claude-code-generator 0.2.4__tar.gz → 0.4.1__tar.gz

{claude_code_generator-0.2.4/src/claude_code_generator.egg-info → claude_code_generator-0.4.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: claude-code-generator
-Version: 0.2.4
+Version: 0.4.1
 Summary: Orchestrator CLI that drives Claude Code end-to-end to generate whole projects from a requirements.md file.
 Author: Silvio Baratto
 License: MIT
@@ -20,13 +20,22 @@ Classifier: Topic :: Software Development :: Code Generators
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: claude-agent-sdk
+Requires-Dist: claude-agent-sdk<0.2.0,>=0.1.61
 Requires-Dist: typer
 Requires-Dist: rich
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Requires-Dist: pytest-asyncio; extra == "dev"
 Requires-Dist: ruff; extra == "dev"
+Provides-Extra: repomap
+Requires-Dist: tree-sitter<0.22,>=0.21; extra == "repomap"
+Requires-Dist: tree-sitter-languages>=1.10; extra == "repomap"
+Provides-Extra: mcp
+Requires-Dist: mcp<2.0,>=1.0; extra == "mcp"
+Provides-Extra: batch
+Requires-Dist: anthropic>=0.35; extra == "batch"
+Provides-Extra: all
+Requires-Dist: claude-code-generator[batch,mcp,repomap]; extra == "all"
 Dynamic: license-file
 
 # code-generator
@@ -72,7 +81,7 @@ $EDITOR .code-generator/requirements.md
 
 Pre-flight rewrite of `.code-generator/requirements.md` into the canonical structure the orchestrator expects: `# Title` → `## Description` → `## Tech Stack` → `## Goals` → `## Scope` (numbered `### N.` subsections with per-section `Acceptance criteria` checklists) → `## Non-goals` → `## Constraints` → `## Global acceptance criteria`.
 
-Runs a single Opus 4.6 session with a dedicated prompt, preserves the original intent verbatim when ambiguous (never invents new scope), and writes the rewrite back to disk atomically. The pre-optimize original is saved to `.code-generator/requirements.backup-<UTC-timestamp>.md` so you can always `diff` and recover. The command is stateless — it never reads or writes `state.json` and never calls `gh`.
+Runs a single Opus 4.7 session with a dedicated prompt, preserves the original intent verbatim when ambiguous (never invents new scope), and writes the rewrite back to disk atomically. The pre-optimize original is saved to `.code-generator/requirements.backup-<UTC-timestamp>.md` so you can always `diff` and recover. The command is stateless — it never reads or writes `state.json` and never calls `gh`.
 
 Short-circuits and exits 0 if the file is already canonical. Use `--force` to re-optimize anyway. Use `--dry-run` to print the proposed rewrite to stdout without touching the file or creating a backup.
 
@@ -88,13 +97,13 @@ Read `.code-generator/requirements.md` and orchestrate Claude Code through the 0
 
 | Phase | Action | Model |
 |---|---|---|
-| 0 | Complexity analysis (single vs multi-cycle) | Opus 4.6 |
-| 1 | Planning — creates GitHub issues + milestone | Opus 4.6 |
-| 2 | Per-issue review | Opus 4.6 |
+| 0 | Complexity analysis (single vs multi-cycle) | Opus 4.7 |
+| 1 | Planning — creates GitHub issues + milestone | Opus 4.7 |
+| 2 | Per-issue review | Opus 4.7 |
 | 3-4 | Implementation, fresh SDK session per issue (TDD) | Sonnet 4.6 |
-| 5 | Closure + cross-module review | Opus 4.6 |
+| 5 | Closure + cross-module review | Opus 4.7 |
 | 6 | Test suite, max 3 retries on failure | Sonnet 4.6 |
-| 7 | Commit message + git add/commit/push | Opus 4.6 |
+| 7 | Commit message + git add/commit/push | Opus 4.7 |
 
 Every phase logs a completion summary with token counts (`in`, `cache_read`, `cache_write`, `out`) and persists them to `state.json`.
 
@@ -107,6 +116,11 @@ Flags:
 | `--continue` | Resume from the last completed phase (reads `state.json`) | false |
 | `--mode auto\|single\|multi-cycle` | Force a mode (default `auto` runs Phase 0 first) | auto |
 | `--cycle N` | Resume from cycle N (multi-cycle only) | — |
+| `--max-turns N` | Opt-in per-session turn cap for debugging or strict cost control (positive integer); omit for no cap | — |
+
+#### Turn budgets
+
+No per-session turn budget is set by default. Sessions terminate via `end_turn` on the model side; the real safety nets are rate-limit handling, overage abort (non-negotiable #4), and task budgets — **not** a guessed integer. `--max-turns N` exists on `generate`, `optimize`, and `review` as an opt-in for debugging or strict cost control; it is not for production.
 
 In multi-cycle mode each cycle is a GitHub Milestone and produces a committable, working increment. Each cycle starts a fresh SDK session and re-reads the committed codebase, so the model never relies on stale conversational context.
 
@@ -114,7 +128,7 @@ If the Max subscription hits a rate limit, the tool **pauses** — it persists t
 
 ### `code-generator review [--create-issues] [--severity LEVEL]`
 
-Run a standalone Opus 4.6 review of the current codebase against the design principles in `requirements.md` §4 (SOLID, Clean Code, TDD, YAGNI/KISS/DRY). With `--create-issues`, every finding becomes a GitHub Issue.
+Run a standalone Opus 4.7 review of the current codebase against the design principles in `requirements.md` §4 (SOLID, Clean Code, TDD, YAGNI/KISS/DRY). With `--create-issues`, every finding becomes a GitHub Issue.
 
 ### `code-generator status`
 
@@ -122,7 +136,7 @@ Print a Rich table with the current mode, current cycle, current phase, open/clo
 
 ## Safety constraints
 
-The tool enforces eight non-negotiables (see `CLAUDE.md` for the full list):
+The tool enforces nine non-negotiables (see `CLAUDE.md` for the full list):
 
 1. **Max subscription only, zero API credits.** Before any SDK or subprocess call, the tool strips `ANTHROPIC_API_KEY`, `ANTHROPIC_AUTH_TOKEN`, `ANTHROPIC_BEDROCK_API_KEY`, `ANTHROPIC_VERTEX_PROJECT_ID`, `CLAUDE_CODE_USE_BEDROCK`, `CLAUDE_CODE_USE_VERTEX` from the environment. A startup check refuses to run if any are present.
 2. **Never `--bare`.** The CLI flag `--bare` skips OAuth and forces API credits — it is **never** passed by this tool. If Anthropic ever makes `--bare` the default for `claude -p`, pin the CLI version or pass the opposite flag explicitly.
@@ -131,7 +145,7 @@ The tool enforces eight non-negotiables (see `CLAUDE.md` for the full list):
 5. **Wait-and-resume rate-limit handling.** Rate limits are not retried with exponential backoff — the tool sleeps until `resets_at + 60s` and resumes the same session. Backoff (10→20→40→80→120s) only applies to non-rate-limit transient errors.
 6. **Fresh SDK session per issue.** The implementation phase never reuses conversational context across issues — each issue is a `/clear`-equivalent fresh session.
 7. **Atomic state writes.** `.code-generator/state.json` is updated via `tmp → os.replace(tmp, STATE)` so a crash mid-write cannot corrupt it.
-8. **Fixed model per phase.** Opus 4.6 for phases 0/1/2/5/7; Sonnet 4.6 for phases 3/4/6.
+8. **Fixed model per phase.** Opus 4.7 for phases 0/1/2/5/7; Sonnet 4.6 for phases 3/4/6.
 
 ## Troubleshooting
 
@@ -183,6 +197,12 @@ pytest -q
 ruff check src tests
 ```
 
+## Benchmarking
+
+```bash
+code-generator bench --fake --cycles 3 --output bench.json
+```
+
 ## Reference
 
 The full specification lives in [`requirements.md`](requirements.md). Each prompt file (`prompt-phase-*.md`, `prompt-review.md`) is loaded verbatim by the orchestrator with `{NAME}` placeholders substituted at runtime — edit the prompts there, not inline in Python.

{claude_code_generator-0.2.4 → claude_code_generator-0.4.1}/README.md

@@ -41,7 +41,7 @@ $EDITOR .code-generator/requirements.md
 
 Pre-flight rewrite of `.code-generator/requirements.md` into the canonical structure the orchestrator expects: `# Title` → `## Description` → `## Tech Stack` → `## Goals` → `## Scope` (numbered `### N.` subsections with per-section `Acceptance criteria` checklists) → `## Non-goals` → `## Constraints` → `## Global acceptance criteria`.
 
-Runs a single Opus 4.6 session with a dedicated prompt, preserves the original intent verbatim when ambiguous (never invents new scope), and writes the rewrite back to disk atomically. The pre-optimize original is saved to `.code-generator/requirements.backup-<UTC-timestamp>.md` so you can always `diff` and recover. The command is stateless — it never reads or writes `state.json` and never calls `gh`.
+Runs a single Opus 4.7 session with a dedicated prompt, preserves the original intent verbatim when ambiguous (never invents new scope), and writes the rewrite back to disk atomically. The pre-optimize original is saved to `.code-generator/requirements.backup-<UTC-timestamp>.md` so you can always `diff` and recover. The command is stateless — it never reads or writes `state.json` and never calls `gh`.
 
 Short-circuits and exits 0 if the file is already canonical. Use `--force` to re-optimize anyway. Use `--dry-run` to print the proposed rewrite to stdout without touching the file or creating a backup.
 
@@ -57,13 +57,13 @@ Read `.code-generator/requirements.md` and orchestrate Claude Code through the 0
 
 | Phase | Action | Model |
 |---|---|---|
-| 0 | Complexity analysis (single vs multi-cycle) | Opus 4.6 |
-| 1 | Planning — creates GitHub issues + milestone | Opus 4.6 |
-| 2 | Per-issue review | Opus 4.6 |
+| 0 | Complexity analysis (single vs multi-cycle) | Opus 4.7 |
+| 1 | Planning — creates GitHub issues + milestone | Opus 4.7 |
+| 2 | Per-issue review | Opus 4.7 |
 | 3-4 | Implementation, fresh SDK session per issue (TDD) | Sonnet 4.6 |
-| 5 | Closure + cross-module review | Opus 4.6 |
+| 5 | Closure + cross-module review | Opus 4.7 |
 | 6 | Test suite, max 3 retries on failure | Sonnet 4.6 |
-| 7 | Commit message + git add/commit/push | Opus 4.6 |
+| 7 | Commit message + git add/commit/push | Opus 4.7 |
 
 Every phase logs a completion summary with token counts (`in`, `cache_read`, `cache_write`, `out`) and persists them to `state.json`.
 
@@ -76,6 +76,11 @@ Flags:
 | `--continue` | Resume from the last completed phase (reads `state.json`) | false |
 | `--mode auto\|single\|multi-cycle` | Force a mode (default `auto` runs Phase 0 first) | auto |
 | `--cycle N` | Resume from cycle N (multi-cycle only) | — |
+| `--max-turns N` | Opt-in per-session turn cap for debugging or strict cost control (positive integer); omit for no cap | — |
+
+#### Turn budgets
+
+No per-session turn budget is set by default. Sessions terminate via `end_turn` on the model side; the real safety nets are rate-limit handling, overage abort (non-negotiable #4), and task budgets — **not** a guessed integer. `--max-turns N` exists on `generate`, `optimize`, and `review` as an opt-in for debugging or strict cost control; it is not for production.
 
 In multi-cycle mode each cycle is a GitHub Milestone and produces a committable, working increment. Each cycle starts a fresh SDK session and re-reads the committed codebase, so the model never relies on stale conversational context.
 
@@ -83,7 +88,7 @@ If the Max subscription hits a rate limit, the tool **pauses** — it persists t
 
 ### `code-generator review [--create-issues] [--severity LEVEL]`
 
-Run a standalone Opus 4.6 review of the current codebase against the design principles in `requirements.md` §4 (SOLID, Clean Code, TDD, YAGNI/KISS/DRY). With `--create-issues`, every finding becomes a GitHub Issue.
+Run a standalone Opus 4.7 review of the current codebase against the design principles in `requirements.md` §4 (SOLID, Clean Code, TDD, YAGNI/KISS/DRY). With `--create-issues`, every finding becomes a GitHub Issue.
 
 ### `code-generator status`
 
@@ -91,7 +96,7 @@ Print a Rich table with the current mode, current cycle, current phase, open/clo
 
 ## Safety constraints
 
-The tool enforces eight non-negotiables (see `CLAUDE.md` for the full list):
+The tool enforces nine non-negotiables (see `CLAUDE.md` for the full list):
 
 1. **Max subscription only, zero API credits.** Before any SDK or subprocess call, the tool strips `ANTHROPIC_API_KEY`, `ANTHROPIC_AUTH_TOKEN`, `ANTHROPIC_BEDROCK_API_KEY`, `ANTHROPIC_VERTEX_PROJECT_ID`, `CLAUDE_CODE_USE_BEDROCK`, `CLAUDE_CODE_USE_VERTEX` from the environment. A startup check refuses to run if any are present.
 2. **Never `--bare`.** The CLI flag `--bare` skips OAuth and forces API credits — it is **never** passed by this tool. If Anthropic ever makes `--bare` the default for `claude -p`, pin the CLI version or pass the opposite flag explicitly.
@@ -100,7 +105,7 @@ The tool enforces eight non-negotiables (see `CLAUDE.md` for the full list):
 5. **Wait-and-resume rate-limit handling.** Rate limits are not retried with exponential backoff — the tool sleeps until `resets_at + 60s` and resumes the same session. Backoff (10→20→40→80→120s) only applies to non-rate-limit transient errors.
 6. **Fresh SDK session per issue.** The implementation phase never reuses conversational context across issues — each issue is a `/clear`-equivalent fresh session.
 7. **Atomic state writes.** `.code-generator/state.json` is updated via `tmp → os.replace(tmp, STATE)` so a crash mid-write cannot corrupt it.
-8. **Fixed model per phase.** Opus 4.6 for phases 0/1/2/5/7; Sonnet 4.6 for phases 3/4/6.
+8. **Fixed model per phase.** Opus 4.7 for phases 0/1/2/5/7; Sonnet 4.6 for phases 3/4/6.
 
 ## Troubleshooting
 
@@ -152,6 +157,12 @@ pytest -q
 ruff check src tests
 ```
 
+## Benchmarking
+
+```bash
+code-generator bench --fake --cycles 3 --output bench.json
+```
+
 ## Reference
 
 The full specification lives in [`requirements.md`](requirements.md). Each prompt file (`prompt-phase-*.md`, `prompt-review.md`) is loaded verbatim by the orchestrator with `{NAME}` placeholders substituted at runtime — edit the prompts there, not inline in Python.

{claude_code_generator-0.2.4 → claude_code_generator-0.4.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "claude-code-generator"
-version = "0.2.4"
+version = "0.4.1"
 description = "Orchestrator CLI that drives Claude Code end-to-end to generate whole projects from a requirements.md file."
 readme = "README.md"
 license = { text = "MIT" }
@@ -23,7 +23,7 @@ classifiers = [
 ]
 requires-python = ">=3.11"
 dependencies = [
-    "claude-agent-sdk",
+    "claude-agent-sdk>=0.1.61,<0.2.0",
     "typer",
     "rich",
 ]
@@ -39,6 +39,27 @@ dev = [
     "pytest-asyncio",
     "ruff",
 ]
+repomap = [
+    "tree-sitter>=0.21,<0.22",
+    "tree-sitter-languages>=1.10",
+]
+mcp = [
+    # Python MCP client library — the codebase-memory-mcp and serena binaries
+    # are external and must be installed separately:
+    #   cargo install codebase-memory-mcp   (or brew install codebase-memory-mcp)
+    #   pip install serena                  (or uvx serena for zero-install)
+    "mcp>=1.0,<2.0",
+]
+batch = [
+    # Anthropic SDK for the Messages Batches API (§6).
+    # Install via: pipx install "claude-code-generator[batch]"
+    "anthropic>=0.35",
+]
+all = [
+    # Everything pip can install. External binaries (codebase-memory-mcp Rust
+    # crate, serena) must be installed separately — see README.
+    "claude-code-generator[repomap,mcp,batch]",
+]
 
 [project.scripts]
 code-generator = "code_generator.cli:app"

{claude_code_generator-0.2.4 → claude_code_generator-0.4.1/src/claude_code_generator.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: claude-code-generator
-Version: 0.2.4
+Version: 0.4.1
 Summary: Orchestrator CLI that drives Claude Code end-to-end to generate whole projects from a requirements.md file.
 Author: Silvio Baratto
 License: MIT
@@ -20,13 +20,22 @@ Classifier: Topic :: Software Development :: Code Generators
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: claude-agent-sdk
+Requires-Dist: claude-agent-sdk<0.2.0,>=0.1.61
 Requires-Dist: typer
 Requires-Dist: rich
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Requires-Dist: pytest-asyncio; extra == "dev"
 Requires-Dist: ruff; extra == "dev"
+Provides-Extra: repomap
+Requires-Dist: tree-sitter<0.22,>=0.21; extra == "repomap"
+Requires-Dist: tree-sitter-languages>=1.10; extra == "repomap"
+Provides-Extra: mcp
+Requires-Dist: mcp<2.0,>=1.0; extra == "mcp"
+Provides-Extra: batch
+Requires-Dist: anthropic>=0.35; extra == "batch"
+Provides-Extra: all
+Requires-Dist: claude-code-generator[batch,mcp,repomap]; extra == "all"
 Dynamic: license-file
 
 # code-generator
@@ -72,7 +81,7 @@ $EDITOR .code-generator/requirements.md
 
 Pre-flight rewrite of `.code-generator/requirements.md` into the canonical structure the orchestrator expects: `# Title` → `## Description` → `## Tech Stack` → `## Goals` → `## Scope` (numbered `### N.` subsections with per-section `Acceptance criteria` checklists) → `## Non-goals` → `## Constraints` → `## Global acceptance criteria`.
 
-Runs a single Opus 4.6 session with a dedicated prompt, preserves the original intent verbatim when ambiguous (never invents new scope), and writes the rewrite back to disk atomically. The pre-optimize original is saved to `.code-generator/requirements.backup-<UTC-timestamp>.md` so you can always `diff` and recover. The command is stateless — it never reads or writes `state.json` and never calls `gh`.
+Runs a single Opus 4.7 session with a dedicated prompt, preserves the original intent verbatim when ambiguous (never invents new scope), and writes the rewrite back to disk atomically. The pre-optimize original is saved to `.code-generator/requirements.backup-<UTC-timestamp>.md` so you can always `diff` and recover. The command is stateless — it never reads or writes `state.json` and never calls `gh`.
 
 Short-circuits and exits 0 if the file is already canonical. Use `--force` to re-optimize anyway. Use `--dry-run` to print the proposed rewrite to stdout without touching the file or creating a backup.
 
@@ -88,13 +97,13 @@ Read `.code-generator/requirements.md` and orchestrate Claude Code through the 0
 
 | Phase | Action | Model |
 |---|---|---|
-| 0 | Complexity analysis (single vs multi-cycle) | Opus 4.6 |
-| 1 | Planning — creates GitHub issues + milestone | Opus 4.6 |
-| 2 | Per-issue review | Opus 4.6 |
+| 0 | Complexity analysis (single vs multi-cycle) | Opus 4.7 |
+| 1 | Planning — creates GitHub issues + milestone | Opus 4.7 |
+| 2 | Per-issue review | Opus 4.7 |
 | 3-4 | Implementation, fresh SDK session per issue (TDD) | Sonnet 4.6 |
-| 5 | Closure + cross-module review | Opus 4.6 |
+| 5 | Closure + cross-module review | Opus 4.7 |
 | 6 | Test suite, max 3 retries on failure | Sonnet 4.6 |
-| 7 | Commit message + git add/commit/push | Opus 4.6 |
+| 7 | Commit message + git add/commit/push | Opus 4.7 |
 
 Every phase logs a completion summary with token counts (`in`, `cache_read`, `cache_write`, `out`) and persists them to `state.json`.
 
@@ -107,6 +116,11 @@ Flags:
 | `--continue` | Resume from the last completed phase (reads `state.json`) | false |
 | `--mode auto\|single\|multi-cycle` | Force a mode (default `auto` runs Phase 0 first) | auto |
 | `--cycle N` | Resume from cycle N (multi-cycle only) | — |
+| `--max-turns N` | Opt-in per-session turn cap for debugging or strict cost control (positive integer); omit for no cap | — |
+
+#### Turn budgets
+
+No per-session turn budget is set by default. Sessions terminate via `end_turn` on the model side; the real safety nets are rate-limit handling, overage abort (non-negotiable #4), and task budgets — **not** a guessed integer. `--max-turns N` exists on `generate`, `optimize`, and `review` as an opt-in for debugging or strict cost control; it is not for production.
 
 In multi-cycle mode each cycle is a GitHub Milestone and produces a committable, working increment. Each cycle starts a fresh SDK session and re-reads the committed codebase, so the model never relies on stale conversational context.
 
@@ -114,7 +128,7 @@ If the Max subscription hits a rate limit, the tool **pauses** — it persists t
 
 ### `code-generator review [--create-issues] [--severity LEVEL]`
 
-Run a standalone Opus 4.6 review of the current codebase against the design principles in `requirements.md` §4 (SOLID, Clean Code, TDD, YAGNI/KISS/DRY). With `--create-issues`, every finding becomes a GitHub Issue.
+Run a standalone Opus 4.7 review of the current codebase against the design principles in `requirements.md` §4 (SOLID, Clean Code, TDD, YAGNI/KISS/DRY). With `--create-issues`, every finding becomes a GitHub Issue.
 
 ### `code-generator status`
 
@@ -122,7 +136,7 @@ Print a Rich table with the current mode, current cycle, current phase, open/clo
 
 ## Safety constraints
 
-The tool enforces eight non-negotiables (see `CLAUDE.md` for the full list):
+The tool enforces nine non-negotiables (see `CLAUDE.md` for the full list):
 
 1. **Max subscription only, zero API credits.** Before any SDK or subprocess call, the tool strips `ANTHROPIC_API_KEY`, `ANTHROPIC_AUTH_TOKEN`, `ANTHROPIC_BEDROCK_API_KEY`, `ANTHROPIC_VERTEX_PROJECT_ID`, `CLAUDE_CODE_USE_BEDROCK`, `CLAUDE_CODE_USE_VERTEX` from the environment. A startup check refuses to run if any are present.
 2. **Never `--bare`.** The CLI flag `--bare` skips OAuth and forces API credits — it is **never** passed by this tool. If Anthropic ever makes `--bare` the default for `claude -p`, pin the CLI version or pass the opposite flag explicitly.
@@ -131,7 +145,7 @@ The tool enforces eight non-negotiables (see `CLAUDE.md` for the full list):
 5. **Wait-and-resume rate-limit handling.** Rate limits are not retried with exponential backoff — the tool sleeps until `resets_at + 60s` and resumes the same session. Backoff (10→20→40→80→120s) only applies to non-rate-limit transient errors.
 6. **Fresh SDK session per issue.** The implementation phase never reuses conversational context across issues — each issue is a `/clear`-equivalent fresh session.
 7. **Atomic state writes.** `.code-generator/state.json` is updated via `tmp → os.replace(tmp, STATE)` so a crash mid-write cannot corrupt it.
-8. **Fixed model per phase.** Opus 4.6 for phases 0/1/2/5/7; Sonnet 4.6 for phases 3/4/6.
+8. **Fixed model per phase.** Opus 4.7 for phases 0/1/2/5/7; Sonnet 4.6 for phases 3/4/6.
 
 ## Troubleshooting
 
@@ -183,6 +197,12 @@ pytest -q
 ruff check src tests
 ```
 
+## Benchmarking
+
+```bash
+code-generator bench --fake --cycles 3 --output bench.json
+```
+
 ## Reference
 
 The full specification lives in [`requirements.md`](requirements.md). Each prompt file (`prompt-phase-*.md`, `prompt-review.md`) is loaded verbatim by the orchestrator with `{NAME}` placeholders substituted at runtime — edit the prompts there, not inline in Python.

{claude_code_generator-0.2.4 → claude_code_generator-0.4.1}/src/claude_code_generator.egg-info/SOURCES.txt

@@ -14,12 +14,23 @@ src/code_generator/effort.py
 src/code_generator/env.py
 src/code_generator/git_ops.py
 src/code_generator/logging_setup.py
+src/code_generator/memory.py
+src/code_generator/preflight.py
+src/code_generator/repo_info.py
+src/code_generator/repomap.py
 src/code_generator/requirements_structure.py
 src/code_generator/state.py
+src/code_generator/state_retention.py
 src/code_generator/commands/__init__.py
+src/code_generator/commands/_bench_io.py
+src/code_generator/commands/_crash_recovery.py
 src/code_generator/commands/_detect.py
 src/code_generator/commands/_dispatch.py
 src/code_generator/commands/_resume.py
+src/code_generator/commands/_validators.py
+src/code_generator/commands/bench.py
+src/code_generator/commands/bench_compare.py
+src/code_generator/commands/bench_export.py
 src/code_generator/commands/generate.py
 src/code_generator/commands/init.py
 src/code_generator/commands/optimize.py
@@ -31,7 +42,10 @@ src/code_generator/gh/issues.py
 src/code_generator/gh/labels.py
 src/code_generator/gh/milestones.py
 src/code_generator/orchestrator/__init__.py
+src/code_generator/orchestrator/_client_lifecycle.py
 src/code_generator/orchestrator/_comments.py
+src/code_generator/orchestrator/_memory_writers.py
+src/code_generator/orchestrator/_phase5_precommit.py
 src/code_generator/orchestrator/cycle_loop.py
 src/code_generator/orchestrator/phase0_complexity.py
 src/code_generator/orchestrator/phase1_plan.py
@@ -41,6 +55,7 @@ src/code_generator/orchestrator/phase5_closure.py
 src/code_generator/orchestrator/phase6_test.py
 src/code_generator/orchestrator/phase7_commit.py
 src/code_generator/prompts/__init__.py
+src/code_generator/prompts/hashes.py
 src/code_generator/prompts/prompt-optimize-requirements.md
 src/code_generator/prompts/prompt-phase-0-complexity.md
 src/code_generator/prompts/prompt-phase-1-planning.md
@@ -51,12 +66,18 @@ src/code_generator/prompts/prompt-phase-6-test.md
 src/code_generator/prompts/prompt-phase-7-commit.md
 src/code_generator/prompts/prompt-review.md
 src/code_generator/runner/__init__.py
+src/code_generator/runner/_telemetry.py
+src/code_generator/runner/batch.py
+src/code_generator/runner/fake_runner.py
+src/code_generator/runner/mcp.py
 src/code_generator/runner/message_parsing.py
 src/code_generator/runner/options.py
 src/code_generator/runner/protocol.py
 src/code_generator/runner/rate_limit.py
 src/code_generator/runner/retry.py
 src/code_generator/runner/sdk_runner.py
+src/code_generator/runner/soft_reset.py
+src/code_generator/runner/state_guard.py
 src/code_generator/runner/subprocess_runner.py
 src/code_generator/runner/types.py
 src/code_generator/runner/utils.py
@@ -69,12 +90,23 @@ src/code_generator/templates/fullstack.md
 src/code_generator/templates/nestjs.md
 src/code_generator/templates/python-cli.md
 tests/test_agents.py
+tests/test_bench.py
+tests/test_bench_compare.py
+tests/test_bench_export.py
+tests/test_bench_fixture.py
+tests/test_bench_regression.py
+tests/test_changelog.py
+tests/test_claude_md.py
+tests/test_client_lifecycle.py
 tests/test_comments.py
 tests/test_commit_message.py
+tests/test_crash_recovery.py
 tests/test_cycle_loop.py
 tests/test_cycle_loop_multicycle.py
 tests/test_delta_planning.py
+tests/test_dependencies.py
 tests/test_detect.py
+tests/test_docs_no_default_max_turns.py
 tests/test_effort.py
 tests/test_env.py
 tests/test_generate.py
@@ -82,23 +114,40 @@ tests/test_generate_resume.py
 tests/test_gh.py
 tests/test_gh_labels.py
 tests/test_gh_milestones.py
+tests/test_gh_repo_threading.py
 tests/test_gh_submodules.py
 tests/test_git_ops.py
 tests/test_init.py
 tests/test_logging_setup.py
+tests/test_max_turns_cli_flag.py
+tests/test_mcp.py
+tests/test_memory.py
+tests/test_memory_writers.py
 tests/test_message_parsing.py
+tests/test_no_max_turns_in_call_sites.py
+tests/test_no_max_turns_literal.py
+tests/test_non_goals_grep_guard.py
 tests/test_optimize.py
 tests/test_options.py
 tests/test_phase0.py
 tests/test_phase1.py
 tests/test_phase2.py
+tests/test_phase2_batch.py
 tests/test_phase3_4.py
 tests/test_phase5.py
+tests/test_phase5_precommit.py
 tests/test_phase6.py
 tests/test_phase7.py
+tests/test_phase_mcp_regression.py
 tests/test_phase_token_logging.py
+tests/test_preflight.py
+tests/test_prompt_drift.py
+tests/test_prompt_prefix_snapshots.py
+tests/test_prompt_prefix_stability.py
 tests/test_prompts.py
 tests/test_rate_limit.py
+tests/test_repo_info.py
+tests/test_repomap.py
 tests/test_requirements_structure.py
 tests/test_retry.py
 tests/test_review.py
@@ -107,7 +156,12 @@ tests/test_runner_protocol_annotations.py
 tests/test_runner_types.py
 tests/test_runner_utils.py
 tests/test_sdk_runner.py
+tests/test_sdk_runner_shared.py
+tests/test_session_mode.py
 tests/test_state.py
+tests/test_state_guard.py
+tests/test_state_retention.py
 tests/test_status.py
 tests/test_subprocess_runner.py
+tests/test_telemetry.py
 tests/test_version.py

claude_code_generator-0.4.1/src/claude_code_generator.egg-info/requires.txt

@@ -0,0 +1,21 @@
+claude-agent-sdk<0.2.0,>=0.1.61
+typer
+rich
+
+[all]
+claude-code-generator[batch,mcp,repomap]
+
+[batch]
+anthropic>=0.35
+
+[dev]
+pytest
+pytest-asyncio
+ruff
+
+[mcp]
+mcp<2.0,>=1.0
+
+[repomap]
+tree-sitter<0.22,>=0.21
+tree-sitter-languages>=1.10

{claude_code_generator-0.2.4 → claude_code_generator-0.4.1}/src/code_generator/__init__.py

@@ -1,3 +1,3 @@
 """code-generator: orchestrator CLI for end-to-end project generation."""
 
-__version__ = "0.2.4"
+__version__ = "0.4.1"

{claude_code_generator-0.2.4 → claude_code_generator-0.4.1}/src/code_generator/cli.py

@@ -9,6 +9,7 @@ from typing import Annotated
 import typer
 
 import code_generator
+from code_generator.commands.bench import bench_app
 from code_generator.commands.generate import generate_command
 from code_generator.commands.init import init_command
 from code_generator.commands.optimize import optimize_command
@@ -44,6 +45,7 @@ def root(
     """code-generator CLI root."""
 
 
+app.add_typer(bench_app, name="bench")
 app.command(name="init")(init_command)
 app.command(name="status")(status_command)
 app.command(name="generate")(generate_command)

claude_code_generator-0.4.1/src/code_generator/commands/_bench_io.py

@@ -0,0 +1,39 @@
+"""Shared I/O helpers for bench and bench export subcommands.
+
+Extracted to break the circular-import between bench.py (which registers
+subcommands) and bench_export.py (which needs the schema version and writer).
+"""
+
+from __future__ import annotations
+
+import json
+import os
+from typing import TYPE_CHECKING, Any
+
+import typer
+
+if TYPE_CHECKING:
+    from pathlib import Path
+
+# Bump this + add a migration note for any breaking schema change.
+# v2 (issue #212): per-phase dicts + totals now include ``num_turns``.
+SCHEMA_VERSION: int = 2
+
+
+def write_output(path: Path | None, payload: dict[str, Any]) -> None:
+    """Write *payload* as pretty-printed JSON atomically to *path*, or stdout.
+
+    Uses tmp → os.replace() so a crash mid-write cannot corrupt the file.
+
+    Args:
+        path: Destination file path, or None to print to stdout.
+        payload: JSON-serialisable dict to write.
+    """
+    serialised = json.dumps(payload, indent=2)
+    if path is None:
+        typer.echo(serialised)
+        return
+
+    tmp_path = path.with_suffix(".tmp")
+    tmp_path.write_text(serialised, encoding="utf-8")
+    os.replace(tmp_path, path)

claude_code_generator-0.4.1/src/code_generator/commands/_crash_recovery.py

@@ -0,0 +1,52 @@
+"""Crash-recovery helpers for the --continue entry point.
+
+On --continue, if the previous client was alive within the TTL window
+(default 60 minutes), the client was likely orphaned by a process crash.
+The orchestrator always opens a fresh client on resume; this module
+provides the diagnostic check and log message.
+
+The log message is exact — downstream bench/observability tooling greps for it.
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import TYPE_CHECKING
+
+from code_generator import state as _state
+
+if TYPE_CHECKING:
+    from code_generator.state import State
+
+# Exact log string — do NOT change without updating downstream consumers.
+CRASH_RECOVERY_LOG_MSG = (
+    "previous client vanished; reopening with fresh KV cache (expected on crash recovery)"
+)
+
+_logger = logging.getLogger(__name__)
+
+
+def check_and_log_crash_recovery(
+    state: State,
+    *,
+    ttl_minutes: int = 60,
+    logger: logging.Logger | None = None,
+) -> None:
+    """Log a crash-recovery diagnostic when client_alive_at is within the TTL window.
+
+    Called on --continue before the cycle loop opens a new shared client.
+    When client_alive_at is set and within ttl_minutes, the client was likely
+    orphaned by a crash. The caller will open a fresh client regardless; this
+    function only emits the INFO diagnostic.
+
+    When client_alive_at is None or older than ttl_minutes, no log is emitted
+    (normal resume after a rate-limit pause or a completed run).
+
+    Args:
+        state: Root state to inspect (read-only).
+        ttl_minutes: Age threshold in minutes (exclusive upper bound). Default 60.
+        logger: Logger for the INFO message. Defaults to the module logger.
+    """
+    log = logger or _logger
+    if _state.is_client_presumed_alive(state, ttl_minutes=ttl_minutes):
+        log.info(CRASH_RECOVERY_LOG_MSG)