simplicio-cli 0.5.1__tar.gz → 0.5.13__tar.gz

{simplicio_cli-0.5.1/simplicio_cli.egg-info → simplicio_cli-0.5.13}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: simplicio-cli
-Version: 0.5.1
+Version: 0.5.13
 Summary: Portable task-to-code pipeline that works with any LLM. Turn a one-line task into a verified code change — diff + test + verify loop. +55 pts on a 156-check benchmark, 21% faster, ~same tokens.
 Author-email: Wesley Simplicio <wesleybob4@gmail.com>
 License: MIT
@@ -16,7 +16,6 @@ Classifier: License :: OSI Approved :: MIT License
 Classifier: Operating System :: OS Independent
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3 :: Only
-Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
@@ -24,24 +23,24 @@ Classifier: Topic :: Software Development
 Classifier: Topic :: Software Development :: Code Generators
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
-Requires-Python: >=3.9
+Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: sentence-transformers>=2.2
-Requires-Dist: numpy>=1.23
-Requires-Dist: anthropic>=0.30
-Requires-Dist: openai>=1.30
-Requires-Dist: simplicio-mapper>=0.6.1
-Requires-Dist: simplicio-prompt>=1.12.0
-Requires-Dist: httpx>=0.27
-Requires-Dist: orjson>=3.10
-Requires-Dist: diskcache>=5.6
-Requires-Dist: libcst>=1.8
+Requires-Dist: sentence-transformers>=5.1.2
+Requires-Dist: numpy>=2.0.2
+Requires-Dist: anthropic>=0.105.2
+Requires-Dist: openai>=2.38.0
+Requires-Dist: simplicio-mapper>=0.6.10
+Requires-Dist: simplicio-prompt>=1.12.3
+Requires-Dist: httpx>=0.28.1
+Requires-Dist: orjson>=3.11.5
+Requires-Dist: diskcache>=5.6.3
+Requires-Dist: libcst>=1.8.6
 Provides-Extra: bench
-Requires-Dist: fpdf2>=2.7; extra == "bench"
+Requires-Dist: fpdf2>=2.8.4; extra == "bench"
 Provides-Extra: local
-Requires-Dist: llama-cpp-python>=0.3.2; extra == "local"
-Requires-Dist: huggingface-hub>=0.23; extra == "local"
+Requires-Dist: llama-cpp-python>=0.3.23; extra == "local"
+Requires-Dist: huggingface-hub>=1.8.0; extra == "local"
 Dynamic: license-file
 
 # simplicio-cli
@@ -64,6 +63,23 @@ pip install simplicio-cli
 ```
 
 ---
+## Recommended Default Stack (Official)
+
+The recommended and supported way to use `simplicio-dev-cli` is as part of the full Simplicio execution stack:
+
+**simplicio-dev-cli + simplicio-prompt + agents/skills**
+
+- `simplicio-dev-cli`: 6-layer contract, verification loop, and core task execution.
+- `simplicio-prompt`: Subagent runtime, fan-out, behavior consensus aggregator, and advanced orchestration.
+- **Agents & Skills**: Reusable capabilities from `.skills/`, `.agents/`, and the Simplicio starter (AGENTS.md, specs-as-code, etc.).
+
+This combination is the **official default** across the Simplicio ecosystem. All benchmarks, examples, and documentation assume this full stack.
+
+See the canonical policy:
+- [docs/LLM_USAGE_POLICY.md](docs/LLM_USAGE_POLICY.md)
+
+When bootstrapping a new project with the Simplicio starter, this stack is configured by default.
+
 
 ## Why it works — the numbers
 
@@ -562,10 +578,11 @@ simplicio smoke      # prints provider config + one test call
 simplicio ships an **in-process** backend powered by
 [`llama-cpp-python`](https://github.com/abetlen/llama-cpp-python). When **no
 provider is configured** (`SIMPLICIO_MODEL` *and* `SIMPLICIO_BASE_URL` both
-unset), it runs **Qwen2.5-Coder-1.5B-Instruct (Q5_K_M GGUF)** directly — small,
-code-specialized, fast on CPU, no API key, no Ollama, no HTTP overhead. The
-6-layer contract is what makes a 1.5B usable: it lifts the same model from ~34%
-to ~88% pass-rate on the local benchmark.
+unset), it runs **Qwen3.5-2B Q6_K GGUF** directly. If that GGUF is not
+available yet, it falls back to the previous **Qwen2.5-Coder 1.5B Q8_0** and
+**Q6_K_L** executor files — small, code-specialized, fast on CPU, no API key,
+no Ollama, no HTTP overhead. The 6-layer contract is what keeps tiny local
+models usable for execution work.
 
 ```bash
 pip install 'simplicio-cli[local]'          # pulls llama-cpp-python + huggingface-hub
@@ -579,7 +596,8 @@ simplicio task "add input validation to createUser" \
 Explicit routes (override the default model/weights):
 
 ```bash
-SIMPLICIO_MODEL=local-llama/default                                  # bundled default
+SIMPLICIO_MODEL=local-llama/default                                  # Qwen3.5-2B Q6_K primary, Qwen2.5 fallbacks
+SIMPLICIO_MODEL=local-llama/bartowski/Qwen_Qwen3.5-2B-GGUF::Qwen3.5-2B-Q6_K.gguf
 SIMPLICIO_MODEL=local-llama/bartowski/Qwen2.5-Coder-7B-Instruct-GGUF::Qwen2.5-Coder-7B-Instruct-Q4_K_M.gguf
 SIMPLICIO_MODEL=local-llama//models/my-model.gguf                    # direct local path
 SIMPLICIO_LOCAL_MODEL_PATH=/models/my-model.gguf                     # always wins
@@ -622,6 +640,11 @@ API key from any of the providers above (OpenRouter is the cheapest way to
 try multiple models behind one key), set `SIMPLICIO_API_KEY` +
 `SIMPLICIO_MODEL` + optional `SIMPLICIO_BASE_URL`, run `simplicio task ...`.
 
+**"How do I load `.env.local` safely before running a local API?"** Use
+`eval "$(simplicio env-export .env.local)"` instead of `source .env.local`.
+This preserves values with semicolons, such as PostgreSQL connection strings,
+without executing the dotenv file as shell code.
+
 **"I have Codex CLI / ChatGPT Plus and don't want to pay for an API key."**
 Not auto-wired yet. Workarounds: (a) get an OpenRouter key (~$2 covers
 thousands of tasks at small-model rates), (b) wait for the shell-out provider

{simplicio_cli-0.5.1 → simplicio_cli-0.5.13}/README.md

@@ -18,6 +18,23 @@ pip install simplicio-cli
 ```
 
 ---
+## Recommended Default Stack (Official)
+
+The recommended and supported way to use `simplicio-dev-cli` is as part of the full Simplicio execution stack:
+
+**simplicio-dev-cli + simplicio-prompt + agents/skills**
+
+- `simplicio-dev-cli`: 6-layer contract, verification loop, and core task execution.
+- `simplicio-prompt`: Subagent runtime, fan-out, behavior consensus aggregator, and advanced orchestration.
+- **Agents & Skills**: Reusable capabilities from `.skills/`, `.agents/`, and the Simplicio starter (AGENTS.md, specs-as-code, etc.).
+
+This combination is the **official default** across the Simplicio ecosystem. All benchmarks, examples, and documentation assume this full stack.
+
+See the canonical policy:
+- [docs/LLM_USAGE_POLICY.md](docs/LLM_USAGE_POLICY.md)
+
+When bootstrapping a new project with the Simplicio starter, this stack is configured by default.
+
 
 ## Why it works — the numbers
 
@@ -516,10 +533,11 @@ simplicio smoke      # prints provider config + one test call
 simplicio ships an **in-process** backend powered by
 [`llama-cpp-python`](https://github.com/abetlen/llama-cpp-python). When **no
 provider is configured** (`SIMPLICIO_MODEL` *and* `SIMPLICIO_BASE_URL` both
-unset), it runs **Qwen2.5-Coder-1.5B-Instruct (Q5_K_M GGUF)** directly — small,
-code-specialized, fast on CPU, no API key, no Ollama, no HTTP overhead. The
-6-layer contract is what makes a 1.5B usable: it lifts the same model from ~34%
-to ~88% pass-rate on the local benchmark.
+unset), it runs **Qwen3.5-2B Q6_K GGUF** directly. If that GGUF is not
+available yet, it falls back to the previous **Qwen2.5-Coder 1.5B Q8_0** and
+**Q6_K_L** executor files — small, code-specialized, fast on CPU, no API key,
+no Ollama, no HTTP overhead. The 6-layer contract is what keeps tiny local
+models usable for execution work.
 
 ```bash
 pip install 'simplicio-cli[local]'          # pulls llama-cpp-python + huggingface-hub
@@ -533,7 +551,8 @@ simplicio task "add input validation to createUser" \
 Explicit routes (override the default model/weights):
 
 ```bash
-SIMPLICIO_MODEL=local-llama/default                                  # bundled default
+SIMPLICIO_MODEL=local-llama/default                                  # Qwen3.5-2B Q6_K primary, Qwen2.5 fallbacks
+SIMPLICIO_MODEL=local-llama/bartowski/Qwen_Qwen3.5-2B-GGUF::Qwen3.5-2B-Q6_K.gguf
 SIMPLICIO_MODEL=local-llama/bartowski/Qwen2.5-Coder-7B-Instruct-GGUF::Qwen2.5-Coder-7B-Instruct-Q4_K_M.gguf
 SIMPLICIO_MODEL=local-llama//models/my-model.gguf                    # direct local path
 SIMPLICIO_LOCAL_MODEL_PATH=/models/my-model.gguf                     # always wins
@@ -576,6 +595,11 @@ API key from any of the providers above (OpenRouter is the cheapest way to
 try multiple models behind one key), set `SIMPLICIO_API_KEY` +
 `SIMPLICIO_MODEL` + optional `SIMPLICIO_BASE_URL`, run `simplicio task ...`.
 
+**"How do I load `.env.local` safely before running a local API?"** Use
+`eval "$(simplicio env-export .env.local)"` instead of `source .env.local`.
+This preserves values with semicolons, such as PostgreSQL connection strings,
+without executing the dotenv file as shell code.
+
 **"I have Codex CLI / ChatGPT Plus and don't want to pay for an API key."**
 Not auto-wired yet. Workarounds: (a) get an OpenRouter key (~$2 covers
 thousands of tasks at small-model rates), (b) wait for the shell-out provider

{simplicio_cli-0.5.1 → simplicio_cli-0.5.13}/pyproject.toml

@@ -1,10 +1,10 @@
 [project]
 name = "simplicio-cli"
-version = "0.5.1"
+version = "0.5.13"
 description = "Portable task-to-code pipeline that works with any LLM. Turn a one-line task into a verified code change — diff + test + verify loop. +55 pts on a 156-check benchmark, 21% faster, ~same tokens."
 readme = "README.md"
 license = { text = "MIT" }
-requires-python = ">=3.9"
+requires-python = ">=3.10"
 authors = [{ name = "Wesley Simplicio", email = "wesleybob4@gmail.com" }]
 keywords = [
   "llm",
@@ -31,7 +31,6 @@ classifiers = [
   "Operating System :: OS Independent",
   "Programming Language :: Python :: 3",
   "Programming Language :: Python :: 3 :: Only",
-  "Programming Language :: Python :: 3.9",
   "Programming Language :: Python :: 3.10",
   "Programming Language :: Python :: 3.11",
   "Programming Language :: Python :: 3.12",
@@ -41,23 +40,23 @@ classifiers = [
   "Topic :: Scientific/Engineering :: Artificial Intelligence",
 ]
 dependencies = [
-  "sentence-transformers>=2.2",
-  "numpy>=1.23",
-  "anthropic>=0.30",
-  "openai>=1.30",
-  "simplicio-mapper>=0.6.1",
-  "simplicio-prompt>=1.12.0",
-  "httpx>=0.27",
-  "orjson>=3.10",
-  "diskcache>=5.6",
-  "libcst>=1.8",
+  "sentence-transformers>=5.1.2",
+  "numpy>=2.0.2",
+  "anthropic>=0.105.2",
+  "openai>=2.38.0",
+  "simplicio-mapper>=0.6.10",
+  "simplicio-prompt>=1.12.3",
+  "httpx>=0.28.1",
+  "orjson>=3.11.5",
+  "diskcache>=5.6.3",
+  "libcst>=1.8.6",
 ]
 
 [project.optional-dependencies]
-bench = ["fpdf2>=2.7"]
+bench = ["fpdf2>=2.8.4"]
 # Offline-first in-process inference (Path 4). Pulls the llama.cpp Python
-# bindings plus huggingface-hub to fetch the default Qwen2.5-Coder-1.5B GGUF.
-local = ["llama-cpp-python>=0.3.2", "huggingface-hub>=0.23"]
+# bindings plus huggingface-hub to fetch the default Qwen3.5-2B Q6_K GGUF.
+local = ["llama-cpp-python>=0.3.23", "huggingface-hub>=1.8.0"]
 
 [project.urls]
 Homepage = "https://github.com/wesleysimplicio/simplicio-cli"

simplicio_cli-0.5.13/simplicio/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.5.13"

{simplicio_cli-0.5.1 → simplicio_cli-0.5.13}/simplicio/bench.py

@@ -13,10 +13,12 @@ import os, json, time, subprocess
 from .observability import estimate_tokens, log_run
 from .prompt import build_prompt
 from .providers import generate
+from .runtime_env import wrap_project_command
 
 def _test(output, root, test_cmd):
     os.makedirs(os.path.join(root, ".simplicio"), exist_ok=True)
     open(os.path.join(root, ".simplicio/bench_out.txt"), "w").write(output or "")
+    test_cmd = wrap_project_command(root, test_cmd)
     p = subprocess.run(test_cmd, shell=True, cwd=root, capture_output=True, text=True)
     return p.returncode == 0
 

{simplicio_cli-0.5.1 → simplicio_cli-0.5.13}/simplicio/cli.py

@@ -91,7 +91,7 @@ def _add_task_args(p: argparse.ArgumentParser, *, target_required: bool) -> None
     p.add_argument(
         "--local",
         action="store_true",
-        help="force the in-process local model (Qwen2.5-Coder-1.5B GGUF, "
+        help="force the in-process local model (Qwen3.5-2B Q6_K GGUF, "
         "no API key); overrides SIMPLICIO_MODEL/SIMPLICIO_BASE_URL",
     )
 
@@ -128,7 +128,7 @@ def _force_local_if_requested(a: argparse.Namespace) -> None:
 
 
 def _run_task_command(a: argparse.Namespace) -> int:
-    from .pipeline import run, run_task
+    from .pipeline import run_task
 
     _force_local_if_requested(a)
     stack = a.stack or "angular"
@@ -152,7 +152,7 @@ def _run_task_command(a: argparse.Namespace) -> int:
             for warning in result["warnings"]:
                 print(f"warning: {warning}", file=sys.stderr)
         return 0 if (a.dry_run_task or result["applied"]) else 1
-    run(
+    result = run_task(
         a.root,
         stack,
         a.goal,
@@ -161,7 +161,11 @@ def _run_task_command(a: argparse.Namespace) -> int:
         a.constraints,
         bound_paths=a.bound_paths,
     )
-    return 0
+    status = "DONE" if result["applied"] else "FAILED"
+    print(f"{status}: {result['diff_summary']}")
+    for warning in result["warnings"]:
+        print(f"warning: {warning}", file=sys.stderr)
+    return 0 if result["applied"] else 1
 
 
 def _first_file_signal(signals: list[str]) -> str | None:
@@ -177,6 +181,8 @@ def _run_scratch_command(a: argparse.Namespace) -> int:
     scratch_argv = [a.goal]
     if a.stack:
         scratch_argv += ["--stack", a.stack]
+    if a.root:
+        scratch_argv += ["--root", a.root]
     if a.name:
         scratch_argv += ["--name", a.name]
     if a.dest:
@@ -560,6 +566,7 @@ def main(argv=None):
     sub = ap.add_subparsers(dest="cmd", required=True)
 
     pi = sub.add_parser("index", help="index/cache the repo (once, or after changes)")
+    pi.add_argument("root_arg", nargs="?", help="project root; same as --root")
     pi.add_argument("--root", default=".")
     pi.add_argument("--stack", default="angular")
 
@@ -592,6 +599,7 @@ def main(argv=None):
     p_init.add_argument("--dry-run", action="store_true")
 
     p_det = sub.add_parser("detect", help="heuristic: is a prompt a code-edit task")
+    p_det.add_argument("prompt_words", nargs="*", help=argparse.SUPPRESS)
     p_det.add_argument("--prompt", help="prompt text (default: read from stdin)")
     p_det.add_argument("--quiet", action="store_true")
     p_det.add_argument("--json", action="store_true")
@@ -600,12 +608,19 @@ def main(argv=None):
     p_status.add_argument("--root", default=".")
     p_status.add_argument("--json", action="store_true")
 
+    p_env_export = sub.add_parser(
+        "env-export",
+        help="print shell-safe exports from a dotenv file without sourcing it",
+    )
+    p_env_export.add_argument("env_file")
+    p_env_export.add_argument("--json", action="store_true")
+
     a = ap.parse_args(argv)
     maybe_autoinstall(a.cmd)
     if a.cmd == "index":
         from .precedent import index_repo
 
-        index_repo(a.root, a.stack)
+        index_repo(a.root_arg or a.root, a.stack)
     elif a.cmd == "smoke":
         from .providers import generate, info
 
@@ -650,8 +665,11 @@ def main(argv=None):
         from .detect import main as detect_main
 
         detect_argv = []
-        if a.prompt is not None:
-            detect_argv += ["--prompt", a.prompt]
+        prompt = a.prompt
+        if prompt is None and a.prompt_words:
+            prompt = " ".join(a.prompt_words)
+        if prompt is not None:
+            detect_argv += ["--prompt", prompt]
         if a.quiet:
             detect_argv += ["--quiet"]
         if a.json:
@@ -659,6 +677,22 @@ def main(argv=None):
         return detect_main(detect_argv)
     elif a.cmd == "status":
         return _run_status_command(a)
+    elif a.cmd == "env-export":
+        from .runtime_env import parse_env_file, shell_export_lines
+
+        try:
+            values = parse_env_file(a.env_file)
+        except OSError as exc:
+            print(f"simplicio env-export: {exc}", file=sys.stderr)
+            return 2
+        except ValueError as exc:
+            print(f"simplicio env-export: {exc}", file=sys.stderr)
+            return 2
+        if a.json:
+            print(json.dumps(values, sort_keys=True))
+        else:
+            print("\n".join(shell_export_lines(values)))
+        return 0
     elif a.cmd == "task":
         return _run_task_command(a)
     elif a.cmd == "run":

{simplicio_cli-0.5.1 → simplicio_cli-0.5.13}/simplicio/detect.py

@@ -13,17 +13,25 @@ import sys
 from dataclasses import dataclass
 from pathlib import Path
 
+from .intent import classify_goal
+
 _EDIT_VERBS = (
     # English
     "add", "remove", "delete", "rename", "refactor", "fix", "patch", "update",
     "change", "replace", "hide", "show", "validate", "implement", "wire", "inject",
     "extract", "split", "rewrite", "tweak", "adjust", "introduce", "expose",
+    "map", "inventory", "align", "document", "improve", "use", "run",
+    "execute", "connect", "test", "prove", "evidence",
     # Portuguese
     "adicione", "adicionar", "remova", "remover", "renomeie", "renomear",
     "corrija", "corrigir", "atualize", "atualizar", "altere", "alterar",
     "esconda", "esconder", "mostre", "mostrar", "valide", "validar",
     "implemente", "implementar", "troque", "trocar", "ajuste", "ajustar",
-    "ocultar", "exiba", "exibir", "criar", "crie",
+    "ocultar", "exiba", "exibir", "criar", "crie", "veja", "ver",
+    "alinhe", "alinhar", "mapeie", "mapear", "documente", "documentar",
+    "melhore", "melhorar", "use", "usar", "rode", "rodar", "execute",
+    "executar", "teste", "testar", "conecte", "conectar", "prove", "provar",
+    "evidencie", "evidenciar",
 )
 
 _FILE_EXT_RE = re.compile(
@@ -38,8 +46,9 @@ _CODE_NOUNS = (
     "spec", "validator", "selector", "store", "reducer", "action", "hook",
     "directive", "pipe", "module", "function", "method", "class", "prop",
     "field", "column", "button", "form", "input", "dropdown", "modal", "page",
+    "api", "apis", "database", "postgres", "postgresql", "playwright", "e2e",
     "componente", "função", "funcao", "classe", "tela", "rota", "campo",
-    "botão", "botao", "formulário", "formulario",
+    "botão", "botao", "formulário", "formulario", "banco", "dados",
 )
 
 _NEGATIVE_CUES = (
@@ -55,6 +64,7 @@ class DetectResult:
     score: int
     signals: list
     hint: str
+    scope: str = "task"
 
 
 def detect(prompt: str) -> DetectResult:
@@ -89,21 +99,43 @@ def detect(prompt: str) -> DetectResult:
         score += 5
         signals.append("explicit_invocation")
 
+    intent = classify_goal(prompt)
+    if intent.scope in {"feature", "sprint"} and intent.confidence >= 0.70:
+        score += 2
+        for signal in intent.signals:
+            if signal.startswith(("feature:", "sprint:")) and signal not in signals:
+                signals.append(signal)
+
     is_code = score >= 3
-    hint = _render_hint(prompt, signals) if is_code else ""
-    return DetectResult(is_code, score, signals, hint)
+    hint = _render_hint(prompt, signals, intent.scope) if is_code else ""
+    return DetectResult(is_code, score, signals, hint, intent.scope)
 
 
-def _render_hint(prompt: str, signals: list) -> str:
+def _render_hint(prompt: str, signals: list, scope: str = "task") -> str:
     target_hint = next((s.split(":", 1)[1] for s in signals if s.startswith("file:")), None)
     target_line = f"target = {target_hint}" if target_hint else "target = <ask user or infer via Explore>"
+    if scope == "sprint":
+        scale_line = (
+            "This prompt looks like sprint-scale code work. Use `simplicio run --scope sprint`\n"
+            "or a local sprint plan before editing by hand, then verify each slice."
+        )
+    elif scope == "feature":
+        scale_line = (
+            "This prompt looks like feature-scale code work. Use `simplicio run --scope feature`\n"
+            "or a focused local plan before editing by hand, then verify the full flow."
+        )
+    else:
+        scale_line = (
+            "This prompt looks like a small/medium code edit. Before editing by hand,\n"
+            "invoke the simplicio-cli skill (it stacks precedent + skill-router + 6-layer\n"
+            "prompt + test + verify-loop and measurably boosts pass-rate)."
+        )
     return (
         "[SIMPLICIO_PROMPT_HINT]\n"
-        "This prompt looks like a small/medium code edit. Before editing by hand,\n"
-        "invoke the simplicio-cli skill (it stacks precedent + skill-router + 6-layer\n"
-        "prompt + test + verify-loop and measurably boosts pass-rate).\n"
+        f"{scale_line}\n"
         f"  goal = {prompt.strip()[:120]}\n"
         f"  {target_line}\n"
+        f"  scope = {scope}\n"
         f"  signals = {', '.join(signals)}\n"
         "[/SIMPLICIO_PROMPT_HINT]"
     )
@@ -126,6 +158,7 @@ def main(argv=None) -> int:
         print(json.dumps({
             "is_code_task": result.is_code_task,
             "score": result.score,
+            "scope": result.scope,
             "signals": result.signals,
         }))
 

{simplicio_cli-0.5.1 → simplicio_cli-0.5.13}/simplicio/intent.py

@@ -59,6 +59,11 @@ _TASK_VERBS = (
     "tweak",
     "adjust",
     "introduce",
+    "map",
+    "inventory",
+    "align",
+    "document",
+    "improve",
     "corrija",
     "corrigir",
     "adicione",
@@ -75,6 +80,16 @@ _TASK_VERBS = (
     "ajustar",
     "crie",
     "criar",
+    "veja",
+    "ver",
+    "alinhe",
+    "alinhar",
+    "mapeie",
+    "mapear",
+    "documente",
+    "documentar",
+    "melhore",
+    "melhorar",
 )
 _TASK_NOUNS = (
     "component",
@@ -167,6 +182,12 @@ _BROAD_CUES = (
     "varios arquivos",
     "varios modulos",
 )
+_REAL_STACK_EVIDENCE_CUES = {
+    "browser": ("playwright", "e2e"),
+    "ui": ("web", "frontend", "telas", "screens", "interface", "browser"),
+    "api": ("api", "backend", "azure functions", ".net"),
+    "database": ("postgres", "postgresql", "database", "banco", "db"),
+}
 
 _SPRINT_CUES = (
     "sprint",
@@ -180,11 +201,27 @@ _SPRINT_CUES = (
     ".specs/sprints",
     "all issues",
     "all tasks",
+    "all endpoints",
+    "all screens",
     "every issue",
     "entire repo",
+    "cross-repo",
     "todas as issues",
     "todos os issues",
     "todas as tarefas",
+    "todos os endpoints",
+    "todas as telas",
+    "todos os projetos",
+    "remaining endpoints",
+    "remaining routes",
+    "missing endpoints",
+    "missing routes",
+    "endpoints restantes",
+    "rotas restantes",
+    "lacunas restantes",
+    "endpoints faltantes",
+    "rotas faltantes",
+    "lacunas faltantes",
 )
 _FINISH_CUES = ("finish", "close", "complete", "ship", "terminar", "fechar", "concluir", "finalizar")
 
@@ -299,6 +336,10 @@ def classify_goal(text: str, explicit_scope: str | None = None) -> IntentResult:
         scores["feature"] += 2
         signals.append(f"broad:{broad}")
 
+    if _looks_like_real_stack_evidence(lower):
+        scores["feature"] += 5
+        signals.append("feature:real_stack_evidence")
+
     if not unique_files and (feature_verb or feature_noun):
         scores["feature"] += 1
         signals.append("feature:no_target_file")
@@ -308,6 +349,39 @@ def classify_goal(text: str, explicit_scope: str | None = None) -> IntentResult:
         scores["sprint"] += 4
         signals.append(f"sprint:{sprint}")
 
+    if any(
+        phrase in lower
+        for phrase in (
+            "todos os endpoints",
+            "todas as telas",
+            "todos os projetos",
+            "all endpoints",
+            "all screens",
+            "web com api",
+            "api e ai-agents",
+        )
+    ):
+        scores["sprint"] += 2
+        signals.append("sprint:full_inventory")
+
+    if any(
+        phrase in lower
+        for phrase in (
+            "remaining endpoints",
+            "remaining routes",
+            "missing endpoints",
+            "missing routes",
+            "endpoints restantes",
+            "rotas restantes",
+            "lacunas restantes",
+            "endpoints faltantes",
+            "rotas faltantes",
+            "lacunas faltantes",
+        )
+    ):
+        scores["sprint"] += 5
+        signals.append("sprint:remaining_endpoint_inventory")
+
     issue_refs = _ISSUE_RE.findall(text)
     if len(issue_refs) >= 3:
         scores["sprint"] += 3
@@ -388,3 +462,7 @@ def _first_token(text: str, words: tuple[str, ...]) -> str | None:
 
 def _first_phrase(text: str, phrases: tuple[str, ...]) -> str | None:
     return next((phrase for phrase in phrases if phrase in text), None)
+
+
+def _looks_like_real_stack_evidence(text: str) -> bool:
+    return all(any(cue in text for cue in cues) for cues in _REAL_STACK_EVIDENCE_CUES.values())

{simplicio_cli-0.5.1 → simplicio_cli-0.5.13}/simplicio/orchestrator/feature.py

@@ -7,9 +7,11 @@ existing verify-loop, and a failing task can trigger one bounded replan.
 
 from __future__ import annotations
 
+import os
 from pathlib import Path
 from typing import Callable
 
+from ..scratch.codegen import try_execute
 from ..scratch._pipeline_adapter import run_task as run_plan_task
 from ..scratch.planner import generate_plan
 from ..scratch.stack_registry import StackRegistry, slugify_project
@@ -19,6 +21,32 @@ from .cost_governor import BudgetExceeded, provider_budget
 TaskRunner = Callable[..., tuple[bool, str]]
 
 
+def _codegen_disabled() -> bool:
+    return os.environ.get("SIMPLICIO_DISABLE_CODEGEN", "").lower() in {
+        "1",
+        "true",
+        "yes",
+        "on",
+    }
+
+
+def _run_feature_task(task, project_dir: Path, stack, *, quiet: bool = False):
+    """Run feature tasks through deterministic codegen before the LLM pipeline."""
+    codegen_log = ""
+    if not _codegen_disabled():
+        codegen_result = try_execute(task, project_dir, stack)
+        if codegen_result is not None:
+            codegen_log = codegen_result.log
+            if codegen_result.passed or not codegen_result.fallback_to_llm:
+                mode = codegen_result.executor_name or "codegen"
+                return codegen_result.passed, f"codegen:{mode}: {codegen_log}"
+
+    passed, log = run_plan_task(task, project_dir, stack, quiet=quiet)
+    if codegen_log:
+        log = f"codegen fallback: {codegen_log}\n\n{log}"
+    return passed, log
+
+
 def _ordered_tasks(tasks: list[object]) -> list[object]:
     seen_ids: set[str] = set()
     for task in tasks:
@@ -61,7 +89,8 @@ def run_feature(
     if max_iter < 0:
         raise ValueError("max_iter must be >= 0")
     planner_fn = planner or generate_plan
-    task_runner_fn = task_runner or run_plan_task
+    default_task_runner = task_runner is None
+    task_runner_fn = task_runner or _run_feature_task
 
     reg = StackRegistry()
     stack = reg.get(stack_slug)
@@ -101,7 +130,7 @@ def run_feature(
                 for task in planned_tasks:
                     if task.id in completed_task_ids:
                         continue
-                    if task_runner_fn is run_plan_task:
+                    if default_task_runner:
                         passed, log = task_runner_fn(
                             task,
                             Path(root),