claudeos-core 2.0.2 → 2.1.1

package/pass-prompts/templates/common/pass3-phase1.md

@@ -0,0 +1,131 @@
+## PHASE 1 — Read Once, Extract Facts (DO THIS FIRST, EXACTLY ONCE)
+
+Before generating ANY file, read the following inputs EXACTLY ONCE:
+1. `claudeos-core/generated/project-analysis.json` — structured stack & domain data
+2. `claudeos-core/generated/pass2-merged.json` — deep analysis merged report
+3. `claudeos-core/generated/pass3-context.json` (if it exists) — pre-computed slim summary; prefer this when available
+
+Then extract the following facts into your working memory. You will reference
+this table for EVERY subsequent file generation. Fill in concrete values
+(not placeholders) wherever pass2-merged.json provides them; use `(not in analysis — see source)` otherwise.
+
+| Fact | Value |
+|---|---|
+| Language + Version | ? |
+| Framework + Version | ? |
+| Build tool | ? |
+| Package manager | ? |
+| Database / ORM | ? |
+| Frontend (if any) | ? |
+| Architecture flags (CQRS / BFF / DDD / multi-module) | ? |
+| Module list (if multi-module) | ? |
+| **Response wrapper LAYER** (Controller / Aggregator / Service) | ? |
+| **Response wrapper METHOD** (exact name, e.g. `makeResponse`, `ResponseEntity.ok`) | ? |
+| **Response util class** (exact FQN, e.g. `com.company.util.ApiResponseUtil`) | ? |
+| Exact method names on response util (e.g. `success`, `ok`, `fail` — verbatim) | ? |
+| MyBatis mapper XML path (verbatim, if MyBatis detected) | ? |
+| Aggregator/Orchestrator layer exists? | ? |
+| Aggregator naming convention (e.g. `{Domain}Aggregator`) | ? |
+| Detected domains (flat list) | ? |
+| Per-domain key API methods (terse — 3-5 per domain max) | ? |
+| Base/shared classes (FQN, must not be re-declared) | ? |
+| Shared util package root (e.g. `com.company.shared.util`) | ? |
+| Authentication method (JWT / Session / OAuth2 / etc.) | ? |
+
+## PHASE 2 — Generate Files (using the fact table above)
+
+Proceed to generate files based on the instructions below.
+
+### Rule A — Reference, don't re-read
+
+Once the fact table is filled in PHASE 1, **DO NOT re-read `pass2-merged.json`
+or `pass3-context.json`**. All cross-file consistency checks must use the
+fact table as the authoritative source.
+
+If you find yourself about to `Read pass2-merged.json` again during Phase 2:
+STOP. Reference the fact table. If a detail is missing from the fact table,
+add it to the table ONCE (by reading the relevant JSON file ONCE), then
+reference the table from then on.
+
+**Why this matters**: pass2-merged.json on large multi-module projects can be
+50-500 KB. Re-reading it 10-20 times while generating 30+ files is the #1
+cause of `Prompt is too long` failures in Pass 3. The fact table above is
+your compact in-context cache.
+
+### Rule B — Idempotent file writing (resume-safe)
+
+Before writing ANY target file, check whether it already exists:
+
+1. Use the Read tool on the target path.
+2. If the file EXISTS and has more than 100 characters of real content:
+   - Print `[SKIP] <path> — already generated`
+   - Proceed to the next target WITHOUT regenerating
+   - DO NOT re-read analysis JSON files for a skipped target
+3. If the file is MISSING or has 100 characters or fewer, generate it fresh.
+
+For staged rules (`.claude/rules/*` → `claudeos-core/generated/.staged-rules/*`):
+- Check BOTH the staging path (`claudeos-core/generated/.staged-rules/<subpath>`)
+  AND the final path (`.claude/rules/<subpath>`).
+- If EITHER already has content, skip.
+
+This makes Pass 3 safely re-runnable after any interruption — a crash or
+timeout at file 31 of 50 does not mean restarting from scratch.
+
+### Rule C — Cross-file consistency via the fact table
+
+When generating any file that references another file's contents (e.g.
+`architecture.md` referencing `controller-patterns.md`), use ONLY the values
+from the fact table. Never invent method names, class names, or signatures
+not present in the table.
+
+If the fact table says `Response wrapper LAYER = Controller`, then
+`architecture.md`, `controller-patterns.md`, `response-exception.md`,
+`controller-rules.md`, and ALL skills files MUST show Controller wrapping
+the response. Inconsistencies here are the #1 quality complaint in Pass 3
+output — the fact table exists specifically to prevent them.
+
+### Rule D — Output conciseness (prevents context overflow mid-generation)
+
+Pass 3 generates 30-50+ files in one session. Every token of narration between
+file writes accumulates in the conversation context and eventually causes
+`Prompt is too long` failures around file 30-40. To prevent this:
+
+1. **Between file writes, output AT MOST one short line** — either
+   `[WRITE] <path>` or `[SKIP] <path> — already exists`. No paragraphs,
+   no "Now I will generate...", no "This file contains...".
+2. **Do NOT restate the fact table** between files. It is already in your
+   in-context memory from PHASE 1.
+3. **Do NOT re-explain design decisions** between files. If a decision was
+   recorded in the fact table or in an earlier file you wrote this session,
+   reference it by name, do not re-derive it.
+4. **Each file's content itself should be focused** — write what the project
+   needs, not exhaustive commentary. If a markdown file would exceed ~300 lines,
+   split it by creating additional files with numbered prefixes rather than
+   producing one giant file.
+5. **Do NOT echo file content back in your response after writing.** The Write
+   tool confirms success; there is no need to say "I have written the following
+   content: ...".
+
+The failure mode to avoid: writing file 25, then explaining what you just wrote
+in 200 words, then writing file 26, then explaining again. After 40 files that
+verbose pattern alone adds 15-30K tokens of pure narration — enough to overflow
+context on top of the actual file contents.
+
+### Rule E — Batch idempotent checks at the start
+
+Instead of calling `Read` on every target file individually right before
+writing it (which doubles the number of tool calls and accumulates Read
+response content in context), do this ONCE at the start of PHASE 2:
+
+1. Use `Glob` on each output root (`claudeos-core/standard/**/*.md`,
+   `.claude/rules/**/*.md`, etc.) to get a single list of already-existing files.
+2. Keep that list in your working memory alongside the fact table.
+3. For each target file, check the list (not the filesystem) — if present and
+   non-trivially sized, emit `[SKIP]` and move on.
+4. Only call `Read` on a specific target file if you have a concrete reason
+   to inspect its content.
+
+Rule E makes Rule B idempotency much cheaper in tool call count and context.
+
+---
+

package/pass-prompts/templates/common/pass3a-facts.md

@@ -0,0 +1,143 @@
+## Pass 3a — FACT EXTRACTION (no file generation yet)
+
+This is step 1 of 4 in Pass 3 split mode. Your ONLY task here is to read the
+analysis JSON files and write a single `pass3a-facts.md` summary document.
+**Do NOT generate CLAUDE.md, standard/, rules/, skills/, guide/, or plan/ in
+this step.** Those come in later steps.
+
+## Inputs (read each at most ONCE)
+
+1. `claudeos-core/generated/pass3-context.json` — slim summary (start here)
+2. `claudeos-core/generated/project-analysis.json` — structured stack/domain data
+3. `claudeos-core/generated/pass2-merged.json` — full deep-analysis report
+
+## Output
+
+Write exactly one file: `claudeos-core/generated/pass3a-facts.md`
+
+This file is the authoritative fact sheet that Pass 3b, 3c, and 3d will
+reference. It must be complete enough that the later steps NEVER need to
+re-open pass2-merged.json.
+
+Required structure:
+
+```markdown
+# Pass 3 Fact Sheet
+
+Generated at: <ISO timestamp>
+
+## Stack
+
+- Language: ...
+- Language version: ...
+- Framework: ...
+- Framework version: ...
+- Build tool: ...
+- Package manager: ...
+- Database: ...
+- ORM: ...
+- Frontend: ...
+- Default port: ...
+
+## Architecture
+
+- CQRS: yes/no
+- BFF: yes/no
+- DDD (Aggregate Root): yes/no
+- Multi-module: yes/no (if yes, list module names)
+- Monorepo tool: ...
+- Root package: ...
+
+## Response Flow (CRITICAL — used for cross-file consistency)
+
+- Wrapper layer: <Controller | Aggregator | Service>
+- Wrapper method: <exact method name, e.g. `makeResponse`>
+- Response util class: <FQN, e.g. `com.company.util.ApiResponseUtil`>
+- Response util methods (exact names): `success`, `fail`, ...
+- ResponseEntity usage pattern: ...
+
+## Aggregator/Orchestrator Layer
+
+- Exists: yes/no
+- Naming convention: <e.g. `{Domain}Aggregator`>
+- Return type: <ResponseEntity | DTO | VO>
+- DTO ↔ VO conversion: <MapStruct | extension functions | manual>
+- Dependency rule: <injects Service only | DAO allowed | ...>
+
+## Data Access
+
+- ORM approach: ...
+- Repository/Mapper structure: ...
+- MyBatis mapper XML path (verbatim, if applicable): ...
+- Transaction strategy: ...
+
+## Domains
+
+List every detected domain with its key controller/service/DTO method names.
+For each domain, list AT MOST 10 method signatures (exact names, not paraphrased).
+
+### <domain-1>
+- Controllers: `MethodA()`, `MethodB()`, ...
+- Services: `doX()`, `doY()`, ...
+- Key DTOs: `XRequest`, `XResponse`, ...
+
+### <domain-2>
+...
+
+## Shared / Base Classes (must not be re-declared)
+
+List FQNs of base classes and key utilities that generated code should
+IMPORT, not redefine.
+
+- `com.company.shared.base.BaseEntity`
+- `com.company.shared.util.DateUtils`
+- ...
+
+## Naming Conventions
+
+- Class naming: ...
+- Package structure: ...
+- DTO naming: ...
+- URL pattern: ...
+- DB table/column: ...
+- Module naming: ...
+
+## Authentication & Security
+
+- Authentication method: <JWT | Session | OAuth2 | ...>
+- Authorization pattern: <@PreAuthorize | SecurityConfig | ...>
+- CORS configuration: ...
+
+## Testing
+
+- Test frameworks: ...
+- Test naming convention: ...
+- Mocking library: ...
+
+## Build & Tooling
+
+- Gradle/Maven DSL: ...
+- Version catalog usage: ...
+- Lint/format tools: ...
+
+## Anti-patterns Detected (CRITICAL findings only)
+
+- ...
+- ...
+```
+
+## Rules for Pass 3a
+
+1. **Read each input file AT MOST ONCE.** After reading, all fact extraction
+   must be from your in-context memory of the file.
+2. **Be terse.** This document will be loaded into every subsequent Pass 3
+   step's context. Keep it under 10 KB. Use bullet lists, not prose.
+3. **Exact values only.** Every class name, method name, package path, and
+   file path must be verbatim from the analysis data. If a value is not
+   captured in the analysis, write `(not in analysis)` — do NOT guess.
+4. **Do NOT write any other files.** CLAUDE.md, standard/, rules/, etc.
+   come in later Pass 3 steps. Writing them here is a bug.
+5. **Do NOT read source code.** All information comes from the three JSON
+   inputs. The source has already been analyzed in Pass 1 and Pass 2.
+
+Once `pass3a-facts.md` is written, Pass 3a is complete.

package/pass-prompts/templates/common/pass3b-core-header.md

@@ -0,0 +1,58 @@
+## Pass 3b — CLAUDE.md, standards, and rules (core generation)
+
+This is step 2 of 4 in Pass 3 split mode. Inputs: the fact sheet you wrote
+in Pass 3a, plus the stack-specific template body below.
+
+## Inputs
+
+1. `claudeos-core/generated/pass3a-facts.md` — the fact sheet from Pass 3a
+   **(primary source — reference this for all consistency checks)**
+2. `claudeos-core/generated/pass3-context.json` — slim structured summary
+3. **DO NOT re-read** `pass2-merged.json` or `project-analysis.json`.
+   All facts you need are in `pass3a-facts.md`. Re-reading the full analysis
+   files during this step is the primary cause of context overflow.
+
+## Scope of this step
+
+Generate ONLY the following output categories:
+
+- `CLAUDE.md` (project root)
+- `claudeos-core/standard/` — all standard files per active domains
+- `.claude/rules/` — all rule files (via staging directory, see staging-override)
+
+**Do NOT generate in this step**:
+- `claudeos-core/skills/` (Pass 3c)
+- `claudeos-core/guide/` (Pass 3c)
+- `claudeos-core/plan/` (Pass 3d)
+- `claudeos-core/database/` (Pass 3d)
+- `claudeos-core/mcp-guide/` (Pass 3d)
+
+## Idempotent writes (resume-safe)
+
+Before writing each file:
+
+1. Use the Read tool to check if the target file exists.
+2. If it EXISTS with more than 100 characters of real content:
+   - Print `[SKIP] <path>`
+   - Proceed to next target WITHOUT regenerating
+3. If MISSING or <=100 characters, generate fresh using the fact sheet.
+
+For staged rules, check BOTH:
+- `claudeos-core/generated/.staged-rules/<subpath>` (staging)
+- `.claude/rules/<subpath>` (final — moved by orchestrator after this step)
+
+If either has content, skip.
+
+## Cross-file consistency
+
+Every claim about response flow, method names, class names, or file paths
+MUST match the fact sheet exactly. If `pass3a-facts.md` says
+`Wrapper layer: Controller`, then `architecture.md`,
+`controller-patterns.md`, `response-exception.md`, and
+`controller-rules.md` MUST all show Controller wrapping the response.
+
+---
+
+Below is the stack-specific generation guide. Apply these instructions within
+the scope defined above.
+

package/pass-prompts/templates/common/pass3c-skills-guide-header.md

@@ -0,0 +1,53 @@
+## Pass 3c — Skills and Guides
+
+This is step 3 of 4 in Pass 3 split mode.
+
+## Inputs
+
+1. `claudeos-core/generated/pass3a-facts.md` — **authoritative fact sheet**
+2. `claudeos-core/generated/pass3-context.json` — slim structured summary
+3. Already-generated files (for consistency reference, read on-demand ONLY):
+   - `CLAUDE.md`
+   - `claudeos-core/standard/**/*.md`
+   - `.claude/rules/**/*.md`
+
+**DO NOT re-read** `pass2-merged.json` or `project-analysis.json`.
+Read generated standards/rules only when you need to cite a specific
+section name or file path in a skill or guide.
+
+## Scope of this step
+
+Generate ONLY:
+
+- `claudeos-core/skills/` — orchestrator skill + sub-skills
+  - `10.backend-crud/scaffold-crud-feature/` (if backend present)
+  - `20.frontend-page/scaffold-page-feature/` (if frontend present)
+  - `50.testing/` (if applicable)
+  - `00.shared/` (shared skills)
+- `claudeos-core/guide/` — all 9 expected guide files:
+  - `01.onboarding/*.md`
+  - `02.usage/*.md`
+  - `03.troubleshooting/*.md`
+  - `04.architecture/*.md`
+
+## Idempotent writes
+
+Same rule as Pass 3b: Read the target first, skip if exists with >100 chars.
+Skills with sub-skills: check both the orchestrator `SKILL.md` AND each
+sub-skill `SKILL.md` file independently.
+
+## Cross-file consistency
+
+Skills generate code templates (scaffold outputs). These templates MUST
+use EXACT class/method names from the fact sheet. If a scaffold-crud-feature
+generates a controller stub, that stub's response handling MUST match
+the wrapper layer recorded in `pass3a-facts.md`.
+
+Guide files reference standards and rules. Use the standards/rules file
+paths and section headings that were actually written by Pass 3b —
+read those files if needed to cite accurately. Do NOT invent section names.
+
+---
+
+Below is the stack-specific generation guide for skills and guides:
+

package/pass-prompts/templates/common/pass3d-plan-aux-header.md

@@ -0,0 +1,57 @@
+## Pass 3d — Auxiliary Documentation (database + mcp-guide)
+
+This is step 4 of 4 in Pass 3 split mode (final step).
+
+Master plan aggregation (previously steps 3d-standard / 3d-rules / 3d-skills /
+3d-guide) was removed because master plans are an internal tool backup not
+consumed by Claude Code at runtime, and aggregating many files in a single
+session was the primary source of `Prompt is too long` failures on mid-sized
+projects. The remaining 3d work is limited to two auxiliary directories.
+
+## Inputs
+
+1. `claudeos-core/generated/pass3a-facts.md` — **authoritative fact sheet**
+2. `claudeos-core/generated/pass3-context.json` — slim structured summary
+
+**DO NOT re-read** `pass2-merged.json` or `project-analysis.json`.
+
+**DO NOT re-read** files generated by Pass 3b/3c unless a specific section
+below asks for it. The facts you need are already in `pass3a-facts.md`.
+
+## Scope of this step
+
+Generate ONLY:
+
+- `claudeos-core/database/` — schema docs and SQL reference
+  (only if `pass3a-facts.md` shows a database was detected; otherwise write
+  a single `README.md` stub explaining no database was detected, then stop)
+- `claudeos-core/mcp-guide/` — MCP integration guide
+  (only if the stack has relevant MCP server suggestions; otherwise write
+  a single `README.md` stub, then stop)
+
+**DO NOT generate `claudeos-core/plan/` or any `*-master.md` files.** Master
+plan generation was removed. If a `plan/` directory exists from a previous
+run, leave it untouched (Rule B idempotency).
+
+## Idempotent writes
+
+Same rule as Pass 3b/3c: before writing `database/**/*.md` or
+`mcp-guide/**/*.md`, read the file first. If it exists with substantive
+content (> 100 chars after trim), skip and print `[SKIP] <path>`.
+
+## Output completeness
+
+After this step, the following must exist:
+
+- `claudeos-core/database/` with at least one `.md` file (README.md stub is
+  acceptable when no database was detected).
+- `claudeos-core/mcp-guide/` with at least one `.md` file (README.md stub is
+  acceptable when no relevant MCP integration applies).
+
+Both directories are "best-effort" — their absence is a warning, not a
+failure.
+
+---
+
+Below is the stack-specific generation guide for database and MCP:
+

package/pass-prompts/templates/common/pass4.md

@@ -190,20 +190,9 @@ Proposals with confidence >= 0.70 deserve serious consideration. Do NOT edit pro
 
 ---
 
-## Required output — Master plan (`{{PROJECT_ROOT}}/claudeos-core/plan/`)
-
-### 11. `plan/50.memory-master.md`
-Master plan aggregating all 4 memory files using the `<file path="...">` format.
-Include the full content of each memory file (including the seed decision-log entries
-and project-specific compaction categories).
-
-Write the plan's **title** (e.g., `# 50. Memory Master Plan`) and any **introductory sentence** before the `<file>` blocks in **{{LANG_NAME}}**. Inside each `<file path="...">` block, include the already-translated memory file content verbatim — do NOT re-translate or modify it.
-
----
-
 ## Required output — CLAUDE.md append
 
-### 12. Append a new section to existing `CLAUDE.md`
+### 11. Append a new section to existing `CLAUDE.md`
 
 Do NOT overwrite existing CLAUDE.md content — **append only** at the end.
 The new section must include the `(L4)` marker in its top heading (the marker
@@ -214,7 +203,6 @@ Include:
 - Common rules table (references to `00.core/51.doc-writing-rules.md` and `52.ai-work-rules.md`)
 - L4 Memory table with 4 files (decision-log, failure-patterns, compaction, auto-rule-update)
   — include Purpose and Action columns
-- Master plan reference (`claudeos-core/plan/50.memory-master.md`)
 - Memory workflow (6 numbered steps: scan failure-patterns at session start →
   skim recent decision-log → record decisions → record repeat errors →
   periodic `memory compact` → review `auto-rule-update.md` proposals)
@@ -226,7 +214,7 @@ file paths, and CLI command names in English.
 
 ## Required output — Standard (`{{PROJECT_ROOT}}/claudeos-core/standard/00.core/`)
 
-### 13. `claudeos-core/standard/00.core/XX.doc-writing-guide.md`
+### 12. `claudeos-core/standard/00.core/XX.doc-writing-guide.md`
 Scan `claudeos-core/standard/00.core/` for existing numbered files. Use the **next sequential number**
 (e.g., if `01`, `02`, `03` exist, create `04.doc-writing-guide.md`).
 
@@ -234,7 +222,7 @@ This is the **detailed guide** corresponding to `.claude/rules/00.core/51.doc-wr
 Write entirely in **{{LANG_NAME}}**.
 
 Body must cover:
-1. **Document layer structure** — table with Layer/Path/Role/Loading for CLAUDE.md, Rules, Standard, Skills, Memory, Master Plan. Include a relationship diagram showing CLAUDE.md → Rules → Standard flow.
+1. **Document layer structure** — table with Layer/Path/Role/Loading for CLAUDE.md, Rules, Standard, Skills, Memory. Include a relationship diagram showing CLAUDE.md → Rules → Standard flow.
 2. **Rules file writing rules** — paths frontmatter required, paths scope settings table, file structure template with `## Reference` section. Do not propose narrowing `**/*`.
 3. **Standard file writing rules** — long files are normal, file naming (`number.name.md`), register in `00.standard-reference.md`.
 4. **Skills file writing rules** — MANIFEST.md registration, paths frontmatter in generated rules, verify reference paths.
@@ -244,7 +232,7 @@ Body must cover:
 8. **No domain-specific hardcoding** — use generic placeholder patterns.
 9. **No system absolute path hardcoding** — use relative paths, include ❌/✅ examples.
 10. **Code examples must follow project conventions** — verify directory structure, follow project naming conventions.
-11. **Change synchronization procedure** — table showing Standard→Rules+Plan, Rules→Plan, Skills→MANIFEST+Plan sync targets.
+11. **Change synchronization procedure** — table showing Standard→Rules, Rules→Standard (verify `## Reference` link), Skills→MANIFEST sync targets.
 
 ---
 
@@ -272,9 +260,6 @@ After all files are written, create:
     ".claude/rules/60.memory/03.compaction.md",
     ".claude/rules/60.memory/04.auto-rule-update.md"
   ],
-  "planFiles": [
-    "claudeos-core/plan/50.memory-master.md"
-  ],
   "standardFiles": [
     "claudeos-core/standard/00.core/XX.doc-writing-guide.md"
   ],

package/pass-prompts/templates/java-spring/pass3.md

@@ -139,7 +139,6 @@ Generation targets:
      - claudeos-core/standard/50.verification/01.development-verification.md
      - claudeos-core/standard/50.verification/02.testing-strategy.md
      ## DO NOT Read (context waste)
-     - claudeos-core/plan/ — Master Plan backup files. Never read during coding.
      - claudeos-core/generated/ — Build metadata. Not for coding reference.
      - claudeos-core/guide/ — Onboarding/usage guides for humans. Not for coding reference.
      - claudeos-core/mcp-guide/ — MCP server integration docs. Not for coding reference.
@@ -148,9 +147,9 @@ Generation targets:
 
 4. .claude/rules/50.sync/ (3 sync rules — AI fallback reminders)
    - NOTE: These rules remind AI to run `npx claudeos-core refresh` after modifying standard/rules/skills files.
-   - 01.standard-sync.md — Remind AI to update plan/10 when standard is modified
-   - 02.rules-sync.md — Remind AI to update plan/20 when rules are modified
-   - 03.skills-sync.md — Remind AI to update plan/30 when skills are modified
+   - 01.standard-sync.md — Remind AI to update corresponding rule when standard is modified
+   - 02.rules-sync.md — Remind AI to update corresponding standard when rules are modified
+   - 03.skills-sync.md — Remind AI to update MANIFEST.md when skills are modified
 
 5. claudeos-core/skills/ (active domains only)
    - 10.backend-crud/01.scaffold-crud-feature.md (orchestrator)
@@ -168,18 +167,9 @@ Generation targets:
    - 03.troubleshooting/01.troubleshooting.md
    - 04.architecture/01.file-map.md
    - 04.architecture/02.pros-and-cons.md
-
-7. claudeos-core/plan/ (Master Plan — insert full file content in <file> blocks)
-   - 10.standard-master.md — CLAUDE.md + all standard/ files as <file> blocks
-   - 20.rules-master.md — All rules/ (except sync) as <file> blocks
-   - 21.sync-rules-master.md — All sync rules (code block format)
-   - 30.backend-skills-master.md — All backend skills as <file> blocks
-   - 31.frontend-skills-master.md — All frontend skills as <file> blocks (if frontend detected)
-   - 40.guides-master.md — All guide/ files as <file> blocks
-
-8. claudeos-core/database/ (DB documentation)
+7. claudeos-core/database/ (DB documentation)
    - 01.schema-overview.md — Table list, ER diagram description
    - 02.migration-guide.md — Migration procedure, rollback methods
 
-9. claudeos-core/mcp-guide/ (MCP server integration guide)
+8. claudeos-core/mcp-guide/ (MCP server integration guide)
    - 01.mcp-overview.md — List of MCP servers in use, integration methods

package/pass-prompts/templates/kotlin-spring/pass3.md

@@ -144,7 +144,6 @@ Generation targets:
      - claudeos-core/standard/50.verification/01.development-verification.md
      - claudeos-core/standard/50.verification/02.testing-strategy.md
      ## DO NOT Read (context waste)
-     - claudeos-core/plan/ — Master Plan backup files. Never read during coding.
      - claudeos-core/generated/ — Build metadata. Not for coding reference.
      - claudeos-core/guide/ — Onboarding/usage guides for humans. Not for coding reference.
      - claudeos-core/mcp-guide/ — MCP server integration docs. Not for coding reference.
@@ -153,9 +152,9 @@ Generation targets:
 
 4. .claude/rules/50.sync/ (3 sync rules — AI fallback reminders)
    - NOTE: These rules remind AI to run `npx claudeos-core refresh` after modifying standard/rules/skills files.
-   - 01.standard-sync.md — Remind AI to update plan/10 when standard is modified
-   - 02.rules-sync.md — Remind AI to update plan/20 when rules is modified
-   - 03.skills-sync.md — Remind AI to update plan/30 when skills is modified
+   - 01.standard-sync.md — Remind AI to update corresponding rule when standard is modified
+   - 02.rules-sync.md — Remind AI to update corresponding standard when rules are modified
+   - 03.skills-sync.md — Remind AI to update MANIFEST.md when skills are modified
 
 5. claudeos-core/skills/ (active domains only)
    - 10.backend-crud/01.scaffold-crud-feature.md (orchestrator — CQRS-aware: creates both command and query modules)
@@ -185,18 +184,9 @@ Generation targets:
    - 03.troubleshooting/01.troubleshooting.md
    - 04.architecture/01.file-map.md (show all modules and their relationships)
    - 04.architecture/02.pros-and-cons.md
-
-7. claudeos-core/plan/ (Master Plan — insert full file content in <file> blocks)
-   - 10.standard-master.md — CLAUDE.md + all standard/ files as <file> blocks
-   - 20.rules-master.md — All rules/ (except sync) as <file> blocks
-   - 21.sync-rules-master.md — All sync rules (code block format)
-   - 30.backend-skills-master.md — All backend skills as <file> blocks
-   - 31.frontend-skills-master.md — All frontend skills as <file> blocks (if frontend detected)
-   - 40.guides-master.md — All guide/ files as <file> blocks
-
-8. claudeos-core/database/ (DB documentation)
+7. claudeos-core/database/ (DB documentation)
    - 01.schema-overview.md — Table list, ER diagram description, read/write DB separation
    - 02.migration-guide.md — Migration procedure, rollback methods
 
-9. claudeos-core/mcp-guide/ (MCP server integration guide)
+8. claudeos-core/mcp-guide/ (MCP server integration guide)
    - 01.mcp-overview.md — List of MCP servers in use, integration methods

package/pass-prompts/templates/node-express/pass3.md

@@ -115,7 +115,6 @@ Generation targets:
      - claudeos-core/standard/50.verification/01.development-verification.md
      - claudeos-core/standard/50.verification/02.testing-strategy.md
      ## DO NOT Read (context waste)
-     - claudeos-core/plan/ — Master Plan backup files. Never read during coding.
      - claudeos-core/generated/ — Build metadata. Not for coding reference.
      - claudeos-core/guide/ — Onboarding/usage guides for humans. Not for coding reference.
      - claudeos-core/mcp-guide/ — MCP server integration docs. Not for coding reference.
@@ -124,9 +123,9 @@ Generation targets:
 
 4. .claude/rules/50.sync/ (3 sync rules — AI fallback reminders)
    - NOTE: These rules remind AI to run `npx claudeos-core refresh` after modifying standard/rules/skills files.
-   - 01.standard-sync.md — Remind AI to update plan/10 when standard is modified
-   - 02.rules-sync.md — Remind AI to update plan/20 when rules is modified
-   - 03.skills-sync.md — Remind AI to update plan/30 when skills is modified
+   - 01.standard-sync.md — Remind AI to update corresponding rule when standard is modified
+   - 02.rules-sync.md — Remind AI to update corresponding standard when rules are modified
+   - 03.skills-sync.md — Remind AI to update MANIFEST.md when skills are modified
 
 5. claudeos-core/skills/ (active domains only)
    - 10.backend-crud/01.scaffold-crud-feature.md (orchestrator)
@@ -143,17 +142,9 @@ Generation targets:
    - 03.troubleshooting/01.troubleshooting.md
    - 04.architecture/01.file-map.md
    - 04.architecture/02.pros-and-cons.md
-
-7. claudeos-core/plan/ (Master Plan)
-   - 10.standard-master.md — CLAUDE.md + all standard/ files as <file> blocks
-   - 20.rules-master.md — All rules/ (except sync) as <file> blocks
-   - 21.sync-rules-master.md — All sync rules (code block format)
-   - 30.backend-skills-master.md — All backend skills as <file> blocks
-   - 40.guides-master.md — All guide/ files as <file> blocks
-
-8. claudeos-core/database/
+7. claudeos-core/database/
    - 01.schema-overview.md — Table/collection list, relationship description
    - 02.migration-guide.md — Migration procedure, rollback methods
 
-9. claudeos-core/mcp-guide/
+8. claudeos-core/mcp-guide/
    - 01.mcp-overview.md — List of MCP servers in use, integration methods

package/pass-prompts/templates/node-fastify/pass3.md

@@ -97,17 +97,9 @@ Generation targets:
    - 03.troubleshooting/01.troubleshooting.md
    - 04.architecture/01.file-map.md
    - 04.architecture/02.pros-and-cons.md
-
-7. claudeos-core/plan/ (Master Plan)
-   - 10.standard-master.md
-   - 20.rules-master.md
-   - 21.sync-rules-master.md
-   - 30.backend-skills-master.md
-   - 40.guides-master.md
-
-8. claudeos-core/database/
+7. claudeos-core/database/
    - 01.schema-overview.md
    - 02.migration-guide.md
 
-9. claudeos-core/mcp-guide/
+8. claudeos-core/mcp-guide/
    - 01.mcp-overview.md