Tests unitarios: 51 tests para compactor, key_data, fingerprint y costes

- tests/test_compactor.py: 24 tests (estimate_tokens, extract_facts,
  build_summary, summarize_tool_output, compact_sections)
- tests/test_key_data_extraction.py: 11 tests (extracción de tables,
  records, sections, modules, pages desde tool executions)
- tests/test_fingerprint.py: 8 tests (deduplicación MD5, sort_keys,
  nested args)
- tests/test_cost_calculation.py: 8 tests (pricing formula, custom
  pricing, rounding)
- README.md: sección Tests con instrucciones de ejecución

Todos offline, sin Docker/Redis/LLM. Ejecutar: python3 -m pytest tests/ -v

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Dockerfile

@@ -26,6 +26,12 @@ RUN pip install --no-cache-dir -r requirements.txt
 COPY agenticSystem/mcp-server/package.json agenticSystem/mcp-server/package-lock.json* ./mcp-server/
 RUN cd mcp-server && npm install --production
 
+# Instalar dependencias de sistema de Playwright (como root)
+RUN cd mcp-server && npx playwright install-deps chromium
+
+# Instalar uv (incluye uvx) para mcp-server-fetch
+RUN pip install --no-cache-dir uv
+
 # Copiar codigo fuente del MCP server
 COPY agenticSystem/mcp-server/ ./mcp-server/
 
@@ -45,6 +51,9 @@ RUN useradd -m appuser \
     && chown -R appuser:appuser /opt/acai/webs
 USER appuser
 
+# Descargar Chromium como appuser (queda en ~/.cache/ms-playwright/)
+RUN cd mcp-server && npx playwright install chromium
+
 EXPOSE 8000
 
 CMD ["uvicorn", "src.main:app", "--host", "0.0.0.0", "--port", "8000"]

README.md

@@ -59,6 +59,22 @@ python3 -m uvicorn src.main:app --reload --port 8001
 # 5. Dashboard en http://localhost:8001/dashboard/
 ```
 
+### Tests
+
+```bash
+# Ejecutar todos los tests unitarios (no necesita Docker, Redis ni LLM)
+pip install pytest
+python3 -m pytest tests/ -v
+
+# Ejecutar un archivo específico
+python3 -m pytest tests/test_compactor.py -v
+
+# Ejecutar un test específico
+python3 -m pytest tests/test_cost_calculation.py::TestCostCalculation::test_1m_input_tokens -v
+```
+
+Los tests validan: compactación de contexto, extracción de key_data para historial, fingerprinting de tool calls, y cálculo de costes. Son 100% offline — no consumen tokens ni necesitan servicios externos.
+
 ### Cargar Knowledge Base
 
 ```bash

docs/ACAI-CLAUDE.md

@@ -105,7 +105,7 @@ Do NOT modify web-base files — they are shared across all projects.
 1. **Before working with any area (hooks, modules, templates, CSS/JS, etc.), read the corresponding documentation in `docs/` first.** Do not guess or assume — always consult the docs before taking action.
 2. **NEVER use `mkdir` to create directories.** Instead, use the `Write` tool to create the first file inside the directory — this creates parent directories automatically. For example, to create a new module, directly write the `index-base.tpl` file.
 3. Only edit `index-base.tpl` in modules — `index.tpl`, `index-twig.tpl`, and `builder.json` are auto-generated
-3. **After editing any `index-base.tpl`, ALWAYS call the `compile_module` MCP tool** to compile the module/section. This is mandatory — without compilation, changes won't take effect in the CMS.
+3. **Edit `index-base.tpl` using `acai_write` or `acai_line_replace`** — the server compiles automatically when the file is saved. No need to call `compile_module` manually.
 4. Use Twig **filters** (with `|`), never Twig functions
 5. Table names without `cms_` prefix everywhere
 6. Primary key is `num`, never `id`
@@ -122,8 +122,8 @@ This project has MCP tools for managing modules, records, media, and more. **Bef
 See [docs/mcp-tools-reference.md](docs/mcp-tools-reference.md) for the complete list of available tools and step-by-step workflows.
 
 Key workflows:
-- **Create module**: Read [docs/module-creation-guide.md](docs/module-creation-guide.md) first → `create_module` → `add_module_to_record` (returns sectionId) → `set_module_config_vars` (returns uploadFields) → images via uploadFields
-- **Edit module**: read vars → edit `index-base.tpl` → `compile_module`
+- **Create module**: Read [docs/module-creation-guide.md](docs/module-creation-guide.md) first → Write `index-base.tpl` via `acai_write` → `add_module_to_record` (returns sectionId) → `set_module_config_vars` (returns uploadFields) → images via uploadFields
+- **Edit module**: read vars → edit `index-base.tpl` with `acai_write` or `acai_line_replace` (server compiles automatically)
 - **Add images**: use `uploadFields` from `set_module_config_vars` response → `upload_record_image`
 - **Generate images**: `generate_image` → `upload_record_image` with returned URL
 

docs/mcp-tools-reference.md

@@ -4,7 +4,7 @@
 
 | Tool | Categoría | Acción |
 |------|-----------|--------|
-| `create_module` | Módulos | Crea módulo nuevo (directorio + archivos + compila) |
+| `create_module` | Módulos | (Legacy) Alternativa para crear módulo — preferir acai_write |
 | `compile_module` | Módulos | Compila módulo tras editar index-base.tpl |
 | `check_module` | Módulos | Preview de cómo renderiza un módulo |
 | `check_module_usage` | Módulos | Qué páginas usan un módulo |
@@ -37,7 +37,7 @@
 
 ### Crear un módulo nuevo desde cero
 
-1. `create_module` — Crea el directorio con index-base.tpl, style.css, script.js y compila
+1. `acai_write` — Escribe `index-base.tpl` en `template/estandar/modulos/NOMBRE/`. El server crea la carpeta si no existe, compila y genera todos los archivos derivados (index-twig.tpl, index.tpl, builder.json, screenshots)
 2. `add_module_to_record` — Añade el módulo a una página (tabla padre, ej: `apartados`)
 3. `set_module_config_vars` — Rellena las variables con contenido (textos, colores, opciones). **OBLIGATORIO** — sin esto el módulo no muestra nada. Devuelve:
    - `configVars`: mapa de variables → recordNums
@@ -46,12 +46,13 @@
 4. Para imágenes: `generate_image` o `upload_record_image` usando el `recordNum` y `fieldName` del `uploadFields` devuelto en el paso 3
 5. Verificar con `check_module` o recargando la página
 
+> **Nota:** `create_module` es una alternativa legacy que hace lo mismo pero con menos control sobre el contenido del template.
+
 ### Editar un módulo existente
 
 1. `get_module_config_vars` — Leer el estado actual del módulo (variables, recordNums)
-2. Editar `index-base.tpl` con la tool `Write` o `Edit`
-3. `compile_module` — **OBLIGATORIO** tras cada edición de index-base.tpl
-4. Si cambias variables: `set_module_config_vars` para actualizar valores
+2. Editar `index-base.tpl` con `acai_write` o `acai_line_replace` — el server compila automáticamente al guardar
+3. Si cambias variables: `set_module_config_vars` para actualizar valores
 
 ### Añadir/modificar imágenes de un módulo
 

docs/module-creation-guide.md

@@ -38,7 +38,7 @@ Each module lives in `template/estandar/modulos/<moduleId>/` with:
 ## Creating a Module — Full Workflow
 
 1. **Read style reference** (steps above)
-2. **`create_module`** — Creates the directory with index-base.tpl, style.css, script.js and compiles. Use descriptive `moduleId` and clear `label`.
+2. **`acai_write`** — Write `index-base.tpl` to `template/estandar/modulos/MODULE_ID/index-base.tpl`. The server automatically creates the directory, compiles and generates all derived files. `create_module` is a legacy alternative.
 3. **`add_module_to_record`** — Adds the module to a page. Response includes `sectionId` — use it directly in the next step.
 4. **`set_module_config_vars`** — Fill variables with content. Response includes `uploadFields` with `{ fieldName, recordNum }` for each upload variable.
 5. **Upload images** — Use `generate_image` then `upload_record_image` with the `recordNum` and `fieldName` from step 4's `uploadFields`. No need to read builder.json or call get_module_config_vars.
@@ -72,6 +72,6 @@ Modules with `MJMLModule: true` in their schema are email modules:
 - Use `section_id` variable for unique anchors/scoping
 - Use `interno` variable to detect CMS editor vs public view
 - Include other modules with: `<module_id :param1="value1"></module_id>`
-- After editing `index-base.tpl`, ALWAYS call `compile_module`
+- After editing `index-base.tpl` with `acai_write` or `acai_line_replace`, the server compiles automatically — no need to call `compile_module`
 - Twig uses filters (with `|`), never functions
 - Twig concatenation uses `~`: `'value=' ~ variable`

mcp-server/stdio.js

@@ -22,9 +22,21 @@ registerResources(server);
 
 // Static env vars (web_url and website don't change, token does)
 const projectDir = process.env.ACAI_PROJECT_DIR || "";
-const website = process.env.ACAI_WEBSITE || "";
-const webUrl = process.env.ACAI_WEB_URL || "";
+const acaiFilePath = projectDir ? path.join(projectDir, ".acai") : "";
+
+// Read .acai once at startup for URL fallbacks
+let acaiFileData = {};
+if (acaiFilePath) {
+    try {
+        acaiFileData = JSON.parse(fs.readFileSync(acaiFilePath, "utf-8"));
+    } catch { /* ignore - fall back to env vars */ }
+}
+
+const website = process.env.ACAI_WEBSITE || acaiFileData.domain || "";
+const webUrl = process.env.ACAI_WEB_URL || acaiFileData.local_web_url || "";
 const derivedForgeHost = (() => {
+    // First check .acai for explicit forge host
+    if (acaiFileData.local_forge_host) return acaiFileData.local_forge_host;
     if (!webUrl) return "";
     try {
         const parsed = new URL(webUrl);
@@ -35,7 +47,6 @@ const derivedForgeHost = (() => {
 })();
 const apiWebUrl = process.env.ACAI_API_WEB_URL || (derivedForgeHost ? "http://web:80/" : webUrl);
 const forgeHost = process.env.ACAI_FORGE_HOST || derivedForgeHost;
-const acaiFilePath = projectDir ? path.join(projectDir, ".acai") : "";
 
 // Read fresh credentials from .acai file
 function readFreshCredentials() {

mcp.json

@@ -3,13 +3,15 @@
     "acai-code": {
       "command": "node",
       "args": ["mcp-server/stdio.js"],
-      "env": {},
+      "env": {
+        "LOCAL_SERVER_URL": "http://app:9091"
+      },
       "timeout": 30,
       "startup_timeout": 10
     },
     "playwright": {
       "command": "npx",
-      "args": ["@playwright/mcp", "--headless"],
+      "args": ["@playwright/mcp", "--headless", "--executable-path", "/home/appuser/.cache/ms-playwright/chromium-1212/chrome-linux64/chrome"],
       "timeout": 30,
       "startup_timeout": 15
     },

src/adapters/openai_adapter.py

@@ -44,6 +44,7 @@ class OpenAIAdapter(ModelAdapter):
             "temperature": config.temperature,
             "messages": messages,
             "stream": True,
+            "stream_options": {"include_usage": True},
         }
         if tools:
             kwargs["tools"] = self._format_tools(tools)
@@ -52,9 +53,22 @@ class OpenAIAdapter(ModelAdapter):
 
         tool_calls_acc: dict[int, dict[str, str]] = {}
 
+        final_usage: dict[str, int] = {}
+
         async for chunk in stream:
+            # With include_usage, the last chunk has usage but no choices
+            if chunk.usage:
+                final_usage = {
+                    "input_tokens": chunk.usage.prompt_tokens or 0,
+                    "output_tokens": chunk.usage.completion_tokens or 0,
+                }
+
             choice = chunk.choices[0] if chunk.choices else None
             if not choice:
+                # Usage-only chunk (last one with include_usage) — emit it
+                if final_usage:
+                    yield StreamChunk(usage=final_usage)
+                    final_usage = {}  # Only emit once
                 continue
 
             delta = choice.delta
@@ -99,16 +113,15 @@ class OpenAIAdapter(ModelAdapter):
                             tool_arguments=acc["arguments"],
                             finish_reason="tool_use",
                         )
+                    # Emit usage after tool_use chunks
+                    if final_usage:
+                        yield StreamChunk(usage=final_usage)
                 else:
                     yield StreamChunk(
                         finish_reason="end_turn"
                         if choice.finish_reason == "stop"
                         else choice.finish_reason,
-                        usage={
-                            "output_tokens": chunk.usage.completion_tokens
-                            if chunk.usage
-                            else 0
-                        },
+                        usage=final_usage,
                     )
 
     # ------------------------------------------------------------------

src/config.py

@@ -48,6 +48,10 @@ class Settings(BaseSettings):
     mcp_timeout_seconds: float = 30.0
     mcp_startup_timeout_seconds: float = 10.0
 
+    # --- Pricing (per 1M tokens) ---
+    cost_per_1m_input: float = 2.50
+    cost_per_1m_output: float = 15.00
+
     # --- Orchestrator ---
     max_execution_steps: int = 25
     subagent_max_steps: int = 10

src/context/engine.py

@@ -62,10 +62,15 @@ class ContextEngine:
         session: SessionState,
         agent: AgentProfile,
         artifacts: list[ArtifactSummary] | None = None,
-        working_items: list[dict[str, Any]] | None = None,
+        conversation: list[dict[str, Any]] | None = None,
         extra_instructions: str = "",
     ) -> ContextPackage:
-        """Build a full ContextPackage for the given agent and session."""
+        """Build a full ContextPackage for the given agent and session.
+
+        The conversation parameter contains real assistant/tool messages
+        with complete tool results. These go into the messages array,
+        not the system prompt — like professional agentic tools.
+        """
 
         sections: list[ContextSection] = []
         allowed = set(agent.context_sections)
@@ -88,28 +93,18 @@ class ContextEngine:
         if "task_state" in allowed and session.task_history:
             sections.append(self._build_task_history(session))
 
-        # 5. Task state — current task
+        # 5. Task state — current task (includes compacted previous steps)
         if "task_state" in allowed and session.current_task:
             sections.append(self._build_task_state(session.current_task))
 
-        # 6. Artifact memory — summarised, never raw (only current task's)
-        if "artifact_memory" in allowed and artifacts:
-            sections.append(self._build_artifact_memory(artifacts))
-
-        # 6. Working context — recent relevant items
-        if "working_context" in allowed:
-            sections.append(
-                self._build_working_context(working_items or [], extra_instructions)
-            )
-
         # Compact to fit budget
         sections = self.compactor.compact_sections(sections)
 
         # Assemble system prompt from sections
         system_prompt = self._assemble_system_prompt(sections)
 
-        # Build messages (just user message — no chat history)
-        messages = self._build_messages(session)
+        # Build messages with real conversation history
+        messages = self._build_messages(session, conversation)
 
         total_tokens = estimate_tokens(system_prompt) + sum(
             estimate_tokens(m.get("content", "")) for m in messages
@@ -133,6 +128,7 @@ class ContextEngine:
                 "preview": s.content[:150].replace("\n", " "),
             })
 
+        conv_len = len(conversation) if conversation else 0
         debug_entry = {
             "timestamp": time.time(),
             "agent": agent.role.value,
@@ -144,7 +140,7 @@ class ContextEngine:
             "system_prompt_tokens": estimate_tokens(system_prompt),
             "user_message_preview": messages[0]["content"][:200] if messages else "",
             "artifacts_count": len(artifacts) if artifacts else 0,
-            "working_items_count": len(working_items) if working_items else 0,
+            "conversation_messages": conv_len,
         }
 
         history = self._history[session.session_id]
@@ -153,18 +149,13 @@ class ContextEngine:
             self._history[session.session_id] = history[-self._max_history:]
 
         logger.info(
-            "Context built for [%s/%s] — %d sections, ~%d tokens, artifacts=%d, working_items=%d",
+            "Context built for [%s/%s] — %d sections, ~%d tokens, artifacts=%d, conversation=%d msgs",
             session.session_id[:8],
             agent.role.value,
             len(sections),
             total_tokens,
             len(artifacts) if artifacts else 0,
-            len(working_items) if working_items else 0,
-        )
-        for s in section_summary:
-            logger.debug(
-                "  Section [%s] prio=%d tokens=%d chars=%d",
-                s["type"], s["priority"], s["tokens"], s["chars"],
+            conv_len,
         )
 
         return package
@@ -236,10 +227,11 @@ class ContextEngine:
             [
                 "",
                 "## Contrato de Contexto",
-                "- NUNCA recibirás salidas crudas de herramientas en tu contexto.",
-                "- Los resultados de herramientas se resumen como artefactos.",
-                "- Solicita rehidratación si necesitas el contenido completo.",
+                "- Los resultados de herramientas se incluyen completos en la conversación.",
+                "- Los steps anteriores pueden estar compactados como resúmenes.",
                 "- Mantén las respuestas enfocadas en el paso actual.",
+                "- Si ya tienes la información necesaria, genera tu respuesta final.",
+                "- NO repitas llamadas a herramientas con los mismos argumentos.",
                 "- Responde SIEMPRE en español.",
             ]
         )
@@ -406,6 +398,20 @@ class ContextEngine:
                 lines.append(f"  Result: {summary}")
             if facts:
                 lines.append(f"  Facts: {'; '.join(facts[:5])}")
+            # Key structured data (recordNums, sectionIds, etc.)
+            key_data = entry.get("key_data", {})
+            if key_data:
+                kd_parts = []
+                for table, nums in key_data.get("tables", {}).items():
+                    kd_parts.append(f"{table}: records {nums}")
+                for page, num in key_data.get("pages", {}).items():
+                    kd_parts.append(f"page '{page}' = record {num}")
+                if key_data.get("sections"):
+                    kd_parts.append(f"sections: {key_data['sections']}")
+                if key_data.get("modules"):
+                    kd_parts.append(f"modules: {key_data['modules']}")
+                if kd_parts:
+                    lines.append(f"  Key data: {'; '.join(kd_parts)}")
             review = entry.get("review", "")
             if review:
                 lines.append(f"  Review: {review[:100]}")
@@ -451,6 +457,14 @@ class ContextEngine:
             for c in task.constraints:
                 lines.append(f"- {c}")
 
+        # Show compacted previous steps results
+        compacted_steps = [s for s in task.plan if s.compacted and s.result_summary]
+        if compacted_steps:
+            lines.append("")
+            lines.append("## Previous Steps (compacted)")
+            for step in compacted_steps:
+                lines.append(f"- [{step.agent_role}] {step.description}: {step.result_summary[:300]}")
+
         # Show plan overview (compact)
         if task.plan:
             lines.append("")
@@ -458,8 +472,9 @@ class ContextEngine:
             for i, step in enumerate(task.plan):
                 marker = "→" if i == task.current_step_index else "·"
                 status_label = step.status.value
+                compacted_label = " (compacted)" if step.compacted else ""
                 lines.append(
-                    f"  {marker} Step {i + 1} [{status_label}]: {step.description}"
+                    f"  {marker} Step {i + 1} [{status_label}{compacted_label}]: {step.description}"
                 )
 
         content = "\n".join(lines)
@@ -483,26 +498,6 @@ class ContextEngine:
             token_estimate=estimate_tokens(content),
         )
 
-    def _build_working_context(
-        self,
-        items: list[dict[str, Any]],
-        extra_instructions: str,
-    ) -> ContextSection:
-        lines = ["# Working Context"]
-        if extra_instructions:
-            lines.append(f"\n{extra_instructions}")
-        for item in items[: settings.working_context_max_items]:
-            role = item.get("role", "info")
-            content_val = item.get("content", "")
-            lines.append(f"[{role}] {content_val}")
-        content = "\n".join(lines)
-        return ContextSection(
-            section_type=ContextSectionType.WORKING_CONTEXT,
-            content=content,
-            priority=30,
-            token_estimate=estimate_tokens(content),
-        )
-
     # ------------------------------------------------------------------
     # Assembly
     # ------------------------------------------------------------------
@@ -510,14 +505,11 @@ class ContextEngine:
     def _assemble_system_prompt(self, sections: list[ContextSection]) -> str:
         """Combine sections into a single system prompt string."""
         parts: list[str] = []
-        # Order: rules → profile → task → artifacts → working
         order = [
             ContextSectionType.IMMUTABLE_RULES,
             ContextSectionType.PROJECT_PROFILE,
             ContextSectionType.KNOWLEDGE_BASE,
             ContextSectionType.TASK_STATE,
-            ContextSectionType.ARTIFACT_MEMORY,
-            ContextSectionType.WORKING_CONTEXT,
         ]
         section_map: dict[ContextSectionType, ContextSection] = {
             s.section_type: s for s in sections
@@ -527,11 +519,15 @@ class ContextEngine:
                 parts.append(section_map[st].content)
         return "\n\n---\n\n".join(parts)
 
-    def _build_messages(self, session: SessionState) -> list[dict[str, Any]]:
-        """Build the messages array. We do NOT include chat history.
+    def _build_messages(
+        self,
+        session: SessionState,
+        conversation: list[dict[str, Any]] | None = None,
+    ) -> list[dict[str, Any]]:
+        """Build the messages array with real conversation history.
 
-        The user message is the current task objective (or a sentinel
-        if no task is active).
+        Includes the user objective message followed by the full
+        assistant/tool conversation — like professional agentic tools.
         """
         if session.current_task:
             step = session.current_task.current_step()
@@ -545,4 +541,10 @@ class ContextEngine:
         else:
             user_content = "Awaiting task assignment."
 
-        return [{"role": "user", "content": user_content}]
+        messages: list[dict[str, Any]] = [{"role": "user", "content": user_content}]
+
+        # Append real conversation (assistant messages + tool results)
+        if conversation:
+            messages.extend(conversation)
+
+        return messages

src/mcp/client.py

@@ -74,6 +74,7 @@ class MCPClient:
             stdout=asyncio.subprocess.PIPE,
             stderr=asyncio.subprocess.PIPE,
             env=self._env,
+            limit=1024 * 1024,  # 1MB buffer for large MCP responses
         )
         self._running = True
         self._reader_task = asyncio.create_task(self._read_loop())

src/models/session.py

@@ -36,6 +36,7 @@ class TaskStep(BaseModel):
     status: TaskStatus = TaskStatus.PENDING
     result_summary: str = ""
     tools_used: list[str] = Field(default_factory=list)
+    compacted: bool = False  # True when step results have been compacted
     started_at: datetime | None = None
     completed_at: datetime | None = None
 

src/models/tools.py

@@ -33,7 +33,8 @@ class ToolExecution(BaseModel):
     tool_name: str
     arguments: dict[str, Any] = Field(default_factory=dict)
     status: ToolExecutionStatus = ToolExecutionStatus.PENDING
-    result_summary: str = ""  # Summarised result — raw output is NEVER stored here
+    result_summary: str = ""  # Summarised result for artifacts and compacted history
+    raw_output: str = ""  # Truncated raw output for conversation messages
     error: str = ""
     duration_ms: float = 0.0
     started_at: datetime = Field(default_factory=lambda: datetime.now(timezone.utc))

src/orchestrator/agents/base.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import hashlib
 import json
 import logging
 import time
@@ -47,6 +48,10 @@ class BaseAgent:
     ) -> dict[str, Any]:
         """Run the agent's execution loop.
 
+        Uses real conversation messages with complete tool results,
+        like professional agentic tools (Claude Code, Cursor).
+        Compaction happens at the step level, not per tool result.
+
         Returns a result dict with keys: content, artifacts, tool_executions.
         """
         artifacts: list[ArtifactSummary] = await self.memory.list_artifacts(
@@ -54,15 +59,20 @@ class BaseAgent:
         )
         tool_executions: list[ToolExecution] = []
         accumulated_content = ""
-        working_items: list[dict[str, Any]] = []
+        total_input_tokens = 0
+        total_output_tokens = 0
+        # Real conversation history: assistant messages + tool results
+        conversation: list[dict[str, Any]] = []
+        tool_fingerprints: dict[str, ToolExecution] = {}
+        all_duplicates_streak = 0  # consecutive steps where ALL calls are duplicates
 
         for step in range(max_steps):
-            # Build context — NEVER includes raw tool output
+            # Build context with real conversation
             ctx = await self.context.build_context(
                 session=session,
                 agent=self.profile,
                 artifacts=artifacts,
-                working_items=working_items,
+                conversation=conversation,
             )
 
             # Prepare tool definitions
@@ -77,7 +87,7 @@ class BaseAgent:
 
             full_text = ""
             tool_calls: list[dict[str, Any]] = []
-            current_tool: dict[str, Any] | None = None
+            active_tools: dict[str, dict[str, Any]] = {}
 
             async for chunk in self.model.stream(
                 messages=ctx.to_messages(),
@@ -96,8 +106,9 @@ class BaseAgent:
                         session_id=session.session_id,
                     )
 
-                if chunk.tool_name and (current_tool is None or not current_tool.get("name")):
-                    current_tool = {
+                if chunk.tool_name and chunk.tool_call_id:
+                    if chunk.tool_call_id not in active_tools:
+                        active_tools[chunk.tool_call_id] = {
                             "id": chunk.tool_call_id,
                             "name": chunk.tool_name,
                             "arguments": "",
@@ -108,23 +119,32 @@ class BaseAgent:
                             session_id=session.session_id,
                         )
 
-                if chunk.tool_arguments and current_tool is not None and not chunk.finish_reason:
-                    # Accumulate partial argument chunks (NOT the final one)
-                    current_tool["arguments"] += chunk.tool_arguments
+                if chunk.tool_arguments and chunk.tool_call_id and not chunk.finish_reason:
+                    tool = active_tools.get(chunk.tool_call_id)
+                    if tool:
+                        tool["arguments"] += chunk.tool_arguments
 
-                if chunk.finish_reason == "tool_use" and current_tool is not None and current_tool.get("name"):
-                    # Final chunk carries complete arguments — use those if
-                    # partial accumulation is empty, otherwise use accumulated
-                    final_args = current_tool["arguments"] or chunk.tool_arguments or ""
+                if chunk.finish_reason == "tool_use" and chunk.tool_call_id:
+                    tool = active_tools.pop(chunk.tool_call_id, None)
+                    if not tool:
+                        tool = {
+                            "id": chunk.tool_call_id,
+                            "name": chunk.tool_name or "",
+                            "arguments": "",
+                        }
+                    final_args = tool["arguments"] or chunk.tool_arguments or ""
                     try:
                         args = json.loads(final_args) if final_args else {}
                     except json.JSONDecodeError:
                         logger.warning("Failed to parse tool args: %s", final_args[:200])
                         args = {}
-                    current_tool["parsed_arguments"] = args
-                    logger.debug("Tool call finalized: %s args=%s", current_tool["name"], json.dumps(args)[:200])
-                    tool_calls.append(current_tool)
-                    current_tool = None
+                    tool["parsed_arguments"] = args
+                    tool_calls.append(tool)
+
+                # Accumulate token usage from any chunk that has it
+                if chunk.usage:
+                    total_input_tokens += chunk.usage.get("input_tokens", 0)
+                    total_output_tokens += chunk.usage.get("output_tokens", 0)
 
                 if chunk.finish_reason == "end_turn":
                     break
@@ -133,28 +153,98 @@ class BaseAgent:
 
             # If no tool calls, we're done
             if not tool_calls:
+                # Add final assistant message to conversation
+                if full_text:
+                    conversation.append({"role": "assistant", "content": full_text})
                 break
 
-            # Execute tool calls
+            # Add assistant message with tool calls to conversation
+            # (OpenAI format: assistant message carries tool_calls)
+            assistant_msg: dict[str, Any] = {"role": "assistant"}
+            if full_text:
+                assistant_msg["content"] = full_text
+            assistant_msg["tool_calls"] = [
+                {
+                    "id": tc["id"],
+                    "type": "function",
+                    "function": {
+                        "name": tc["name"],
+                        "arguments": json.dumps(tc.get("parsed_arguments", {})),
+                    },
+                }
+                for tc in tool_calls
+            ]
+            conversation.append(assistant_msg)
+
+            # Execute tool calls and add COMPLETE results to conversation
+            duplicates_this_step = 0
             for tc in tool_calls:
+                fp_raw = f"{tc['name']}:{json.dumps(tc.get('parsed_arguments', {}), sort_keys=True)}"
+                fp = hashlib.md5(fp_raw.encode()).hexdigest()
+
+                if fp in tool_fingerprints:
+                    prev_exec = tool_fingerprints[fp]
+                    tool_executions.append(prev_exec)
+                    duplicates_this_step += 1
+                    # Return cached result as tool message
+                    conversation.append({
+                        "role": "tool",
+                        "tool_call_id": tc["id"],
+                        "content": f"[DUPLICADO] Ya ejecutada con mismos argumentos. Resultado: {prev_exec.raw_output[:2000]}",
+                    })
+                    logger.warning("Duplicate tool call skipped: %s (fingerprint: %s)", tc["name"], fp[:8])
+                    continue
+
                 tool_exec = await self._execute_tool(
                     session=session,
                     tool_name=tc["name"],
                     arguments=tc.get("parsed_arguments", {}),
                     artifacts=artifacts,
                 )
+                tool_fingerprints[fp] = tool_exec
                 tool_executions.append(tool_exec)
 
-                # Add summarised result to working context (NEVER raw)
-                working_items.append({
-                    "role": "tool_result",
-                    "content": f"[{tc['name']}] {tool_exec.result_summary}",
+                # COMPLETE result in conversation (truncated to safe limit)
+                conversation.append({
+                    "role": "tool",
+                    "tool_call_id": tc["id"],
+                    "content": tool_exec.raw_output[:8000] if tool_exec.raw_output else tool_exec.result_summary,
                 })
 
+            # Loop detection: if ALL tool calls in this step were duplicates
+            if duplicates_this_step == len(tool_calls):
+                all_duplicates_streak += 1
+                if all_duplicates_streak >= 2:
+                    logger.warning("Loop detected: %d consecutive steps with all duplicate calls. Breaking.", all_duplicates_streak)
+                    conversation.append({
+                        "role": "user",
+                        "content": "[SISTEMA] Se detectaron llamadas repetidas. Ya tienes toda la información necesaria. Genera tu respuesta final ahora.",
+                    })
+                    # One more chance to generate a final response
+                    ctx = await self.context.build_context(
+                        session=session, agent=self.profile,
+                        artifacts=artifacts, conversation=conversation,
+                    )
+                    async for chunk in self.model.stream(
+                        messages=ctx.to_messages(),
+                        config=config,
+                    ):
+                        if chunk.delta:
+                            accumulated_content += chunk.delta
+                        if chunk.finish_reason:
+                            break
+                    break
+            else:
+                all_duplicates_streak = 0
+
         return {
             "content": accumulated_content,
             "artifacts": artifacts,
             "tool_executions": tool_executions,
+            "usage": {
+                "input_tokens": total_input_tokens,
+                "output_tokens": total_output_tokens,
+            },
         }
 
     async def _execute_tool(
@@ -200,6 +290,7 @@ class BaseAgent:
 
             tool_exec.status = ToolExecutionStatus.COMPLETED
             tool_exec.result_summary = artifact.summary
+            tool_exec.raw_output = raw_output[:8000]
             tool_exec.duration_ms = duration
 
             await self.sse.emit(

src/orchestrator/agents/coder.py

@@ -5,7 +5,7 @@ from __future__ import annotations
 from ...models.agent import AgentProfile, AgentRole
 from .base import BaseAgent
 
-CODER_SYSTEM_PROMPT = """Eres un Agente Programador. Tu rol es ejecutar tareas de implementación usando las herramientas disponibles.
+CODER_SYSTEM_PROMPT = """Eres un Agente Programador de Acai CMS. Tu rol es ejecutar tareas de implementación usando las herramientas MCP disponibles.
 
 ## Instrucciones
 - Concéntrate en la descripción del paso actual.
@@ -16,9 +16,17 @@ CODER_SYSTEM_PROMPT = """Eres un Agente Programador. Tu rol es ejecutar tareas d
 - Responde SIEMPRE en español.
 
 ## Uso de herramientas
-- Usa herramientas cuando necesites leer archivos, escribir código o ejecutar comandos.
-- Los resultados de herramientas se te presentarán resumidos — no verás la salida cruda.
-- Si necesitas más detalle de un resultado, solicita rehidratación.
+- CONSULTA la Knowledge Base ANTES de actuar — tiene la referencia completa de tools y flujos de trabajo.
+- Para CREAR/EDITAR MÓDULOS usa `acai_write` sobre `template/estandar/modulos/NOMBRE/index-base.tpl`. El server crea la carpeta si no existe, compila y genera todos los archivos derivados automáticamente. NO necesitas compile_module.
+- `create_module` es legacy — funciona pero `acai_write` es el flujo estándar.
+- Para GESTIONAR REGISTROS de tablas (apartados, travesias, etc.) usa `create_or_update_record`.
+- Flujo de módulo nuevo: acai_write index-base.tpl → add_module_to_record → set_module_config_vars.
+- tableName siempre SIN prefijo cms_ (ej: apartados, NO cms_apartados).
+- La primary key es siempre `num`, nunca `id`.
+
+## Datos del historial
+- Si el historial de sesión incluye Key Data con recordNums o sectionIds, ÚSALOS directamente sin re-consultar.
+- Ejemplo: si el historial dice "pages: Inicio = record 2", usa recordNum=2 para la portada.
 """
 
 
@@ -35,8 +43,6 @@ def create_coder_profile() -> AgentProfile:
             "project_profile",
             "knowledge_base",
             "task_state",
-            "artifact_memory",
-            "working_context",
         ],
     )
 

src/orchestrator/agents/planner.py

@@ -55,9 +55,10 @@ def create_planner_profile() -> AgentProfile:
 class PlannerAgent(BaseAgent):
     """Generates execution plans from objectives."""
 
-    async def plan(self, session: SessionState) -> list[TaskStep]:
-        """Generate a plan and return TaskSteps."""
+    async def plan(self, session: SessionState) -> tuple[list[TaskStep], dict[str, int]]:
+        """Generate a plan and return (TaskSteps, usage)."""
         result = await self.execute(session, max_steps=1)
+        usage = result.get("usage", {"input_tokens": 0, "output_tokens": 0})
         content = result["content"].strip()
 
         # Parse the JSON plan from the model output
@@ -92,7 +93,7 @@ class PlannerAgent(BaseAgent):
                     parsed.get("facts", [])
                 )
 
-            return steps
+            return steps, usage
 
         except (json.JSONDecodeError, KeyError) as e:
             logger.warning("Failed to parse planner output: %s", e)
@@ -104,4 +105,4 @@ class PlannerAgent(BaseAgent):
                     else "Execute task",
                     agent_role="coder",
                 )
-            ]
+            ], usage

src/orchestrator/engine.py

@@ -16,7 +16,7 @@ from ..context.engine import ContextEngine
 from ..mcp.manager import MCPManager
 from ..memory.store import MemoryStore
 from ..models.agent import AgentRole
-from ..models.session import SessionState, SessionStatus, TaskStatus
+from ..models.session import SessionState, SessionStatus, TaskState, TaskStatus
 from ..streaming.sse import SSEEmitter, EventType
 from .agents.coder import CoderAgent, create_coder_profile
 from .agents.collector import CollectorAgent, create_collector_profile
@@ -115,9 +115,10 @@ class OrchestratorEngine:
 
         # 2. Plan
         task.status = TaskStatus.PLANNING
+        planner_usage: dict[str, int] = {"input_tokens": 0, "output_tokens": 0}
         try:
             planner = self._create_agent(AgentRole.PLANNER)
-            plan_steps = await planner.plan(session)
+            plan_steps, planner_usage = await planner.plan(session)
             task.plan = plan_steps
             task.status = TaskStatus.EXECUTING
         except Exception as e:
@@ -181,6 +182,10 @@ class OrchestratorEngine:
                 for artifact in step_result.get("artifacts", []):
                     task.facts_extracted.extend(artifact.facts[:5])
 
+                # Decide if previous steps should be compacted
+                if i > 0:
+                    self._maybe_compact_previous_steps(task, current_index=i)
+
             except Exception as e:
                 logger.error("Step %d failed: %s", i + 1, e)
                 step.status = TaskStatus.FAILED
@@ -230,6 +235,21 @@ class OrchestratorEngine:
             session_id=session.session_id,
         )
 
+        # Accumulate token usage: planner + all steps + review
+        total_input = planner_usage.get("input_tokens", 0)
+        total_output = planner_usage.get("output_tokens", 0)
+        for r in results:
+            total_input += r.get("usage", {}).get("input_tokens", 0)
+            total_output += r.get("usage", {}).get("output_tokens", 0)
+        # Add review usage if any
+        total_input += review_result.get("usage", {}).get("input_tokens", 0)
+        total_output += review_result.get("usage", {}).get("output_tokens", 0)
+        # Calculate cost
+        cost_usd = (
+            (total_input / 1_000_000) * settings.cost_per_1m_input
+            + (total_output / 1_000_000) * settings.cost_per_1m_output
+        )
+
         return {
             "session_id": session.session_id,
             "task_id": task.task_id,
@@ -241,6 +261,11 @@ class OrchestratorEngine:
             ),
             "review": review_result.get("content", ""),
             "status": status,
+            "usage": {
+                "input_tokens": total_input,
+                "output_tokens": total_output,
+            },
+            "total_cost_usd": round(cost_usd, 6),
         }
 
     def _error_result(self, session: SessionState, error: str) -> dict[str, Any]:
@@ -292,12 +317,16 @@ class OrchestratorEngine:
         for step in task.plan:
             tools_used.update(step.tools_used)
 
+        # Extract key structured data from tool executions
+        key_data = self._extract_key_data_from_results(results)
+
         history_entry = {
             "task_id": task.task_id,
             "objective": task.objective,
             "status": task.status.value,
             "steps": len(task.plan),
             "facts": task.facts_extracted[-10:],
+            "key_data": key_data,
             "tools_used": list(tools_used)[:10],
             "artifacts_count": len(task_artifacts),
             "summary": "; ".join(step_summaries)[:300],
@@ -323,6 +352,102 @@ class OrchestratorEngine:
             task.task_id, len(task.facts_extracted), len(tools_used), len(task_artifacts),
         )
 
+    @staticmethod
+    def _extract_key_data_from_results(results: list[dict[str, Any]]) -> dict[str, Any]:
+        """Extract structured data from tool executions for task history.
+
+        Preserves key identifiers (recordNum, sectionId, tableName, moduleId)
+        so the model retains context across tasks without re-querying.
+        """
+        key_data: dict[str, Any] = {}
+        seen_tables: dict[str, list[int]] = {}  # tableName -> recordNums
+        seen_sections: list[str] = []
+        seen_modules: list[str] = []
+        seen_pages: dict[str, int] = {}  # page name/url -> recordNum
+
+        for result in results:
+            for te in result.get("tool_executions", []):
+                args = te.arguments
+                name = te.tool_name
+
+                # Track table + record relationships
+                table = args.get("tableName", "")
+                record = args.get("recordNum")
+                if table and record:
+                    record_int = int(record) if str(record).isdigit() else None
+                    if record_int and table not in seen_tables:
+                        seen_tables[table] = []
+                    if record_int and record_int not in seen_tables.get(table, []):
+                        seen_tables[table].append(record_int)
+
+                # Track section IDs
+                section = args.get("sectionId", "")
+                if section and section not in seen_sections:
+                    seen_sections.append(section)
+
+                # Track modules
+                module = args.get("moduleId", "") or args.get("moduleName", "")
+                if module and module not in seen_modules:
+                    seen_modules.append(module)
+
+                # Extract page info from raw output (enlace, name)
+                if te.raw_output and "enlace" in te.raw_output:
+                    try:
+                        import json as _json
+                        # Try to parse structured data from output
+                        for line in te.raw_output.splitlines():
+                            line = line.strip()
+                            if line.startswith("{"):
+                                try:
+                                    data = _json.loads(line)
+                                    if "enlace" in data and "num" in data:
+                                        page_key = data.get("name", data["enlace"])
+                                        seen_pages[page_key] = int(data["num"])
+                                except _json.JSONDecodeError:
+                                    pass
+                    except Exception:
+                        pass
+
+        if seen_tables:
+            key_data["tables"] = {t: nums[:10] for t, nums in seen_tables.items()}
+        if seen_sections:
+            key_data["sections"] = seen_sections[:20]
+        if seen_modules:
+            key_data["modules"] = seen_modules[:20]
+        if seen_pages:
+            key_data["pages"] = dict(list(seen_pages.items())[:20])
+
+        return key_data
+
+    def _maybe_compact_previous_steps(
+        self, task: TaskState, current_index: int
+    ) -> None:
+        """Decide if previous steps should be compacted. Deterministic rules."""
+        current_step = task.plan[current_index]
+
+        for i in range(current_index):
+            prev = task.plan[i]
+            if prev.compacted or prev.status != TaskStatus.COMPLETED:
+                continue
+
+            # Rule 1: Change of agent role → previous steps are a different focus
+            if prev.agent_role != current_step.agent_role:
+                prev.compacted = True
+                logger.info(
+                    "Compacted step %d (%s) — agent changed to %s",
+                    i + 1, prev.agent_role, current_step.agent_role,
+                )
+                continue
+
+            # Rule 2: More than 3 completed non-compacted steps → compact oldest
+            non_compacted = [
+                s for s in task.plan[:current_index]
+                if s.status == TaskStatus.COMPLETED and not s.compacted
+            ]
+            if len(non_compacted) > 3:
+                non_compacted[0].compacted = True
+                logger.info("Compacted oldest step to stay within budget")
+
     def _create_agent(self, role: AgentRole) -> PlannerAgent | CoderAgent | CollectorAgent | ReviewerAgent:
         """Instantiate a subagent for the given role."""
         profile = self._profiles[role]

tests/conftest.py

@@ -0,0 +1,8 @@
+"""Configuracion de pytest para agenticSystem tests.
+
+Estos tests son 100% standalone — no importan desde src/ directamente
+porque el entorno de CI puede no tener las dependencias pesadas
+(anthropic, tiktoken, pydantic, etc.) ni Python 3.11+.
+
+La logica bajo test se replica o se extrae como funciones puras.
+"""

tests/test_compactor.py

@@ -0,0 +1,362 @@
+"""Tests para la logica de context/compactor.py — estimacion de tokens,
+extraccion de facts, construccion de summaries y compactacion de secciones.
+
+Se replica la logica pura sin importar src/ (evita dependencias pesadas).
+"""
+
+import hashlib
+import re
+from dataclasses import dataclass, field
+from typing import List
+
+import pytest
+
+
+# =====================================================================
+# Replicas de la logica del compactor (funciones puras)
+# =====================================================================
+
+
+def estimate_tokens_fallback(text: str) -> int:
+    """Replica del fallback de estimate_tokens (sin tiktoken)."""
+    if not text:
+        return 0
+    return max(1, len(text) // 4)
+
+
+def extract_facts(raw_output: str) -> list:
+    """Replica exacta de ContextCompactor._extract_facts."""
+    facts = []
+    lines = raw_output.strip().splitlines()
+
+    for line in lines[:100]:
+        line = line.strip()
+        if not line or len(line) < 10:
+            continue
+        if re.match(r"^[\w\s]+:\s+.+", line) and len(line) < 200:
+            facts.append(line)
+        elif re.match(r"^(✓|✗|PASS|FAIL|ERROR|OK|INFO|WARNING)", line):
+            facts.append(line)
+        elif re.match(r"^[\w/\\.]+\s*[:\-]\s*.+", line) and len(line) < 200:
+            facts.append(line)
+
+    seen = set()
+    unique = []
+    for f in facts:
+        if f not in seen:
+            seen.add(f)
+            unique.append(f)
+    return unique[:15]
+
+
+def build_summary(tool_name: str, raw_output: str, facts: list) -> str:
+    """Replica exacta de ContextCompactor._build_summary."""
+    lines = raw_output.strip().splitlines()
+    total_lines = len(lines)
+    char_count = len(raw_output)
+
+    parts = [f"Tool '{tool_name}' returned {total_lines} lines ({char_count} chars)."]
+
+    if facts:
+        parts.append(f"Key findings: {'; '.join(facts[:5])}")
+
+    meaningful = [l.strip() for l in lines if l.strip()]
+    if meaningful:
+        parts.append(f"First: {meaningful[0][:120]}")
+        if len(meaningful) > 1:
+            parts.append(f"Last: {meaningful[-1][:120]}")
+
+    return " ".join(parts)
+
+
+def infer_artifact_type(tool_name: str) -> str:
+    """Replica de ContextCompactor._infer_artifact_type."""
+    tool_lower = tool_name.lower()
+    if any(k in tool_lower for k in ("read", "file", "code", "write", "edit")):
+        return "code"
+    if any(k in tool_lower for k in ("test", "check", "lint", "validate")):
+        return "test_result"
+    if any(k in tool_lower for k in ("search", "find", "grep", "glob")):
+        return "analysis"
+    if any(k in tool_lower for k in ("plan", "design", "architect")):
+        return "plan"
+    return "general"
+
+
+def summarize_tool_output(tool_name: str, raw_output: str, session_id: str, task_id: str) -> dict:
+    """Replica simplificada de ContextCompactor.summarize_tool_output.
+    Devuelve un dict con los mismos campos que ArtifactSummary.
+    """
+    facts = extract_facts(raw_output)
+    summary = build_summary(tool_name, raw_output, facts)
+    artifact_type = infer_artifact_type(tool_name)
+    artifact_id = hashlib.sha256(
+        f"{session_id}:{task_id}:{tool_name}:{raw_output[:200]}".encode()
+    ).hexdigest()[:16]
+
+    return {
+        "artifact_id": artifact_id,
+        "session_id": session_id,
+        "task_id": task_id,
+        "artifact_type": artifact_type,
+        "title": f"Output of {tool_name}",
+        "summary": summary,
+        "facts": facts,
+        "source_tool": tool_name,
+        "char_count": len(raw_output),
+    }
+
+
+# --- Modelo simplificado de ContextSection para test de compactacion ---
+
+
+@dataclass
+class Section:
+    section_type: str  # "immutable_rules", "working_context", "task_state", etc.
+    content: str
+    priority: int = 0
+    token_estimate: int = 0
+
+
+def compact_sections(sections: list, max_tokens: int) -> list:
+    """Replica de ContextCompactor.compact_sections (logica pura)."""
+    # 1. Deduplicar
+    seen = set()
+    unique = []
+    for s in sections:
+        h = hashlib.md5(s.content.encode()).hexdigest()
+        if h not in seen:
+            seen.add(h)
+            unique.append(s)
+    sections = unique
+
+    # 2. Estimar tokens
+    for s in sections:
+        s.token_estimate = estimate_tokens_fallback(s.content)
+
+    total = sum(s.token_estimate for s in sections)
+    if total <= max_tokens:
+        return sections
+
+    # 3. Ordenar por prioridad (mayor primero)
+    sections.sort(key=lambda s: s.priority, reverse=True)
+
+    # 4. Trim de menor prioridad
+    while total > max_tokens and sections:
+        lowest = sections[-1]
+        if lowest.section_type == "immutable_rules":
+            break
+        # Compactacion simple: eliminar lineas vacias
+        compacted_lines = [l.rstrip() for l in lowest.content.splitlines() if l.strip()]
+        compacted = "\n".join(compacted_lines)
+        new_est = estimate_tokens_fallback(compacted)
+        saved = lowest.token_estimate - new_est
+        if saved > 0:
+            lowest.content = compacted
+            lowest.token_estimate = new_est
+            total -= saved
+        else:
+            total -= lowest.token_estimate
+            sections.pop()
+
+    return sections
+
+
+# =====================================================================
+# Tests: estimate_tokens
+# =====================================================================
+
+
+class TestEstimateTokens:
+    def test_positive_for_nonempty_text(self):
+        result = estimate_tokens_fallback("Hello world, this is a test string.")
+        assert isinstance(result, int)
+        assert result > 0
+
+    def test_zero_for_empty_string(self):
+        assert estimate_tokens_fallback("") == 0
+
+    def test_longer_text_more_tokens(self):
+        short = estimate_tokens_fallback("hi")
+        long = estimate_tokens_fallback("hi " * 500)
+        assert long > short
+
+    def test_returns_int_type(self):
+        assert isinstance(estimate_tokens_fallback("cualquier texto"), int)
+
+    def test_minimum_is_one_for_short_text(self):
+        # "ab" -> len 2 // 4 = 0, pero max(1, 0) = 1
+        assert estimate_tokens_fallback("ab") == 1
+
+
+# =====================================================================
+# Tests: _extract_facts
+# =====================================================================
+
+
+class TestExtractFacts:
+    def test_extracts_key_value_lines(self):
+        raw = "Status: running\nVersion: 3.2.1\nIgnored short\nName: my-module"
+        facts = extract_facts(raw)
+        assert any("Status: running" in f for f in facts)
+        assert any("Version: 3.2.1" in f for f in facts)
+        assert any("Name: my-module" in f for f in facts)
+
+    def test_extracts_status_indicators(self):
+        raw = "PASS test_login completed\nFAIL test_logout broken\nOK everything fine"
+        facts = extract_facts(raw)
+        assert any("PASS" in f for f in facts)
+        assert any("FAIL" in f for f in facts)
+
+    def test_ignores_short_lines(self):
+        raw = "ok\nhi\nyes\nStatus: this is long enough to be a fact"
+        facts = extract_facts(raw)
+        assert not any(f in ("ok", "hi", "yes") for f in facts)
+
+    def test_deduplicates(self):
+        raw = "Status: running value\nStatus: running value\nStatus: running value"
+        facts = extract_facts(raw)
+        assert facts.count("Status: running value") == 1
+
+    def test_limits_to_15(self):
+        lines = [f"Key{i}: value number {i} with enough length" for i in range(30)]
+        raw = "\n".join(lines)
+        facts = extract_facts(raw)
+        assert len(facts) <= 15
+
+    def test_empty_input(self):
+        facts = extract_facts("")
+        assert facts == []
+
+
+# =====================================================================
+# Tests: _build_summary
+# =====================================================================
+
+
+class TestBuildSummary:
+    def test_includes_tool_name(self):
+        summary = build_summary("read_file", "line1\nline2\nline3", [])
+        assert "read_file" in summary
+
+    def test_includes_line_count(self):
+        raw = "line1\nline2\nline3"
+        summary = build_summary("my_tool", raw, [])
+        assert "3 lines" in summary
+
+    def test_includes_char_count(self):
+        raw = "some content here"
+        summary = build_summary("my_tool", raw, [])
+        assert str(len(raw)) in summary
+
+    def test_includes_facts_when_present(self):
+        facts = ["Status: ok", "Count: 42"]
+        summary = build_summary("my_tool", "data", facts)
+        assert "Status: ok" in summary
+
+    def test_includes_first_line(self):
+        raw = "primera linea importante\nsegunda\ntercera"
+        summary = build_summary("tool", raw, [])
+        assert "primera linea importante" in summary
+
+
+# =====================================================================
+# Tests: summarize_tool_output
+# =====================================================================
+
+
+class TestSummarizeToolOutput:
+    def test_returns_dict_with_correct_fields(self):
+        result = summarize_tool_output(
+            tool_name="read_file",
+            raw_output="Status: ok\nContent: hello world here",
+            session_id="sess-001",
+            task_id="task-001",
+        )
+        assert isinstance(result, dict)
+        assert result["session_id"] == "sess-001"
+        assert result["task_id"] == "task-001"
+        assert result["source_tool"] == "read_file"
+        assert result["title"] == "Output of read_file"
+        assert result["artifact_id"]  # no vacio
+        assert result["summary"]  # no vacio
+        assert result["char_count"] > 0
+
+    def test_artifact_type_inference(self):
+        assert summarize_tool_output("read_file", "x", "s", "t")["artifact_type"] == "code"
+        assert summarize_tool_output("test_run", "x", "s", "t")["artifact_type"] == "test_result"
+        assert summarize_tool_output("search_records", "x", "s", "t")["artifact_type"] == "analysis"
+        assert summarize_tool_output("deploy_app", "x", "s", "t")["artifact_type"] == "general"
+
+    def test_artifact_id_is_deterministic(self):
+        r1 = summarize_tool_output("tool", "output", "s", "t")
+        r2 = summarize_tool_output("tool", "output", "s", "t")
+        assert r1["artifact_id"] == r2["artifact_id"]
+
+    def test_artifact_id_length(self):
+        result = summarize_tool_output("tool", "output", "s", "t")
+        assert len(result["artifact_id"]) == 16
+
+
+# =====================================================================
+# Tests: compact_sections
+# =====================================================================
+
+
+class TestCompactSections:
+    def test_never_removes_immutable_rules(self):
+        sections = [
+            Section(
+                section_type="immutable_rules",
+                content="You must always follow these rules " * 20,
+                priority=100,
+            ),
+            Section(
+                section_type="working_context",
+                content="Some working context data " * 50,
+                priority=1,
+            ),
+        ]
+        result = compact_sections(sections, max_tokens=50)
+        types = [s.section_type for s in result]
+        assert "immutable_rules" in types
+
+    def test_respects_priority_order(self):
+        """Secciones de mayor prioridad sobreviven a la compactacion.
+        Usamos un budget que cabe la seccion alta pero no ambas."""
+        high = Section(
+            section_type="task_state",
+            content="Important task data here",  # ~6 tokens
+            priority=90,
+        )
+        low = Section(
+            section_type="working_context",
+            content="Low priority stuff " * 50,  # ~250 tokens
+            priority=1,
+        )
+        # Budget suficiente para high (~6) pero no para high+low (~256)
+        result = compact_sections([high, low], max_tokens=20)
+        types = [s.section_type for s in result]
+        assert "task_state" in types
+        # La de baja prioridad deberia haberse eliminado o compactado
+        assert len(result) <= 2
+
+    def test_no_compaction_when_within_budget(self):
+        sections = [
+            Section(
+                section_type="task_state",
+                content="Short content",
+                priority=50,
+            ),
+        ]
+        result = compact_sections(sections, max_tokens=999_999)
+        assert len(result) == 1
+        assert result[0].content == "Short content"
+
+    def test_deduplicates_identical_sections(self):
+        sections = [
+            Section(section_type="working_context", content="duplicated content", priority=10),
+            Section(section_type="working_context", content="duplicated content", priority=10),
+        ]
+        result = compact_sections(sections, max_tokens=999_999)
+        assert len(result) == 1

tests/test_cost_calculation.py

@@ -0,0 +1,71 @@
+"""Tests para el calculo de costes del orquestador.
+
+Replica la formula de coste de OrchestratorEngine._run_pipeline():
+    cost_usd = (input_tokens / 1_000_000) * cost_per_1m_input
+             + (output_tokens / 1_000_000) * cost_per_1m_output
+
+Defaults: cost_per_1m_input=2.50, cost_per_1m_output=15.00
+"""
+
+import pytest
+
+
+def calculate_cost(
+    input_tokens: int,
+    output_tokens: int,
+    cost_per_1m_input: float = 2.50,
+    cost_per_1m_output: float = 15.00,
+) -> float:
+    """Replica exacta de la formula de coste en engine.py."""
+    return (
+        (input_tokens / 1_000_000) * cost_per_1m_input
+        + (output_tokens / 1_000_000) * cost_per_1m_output
+    )
+
+
+class TestCostCalculation:
+    def test_1m_input_tokens(self):
+        cost = calculate_cost(1_000_000, 0)
+        assert cost == pytest.approx(2.50)
+
+    def test_1m_output_tokens(self):
+        cost = calculate_cost(0, 1_000_000)
+        assert cost == pytest.approx(15.00)
+
+    def test_500k_input_100k_output(self):
+        cost = calculate_cost(500_000, 100_000)
+        # (500_000 / 1_000_000) * 2.50 + (100_000 / 1_000_000) * 15.00
+        # = 1.25 + 1.50 = 2.75
+        assert cost == pytest.approx(2.75)
+
+    def test_zero_tokens(self):
+        cost = calculate_cost(0, 0)
+        assert cost == 0.0
+
+    def test_custom_pricing(self):
+        cost = calculate_cost(
+            1_000_000, 1_000_000,
+            cost_per_1m_input=3.00,
+            cost_per_1m_output=10.00,
+        )
+        assert cost == pytest.approx(13.00)
+
+    def test_small_token_count(self):
+        """Pocos tokens = coste muy bajo pero no cero."""
+        cost = calculate_cost(100, 50)
+        assert cost > 0
+        assert cost < 0.01
+
+    def test_round_to_6_decimals(self):
+        """El engine hace round(cost_usd, 6)."""
+        cost = calculate_cost(1, 1)
+        rounded = round(cost, 6)
+        # (1/1M)*2.50 + (1/1M)*15.00 = 1.75e-05
+        # round(1.75e-05, 6) = 1.7e-05 (banker's rounding: 5 rounds to even)
+        assert rounded == pytest.approx(0.000017, abs=1e-7)
+
+    def test_output_more_expensive_than_input(self):
+        """Con defaults, output es 6x mas caro que input."""
+        input_cost = calculate_cost(1_000_000, 0)
+        output_cost = calculate_cost(0, 1_000_000)
+        assert output_cost == pytest.approx(input_cost * 6.0)

tests/test_fingerprint.py

@@ -0,0 +1,61 @@
+"""Tests para la logica de fingerprinting/deduplicacion de tool calls.
+
+Replica la logica de BaseAgent.execute() (lineas con hashlib.md5) sin
+necesidad de instanciar BaseAgent ni sus dependencias.
+"""
+
+import hashlib
+import json
+
+import pytest
+
+
+def compute_fingerprint(tool_name: str, args: dict) -> str:
+    """Replica exacta de la logica de fingerprint en BaseAgent.execute()."""
+    fp_raw = f"{tool_name}:{json.dumps(args, sort_keys=True)}"
+    return hashlib.md5(fp_raw.encode()).hexdigest()
+
+
+class TestFingerprint:
+    def test_same_tool_same_args_same_fingerprint(self):
+        fp1 = compute_fingerprint("read_file", {"path": "/index.html"})
+        fp2 = compute_fingerprint("read_file", {"path": "/index.html"})
+        assert fp1 == fp2
+
+    def test_same_tool_different_args_different_fingerprint(self):
+        fp1 = compute_fingerprint("read_file", {"path": "/index.html"})
+        fp2 = compute_fingerprint("read_file", {"path": "/style.css"})
+        assert fp1 != fp2
+
+    def test_different_tool_same_args_different_fingerprint(self):
+        fp1 = compute_fingerprint("read_file", {"path": "/index.html"})
+        fp2 = compute_fingerprint("write_file", {"path": "/index.html"})
+        assert fp1 != fp2
+
+    def test_fingerprint_is_md5_hex_32_chars(self):
+        fp = compute_fingerprint("any_tool", {"key": "value"})
+        assert len(fp) == 32
+        assert all(c in "0123456789abcdef" for c in fp)
+
+    def test_arg_order_does_not_matter(self):
+        """json.dumps con sort_keys=True normaliza el orden."""
+        fp1 = compute_fingerprint("tool", {"b": 2, "a": 1})
+        fp2 = compute_fingerprint("tool", {"a": 1, "b": 2})
+        assert fp1 == fp2
+
+    def test_empty_args(self):
+        fp = compute_fingerprint("tool", {})
+        assert len(fp) == 32
+        # Debe ser determinista
+        assert fp == compute_fingerprint("tool", {})
+
+    def test_nested_args(self):
+        args = {"filter": {"table": "pages", "status": "active"}, "limit": 10}
+        fp1 = compute_fingerprint("search", args)
+        fp2 = compute_fingerprint("search", args)
+        assert fp1 == fp2
+
+    def test_different_nested_values(self):
+        fp1 = compute_fingerprint("search", {"filter": {"status": "active"}})
+        fp2 = compute_fingerprint("search", {"filter": {"status": "draft"}})
+        assert fp1 != fp2

tests/test_key_data_extraction.py

@@ -0,0 +1,152 @@
+"""Tests para la logica de _extract_key_data_from_results del OrchestratorEngine.
+
+Se replica la funcion como logica pura, sin importar src/ (evita dependencias).
+Los ToolExecution se representan como SimpleNamespace con .arguments y .tool_name.
+"""
+
+import json
+from types import SimpleNamespace
+from typing import Any
+
+import pytest
+
+
+def _make_tool_execution(tool_name: str, arguments: dict, raw_output: str = "") -> SimpleNamespace:
+    """Crea un objeto similar a ToolExecution con los atributos necesarios."""
+    return SimpleNamespace(
+        tool_name=tool_name,
+        arguments=arguments,
+        raw_output=raw_output,
+    )
+
+
+def _make_result(*tool_executions) -> dict:
+    return {"tool_executions": list(tool_executions), "content": "ok"}
+
+
+def extract_key_data_from_results(results: list) -> dict:
+    """Replica exacta de OrchestratorEngine._extract_key_data_from_results."""
+    key_data: dict[str, Any] = {}
+    seen_tables: dict[str, list] = {}
+    seen_sections: list = []
+    seen_modules: list = []
+    seen_pages: dict[str, int] = {}
+
+    for result in results:
+        for te in result.get("tool_executions", []):
+            args = te.arguments
+            name = te.tool_name
+
+            table = args.get("tableName", "")
+            record = args.get("recordNum")
+            if table and record:
+                record_int = int(record) if str(record).isdigit() else None
+                if record_int and table not in seen_tables:
+                    seen_tables[table] = []
+                if record_int and record_int not in seen_tables.get(table, []):
+                    seen_tables[table].append(record_int)
+
+            section = args.get("sectionId", "")
+            if section and section not in seen_sections:
+                seen_sections.append(section)
+
+            module = args.get("moduleId", "") or args.get("moduleName", "")
+            if module and module not in seen_modules:
+                seen_modules.append(module)
+
+            if te.raw_output and "enlace" in te.raw_output:
+                try:
+                    for line in te.raw_output.splitlines():
+                        line = line.strip()
+                        if line.startswith("{"):
+                            try:
+                                data = json.loads(line)
+                                if "enlace" in data and "num" in data:
+                                    page_key = data.get("name", data["enlace"])
+                                    seen_pages[page_key] = int(data["num"])
+                            except json.JSONDecodeError:
+                                pass
+                except Exception:
+                    pass
+
+    if seen_tables:
+        key_data["tables"] = {t: nums[:10] for t, nums in seen_tables.items()}
+    if seen_sections:
+        key_data["sections"] = seen_sections[:20]
+    if seen_modules:
+        key_data["modules"] = seen_modules[:20]
+    if seen_pages:
+        key_data["pages"] = dict(list(seen_pages.items())[:20])
+
+    return key_data
+
+
+# =====================================================================
+# Tests
+# =====================================================================
+
+
+class TestExtractKeyDataFromResults:
+    def test_extracts_table_and_record(self):
+        te = _make_tool_execution("update_record", {"tableName": "pages", "recordNum": "42"})
+        key_data = extract_key_data_from_results([_make_result(te)])
+        assert "tables" in key_data
+        assert "pages" in key_data["tables"]
+        assert 42 in key_data["tables"]["pages"]
+
+    def test_extracts_section_id(self):
+        te = _make_tool_execution("get_section", {"sectionId": "hero-banner"})
+        key_data = extract_key_data_from_results([_make_result(te)])
+        assert "sections" in key_data
+        assert "hero-banner" in key_data["sections"]
+
+    def test_extracts_module_id(self):
+        te = _make_tool_execution("compile_module", {"moduleId": "gallery-slider"})
+        key_data = extract_key_data_from_results([_make_result(te)])
+        assert "modules" in key_data
+        assert "gallery-slider" in key_data["modules"]
+
+    def test_extracts_module_name_fallback(self):
+        te = _make_tool_execution("compile_module", {"moduleName": "contact-form"})
+        key_data = extract_key_data_from_results([_make_result(te)])
+        assert "modules" in key_data
+        assert "contact-form" in key_data["modules"]
+
+    def test_empty_results(self):
+        key_data = extract_key_data_from_results([])
+        assert key_data == {}
+
+    def test_no_tool_executions_in_result(self):
+        key_data = extract_key_data_from_results([{"content": "x", "tool_executions": []}])
+        assert key_data == {}
+
+    def test_result_without_tool_executions_key(self):
+        key_data = extract_key_data_from_results([{"content": "just text"}])
+        assert key_data == {}
+
+    def test_tool_execution_without_relevant_args(self):
+        te = _make_tool_execution("read_file", {"path": "/var/www/index.html"})
+        key_data = extract_key_data_from_results([_make_result(te)])
+        assert key_data == {}
+
+    def test_multiple_tables_and_records(self):
+        te1 = _make_tool_execution("update_record", {"tableName": "pages", "recordNum": "1"})
+        te2 = _make_tool_execution("update_record", {"tableName": "pages", "recordNum": "5"})
+        te3 = _make_tool_execution("get_record", {"tableName": "blog", "recordNum": "10"})
+        key_data = extract_key_data_from_results([_make_result(te1, te2, te3)])
+        assert 1 in key_data["tables"]["pages"]
+        assert 5 in key_data["tables"]["pages"]
+        assert 10 in key_data["tables"]["blog"]
+
+    def test_deduplicates_records(self):
+        te1 = _make_tool_execution("a", {"tableName": "t", "recordNum": "7"})
+        te2 = _make_tool_execution("b", {"tableName": "t", "recordNum": "7"})
+        key_data = extract_key_data_from_results([_make_result(te1, te2)])
+        assert key_data["tables"]["t"].count(7) == 1
+
+    def test_extracts_pages_from_raw_output(self):
+        raw = '{"enlace": "/contacto", "num": 15, "name": "Contacto"}\nother line'
+        te = _make_tool_execution("list_pages", {"tableName": "web"}, raw_output=raw)
+        key_data = extract_key_data_from_results([_make_result(te)])
+        assert "pages" in key_data
+        assert key_data["pages"]["Contacto"] == 15