT `domains`
+
+#### Scenario: Function names use stack terminology
+- **WHEN** a function generates an index of logical groups
+- **THEN** the function name SHALL include `stacks` and NOT `domains`
+