rlm-code 0.1.5__tar.gz → 0.1.6__tar.gz

{rlm_code-0.1.5 → rlm_code-0.1.6}/CHANGELOG.md

@@ -5,6 +5,30 @@ All notable changes to this project are documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.1.6] - 2026-02-20
+
+### Added
+- Harness strategy selector with `tool_call` (default) and opt-in `codemode`.
+- CodeMode execution flow in harness: MCP tool discovery (`search_tools`), typed tool surface prompt, single-program generation, guardrail validation, and MCP chain execution (`call_tool_chain`).
+- Benchmark support for harness strategy comparison with CodeMode telemetry fields (`harness_strategy`, `codemode_chain_calls`, `codemode_search_calls`, `codemode_discovery_calls`, `codemode_guardrail_blocked`).
+- New top-level CodeMode docs section with dedicated pages for quickstart, architecture, guardrails, and evaluation.
+- Release documentation set for CodeMode:
+  - quickstart and operator workflow
+  - integration architecture and runtime controls
+  - provider/bridge separation model (Cloudflare-based, UTCP, custom)
+  - CodeMode sandbox responsibility and deployment matrix
+  - guardrail policy and safety runbook
+  - benchmark evaluation and promotion-gate criteria
+
+### Changed
+- `/harness run` supports `strategy=tool_call|codemode` and `mcp_server=<name>`.
+- `/rlm bench` in `mode=harness` supports `strategy=tool_call|codemode`.
+- Harness and benchmark command handling now auto-enables MCP when `strategy=codemode` is selected.
+
+### Security
+- Added explicit CodeMode guardrail policy documentation with blocked API classes and runtime limit defaults.
+- Codemode path remains opt-in; default harness behavior remains strict baseline `strategy=tool_call`.
+
 ## [0.1.5] - 2026-02-15
 
 Initial public release of **RLM Code**.
@@ -31,3 +55,4 @@ Initial public release of **RLM Code**.
 - Unsafe local `exec` usage preserved only as an explicit, opt-in path for advanced development scenarios.
 
 [0.1.5]: https://github.com/SuperagenticAI/rlm-code/releases/tag/v0.1.5
+[0.1.6]: https://github.com/SuperagenticAI/rlm-code/releases/tag/v0.1.6

{rlm_code-0.1.5 → rlm_code-0.1.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rlm-code
-Version: 0.1.5
+Version: 0.1.6
 Summary: RLM Code: Research Playground & Evaluation OS for Recursive Language Model Agentic Systems
 Project-URL: Homepage, https://github.com/SuperagenticAI/rlm-code
 Project-URL: Documentation, https://superagenticai.github.io/rlm-code/
@@ -99,20 +99,18 @@ Description-Content-Type: text/markdown
   </a>
 </p>
 
-<p align="center">
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Version" src="https://img.shields.io/pypi/v/rlm-code"></a>
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Python Versions" src="https://img.shields.io/pypi/pyversions/rlm-code"></a>
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Downloads" src="https://img.shields.io/pypi/dm/rlm-code"></a>
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Wheel" src="https://img.shields.io/pypi/wheel/rlm-code"></a>
-  <a href="LICENSE"><img alt="License" src="https://img.shields.io/pypi/l/rlm-code"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml"><img alt="CI" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml"><img alt="Pre-commit" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml"><img alt="Docs Deploy" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml"><img alt="Release" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/stargazers"><img alt="GitHub Stars" src="https://img.shields.io/github/stars/SuperagenticAI/rlm-code?style=social"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/issues"><img alt="GitHub Issues" src="https://img.shields.io/github/issues/SuperagenticAI/rlm-code"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/pulls"><img alt="GitHub Pull Requests" src="https://img.shields.io/github/issues-pr/SuperagenticAI/rlm-code"></a>
-</p>
+[![PyPI Version](https://img.shields.io/pypi/v/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![Python Versions](https://img.shields.io/pypi/pyversions/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![PyPI Wheel](https://img.shields.io/pypi/wheel/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![License](https://img.shields.io/pypi/l/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![CI](https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml)
+[![Pre-commit](https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml)
+[![Docs Deploy](https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml)
+[![Release](https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml)
+[![Docs](https://img.shields.io/badge/Docs-RLM%20Code-ff7a18.svg?logo=readthedocs&logoColor=white)](https://superagenticai.github.io/rlm-code/)
+[![GitHub Stars](https://img.shields.io/github/stars/SuperagenticAI/rlm-code.svg)](https://github.com/SuperagenticAI/rlm-code/stargazers)
+[![GitHub Issues](https://img.shields.io/github/issues/SuperagenticAI/rlm-code.svg)](https://github.com/SuperagenticAI/rlm-code/issues)
+[![GitHub Pull Requests](https://img.shields.io/github/issues-pr/SuperagenticAI/rlm-code.svg)](https://github.com/SuperagenticAI/rlm-code/pulls)
 
 **Run LLM-powered agents in a REPL loop, benchmark them, and compare results.**
 
@@ -120,6 +118,34 @@ RLM Code implements the [Recursive Language Models](https://arxiv.org/abs/2502.0
 
 RLM Code wraps this algorithm in an interactive terminal UI with built-in benchmarks, trajectory replay, and observability.
 
+## Release v0.1.6
+
+This release adds the new CodeMode path as an opt-in harness strategy.
+
+- New harness strategy: `strategy=codemode` (default remains `strategy=tool_call`)
+- MCP bridge flow for CodeMode: `search_tools` -> typed tool surface -> `call_tool_chain`
+- Guardrails before execution: blocked API classes plus timeout/size/tool-call caps
+- Benchmark telemetry for side-by-side comparison: `tool_call` vs `codemode`
+- Dedicated docs section for CodeMode: quickstart, architecture, guardrails, evaluation
+
+Example:
+
+```text
+/harness run "implement feature and add tests" steps=8 mcp=on strategy=codemode mcp_server=codemode
+```
+
+## Documentation
+
+<p align="center">
+  <a href="https://superagenticai.github.io/rlm-code/">
+    <img alt="Read the RLM Code Docs" src="https://img.shields.io/badge/Read%20the%20Docs-RLM%20Code-ff7a18?style=for-the-badge&logo=readthedocs&logoColor=white">
+  </a>
+</p>
+
+<p align="center">
+  <a href="https://superagenticai.github.io/rlm-code/"><strong>Open the full documentation</strong></a>
+</p>
+
 ## Install
 
 ```bash
@@ -399,7 +425,7 @@ rlm_code/
   harness/          # Tool-using coding harness (/harness)
 ```
 
-## Documentation
+## Resources
 
 Full docs: https://superagenticai.github.io/rlm-code/
 

{rlm_code-0.1.5 → rlm_code-0.1.6}/README.md

@@ -6,20 +6,18 @@
   </a>
 </p>
 
-<p align="center">
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Version" src="https://img.shields.io/pypi/v/rlm-code"></a>
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Python Versions" src="https://img.shields.io/pypi/pyversions/rlm-code"></a>
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Downloads" src="https://img.shields.io/pypi/dm/rlm-code"></a>
-  <a href="https://pypi.org/project/rlm-code/"><img alt="PyPI Wheel" src="https://img.shields.io/pypi/wheel/rlm-code"></a>
-  <a href="LICENSE"><img alt="License" src="https://img.shields.io/pypi/l/rlm-code"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml"><img alt="CI" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml"><img alt="Pre-commit" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml"><img alt="Docs Deploy" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml"><img alt="Release" src="https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml/badge.svg"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/stargazers"><img alt="GitHub Stars" src="https://img.shields.io/github/stars/SuperagenticAI/rlm-code?style=social"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/issues"><img alt="GitHub Issues" src="https://img.shields.io/github/issues/SuperagenticAI/rlm-code"></a>
-  <a href="https://github.com/SuperagenticAI/rlm-code/pulls"><img alt="GitHub Pull Requests" src="https://img.shields.io/github/issues-pr/SuperagenticAI/rlm-code"></a>
-</p>
+[![PyPI Version](https://img.shields.io/pypi/v/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![Python Versions](https://img.shields.io/pypi/pyversions/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![PyPI Wheel](https://img.shields.io/pypi/wheel/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![License](https://img.shields.io/pypi/l/rlm-code.svg)](https://pypi.org/project/rlm-code/)
+[![CI](https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/ci.yml)
+[![Pre-commit](https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/pre-commit.yml)
+[![Docs Deploy](https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/deploy-docs.yml)
+[![Release](https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml/badge.svg?branch=main)](https://github.com/SuperagenticAI/rlm-code/actions/workflows/release.yml)
+[![Docs](https://img.shields.io/badge/Docs-RLM%20Code-ff7a18.svg?logo=readthedocs&logoColor=white)](https://superagenticai.github.io/rlm-code/)
+[![GitHub Stars](https://img.shields.io/github/stars/SuperagenticAI/rlm-code.svg)](https://github.com/SuperagenticAI/rlm-code/stargazers)
+[![GitHub Issues](https://img.shields.io/github/issues/SuperagenticAI/rlm-code.svg)](https://github.com/SuperagenticAI/rlm-code/issues)
+[![GitHub Pull Requests](https://img.shields.io/github/issues-pr/SuperagenticAI/rlm-code.svg)](https://github.com/SuperagenticAI/rlm-code/pulls)
 
 **Run LLM-powered agents in a REPL loop, benchmark them, and compare results.**
 
@@ -27,6 +25,34 @@ RLM Code implements the [Recursive Language Models](https://arxiv.org/abs/2502.0
 
 RLM Code wraps this algorithm in an interactive terminal UI with built-in benchmarks, trajectory replay, and observability.
 
+## Release v0.1.6
+
+This release adds the new CodeMode path as an opt-in harness strategy.
+
+- New harness strategy: `strategy=codemode` (default remains `strategy=tool_call`)
+- MCP bridge flow for CodeMode: `search_tools` -> typed tool surface -> `call_tool_chain`
+- Guardrails before execution: blocked API classes plus timeout/size/tool-call caps
+- Benchmark telemetry for side-by-side comparison: `tool_call` vs `codemode`
+- Dedicated docs section for CodeMode: quickstart, architecture, guardrails, evaluation
+
+Example:
+
+```text
+/harness run "implement feature and add tests" steps=8 mcp=on strategy=codemode mcp_server=codemode
+```
+
+## Documentation
+
+<p align="center">
+  <a href="https://superagenticai.github.io/rlm-code/">
+    <img alt="Read the RLM Code Docs" src="https://img.shields.io/badge/Read%20the%20Docs-RLM%20Code-ff7a18?style=for-the-badge&logo=readthedocs&logoColor=white">
+  </a>
+</p>
+
+<p align="center">
+  <a href="https://superagenticai.github.io/rlm-code/"><strong>Open the full documentation</strong></a>
+</p>
+
 ## Install
 
 ```bash
@@ -306,7 +332,7 @@ rlm_code/
   harness/          # Tool-using coding harness (/harness)
 ```
 
-## Documentation
+## Resources
 
 Full docs: https://superagenticai.github.io/rlm-code/
 

{rlm_code-0.1.5 → rlm_code-0.1.6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "rlm-code"
-version = "0.1.5"
+version = "0.1.6"
 description = "RLM Code: Research Playground & Evaluation OS for Recursive Language Model Agentic Systems"
 readme = "README.md"
 license = "Apache-2.0"

{rlm_code-0.1.5 → rlm_code-0.1.6}/rlm_code/__init__.py

@@ -5,5 +5,5 @@ This package provides tools for creating, managing, and optimizing DSPy componen
 through natural language interactions.
 """
 
-__version__ = "0.1.5"
+__version__ = "0.1.6"
 __author__ = "Super Agentic AI"

{rlm_code-0.1.5 → rlm_code-0.1.6}/rlm_code/commands/slash_commands.py

@@ -112,6 +112,7 @@ class SlashCommandHandler:
         self.rlm_runner = RLMRunner(
             llm_connector=self.llm_connector,
             execution_engine=self.execution_engine,
+            mcp_manager=self.mcp_manager,
             reward_profile=reward_profile,
             benchmark_pack_paths=benchmark_pack_paths,
         )
@@ -1442,7 +1443,7 @@ class SlashCommandHandler:
         Usage:
             /harness tools [mcp=on|off]
             /harness doctor
-            /harness run <task> [steps=N] [mcp=on|off] [tools=name[,name2]]
+            /harness run <task> [steps=N] [mcp=on|off] [mcp_server=name] [strategy=tool_call|codemode] [tools=name[,name2]]
         """
         if not args or args[0].lower() in {"help", "--help"}:
             console.print()
@@ -1450,7 +1451,8 @@ class SlashCommandHandler:
             console.print("  [yellow]/harness tools [mcp=on|off][/yellow]")
             console.print("  [yellow]/harness doctor[/yellow]")
             console.print(
-                "  [yellow]/harness run <task> [steps=N] [mcp=on|off] [tools=name[,name2]][/yellow]"
+                "  [yellow]/harness run <task> [steps=N] [mcp=on|off] [mcp_server=name] "
+                "[strategy=tool_call|codemode] [tools=name[,name2]][/yellow]"
             )
             console.print()
             return
@@ -1555,6 +1557,8 @@ class SlashCommandHandler:
             include_mcp = True
             max_steps = 10
             allowlist: list[str] | None = None
+            strategy = "tool_call"
+            mcp_server: str | None = None
             task_tokens: list[str] = []
 
             for token in args[1:]:
@@ -1568,6 +1572,16 @@ class SlashCommandHandler:
                 elif lowered.startswith("mcp="):
                     value = token.split("=", 1)[1].strip().lower()
                     include_mcp = value not in {"off", "false", "0", "no"}
+                elif lowered.startswith("mcp_server="):
+                    mcp_server = token.split("=", 1)[1].strip() or None
+                elif lowered.startswith("strategy="):
+                    raw_strategy = token.split("=", 1)[1].strip().lower().replace("-", "_")
+                    if raw_strategy not in {"tool_call", "codemode"}:
+                        show_error_message(
+                            "Invalid strategy value. Use strategy=tool_call|codemode."
+                        )
+                        return
+                    strategy = raw_strategy
                 elif lowered.startswith("tools="):
                     raw = token.split("=", 1)[1].strip()
                     parsed = [part.strip() for part in raw.split(",") if part.strip()]
@@ -1578,15 +1592,25 @@ class SlashCommandHandler:
             task = " ".join(task_tokens).strip()
             if not task:
                 show_error_message(
-                    "Usage: /harness run <task> [steps=N] [mcp=on|off] [tools=name[,name2]]"
+                    "Usage: /harness run <task> [steps=N] [mcp=on|off] [mcp_server=name] "
+                    "[strategy=tool_call|codemode] [tools=name[,name2]]"
                 )
                 return
+            if strategy == "codemode" and not include_mcp:
+                show_warning_message("strategy=codemode requires mcp=on. Enabling MCP.")
+                include_mcp = True
+            if strategy == "codemode" and allowlist:
+                show_warning_message("tools=... allowlist is ignored for strategy=codemode.")
+                allowlist = None
 
             console.print()
             console.print("[bold cyan]🛠 Running Harness[/bold cyan]")
             console.print(f"  Task: [cyan]{task}[/cyan]")
             console.print(f"  Max steps: [cyan]{max_steps}[/cyan]")
             console.print(f"  MCP tools: [cyan]{'on' if include_mcp else 'off'}[/cyan]")
+            console.print(f"  Strategy: [cyan]{strategy}[/cyan]")
+            if mcp_server:
+                console.print(f"  MCP server: [cyan]{mcp_server}[/cyan]")
             if allowlist:
                 console.print(f"  Tool allowlist: [cyan]{', '.join(allowlist)}[/cyan]")
             console.print()
@@ -1596,6 +1620,8 @@ class SlashCommandHandler:
                 max_steps=max_steps,
                 include_mcp=include_mcp,
                 tool_allowlist=allowlist,
+                strategy=strategy,
+                mcp_server=mcp_server,
             )
 
             self.current_context["harness_last_response"] = result.final_response
@@ -1659,7 +1685,7 @@ class SlashCommandHandler:
 
         Usage:
             /rlm run <task> [steps=N] [timeout=N] [branch=N] [depth=N] [children=N] [parallel=N] [budget=N] [framework=<see /rlm frameworks>] [env=generic|dspy|pure_rlm] [sub=provider/model]
-            /rlm bench [list|preset=name] [mode=native|harness|direct-llm] [pack=path[,path2]] [limit=N] [steps=N] [timeout=N] [branch=N] [framework=<see /rlm frameworks>] [env=generic|dspy|pure_rlm] [sub=provider/model]
+            /rlm bench [list|preset=name] [mode=native|harness|direct-llm] [strategy=tool_call|codemode] [mcp=on|off] [mcp_server=name] [pack=path[,path2]] [limit=N] [steps=N] [timeout=N] [branch=N] [framework=<see /rlm frameworks>] [env=generic|dspy|pure_rlm] [sub=provider/model]
             /rlm bench compare [candidate=<id|path|latest>] [baseline=<id|path|previous>] [min_reward_delta=N] [min_completion_delta=N] [max_steps_increase=N]
             /rlm bench validate [candidate=<id|path|latest>] [baseline=<id|path|previous>] [min_reward_delta=N] [min_completion_delta=N] [max_steps_increase=N] [--json]
             /rlm bench report [candidate=<id|path|latest>] [baseline=<id|path|previous>] [format=markdown|csv|json] [output=path]
@@ -1687,6 +1713,7 @@ class SlashCommandHandler:
             )
             console.print(
                 "  [yellow]/rlm bench [list|preset=name] [mode=native|harness|direct-llm] "
+                "[strategy=tool_call|codemode] [mcp=on|off] [mcp_server=name] "
                 "[pack=path[,path2]] [limit=N] [steps=N] "
                 f"[timeout=N] [branch=N] [framework={framework_opts}] [env=generic|dspy|pure_rlm] [sub=provider/model][/yellow]"
             )
@@ -2521,6 +2548,9 @@ class SlashCommandHandler:
             environment: str | None = None
             sub_model: str | None = None
             sub_provider: str | None = None
+            include_mcp = False
+            mcp_server: str | None = None
+            harness_strategy = "tool_call"
 
             for token in args[1:]:
                 lowered = token.lower()
@@ -2537,6 +2567,19 @@ class SlashCommandHandler:
                         )
                         return
                     mode = resolved_mode
+                elif lowered.startswith("mcp="):
+                    value = token.split("=", 1)[1].strip().lower()
+                    include_mcp = value not in {"off", "false", "0", "no"}
+                elif lowered.startswith("strategy="):
+                    strategy_token = token.split("=", 1)[1].strip().lower().replace("-", "_")
+                    if strategy_token not in {"tool_call", "codemode"}:
+                        show_error_message(
+                            "Invalid strategy value. Use strategy=tool_call|codemode."
+                        )
+                        return
+                    harness_strategy = strategy_token
+                elif lowered.startswith("mcp_server="):
+                    mcp_server = token.split("=", 1)[1].strip() or None
                 elif lowered.startswith("pack="):
                     raw_paths = token.split("=", 1)[1].strip()
                     if not raw_paths:
@@ -2593,8 +2636,10 @@ class SlashCommandHandler:
                 else:
                     show_error_message(
                         "Usage: /rlm bench [list|preset=name] [mode=native|harness|direct-llm] "
+                        "[strategy=tool_call|codemode] [mcp=on|off] [mcp_server=name] "
                         "[pack=path[,path2]] [limit=N] "
-                        f"[steps=N] [timeout=N] [branch=N] [framework={framework_opts}] [env=generic|dspy|pure_rlm] [sub=provider/model]\n"
+                        f"[steps=N] [timeout=N] [branch=N] [framework={framework_opts}] "
+                        "[env=generic|dspy|pure_rlm] [sub=provider/model]\n"
                         "       /rlm bench compare [candidate=<id|path|latest>] [baseline=<id|path|previous>] ...\n"
                         "       /rlm bench validate [candidate=<id|path|latest>] [baseline=<id|path|previous>] ...\n"
                         "       /rlm bench report [candidate=<id|path|latest>] [baseline=<id|path|previous>] "
@@ -2602,6 +2647,30 @@ class SlashCommandHandler:
                     )
                     return
 
+            if mode == "harness" and harness_strategy == "codemode" and not include_mcp:
+                show_warning_message("strategy=codemode requires mcp=on. Enabling MCP.")
+                include_mcp = True
+
+            if mode != "harness" and include_mcp:
+                show_warning_message("mcp=on is only used for mode=harness. Ignoring MCP settings.")
+                include_mcp = False
+                mcp_server = None
+            elif mode != "harness" and mcp_server:
+                show_warning_message(
+                    "mcp_server is only used for mode=harness with mcp=on. Ignoring."
+                )
+                mcp_server = None
+            elif mode == "harness" and mcp_server and not include_mcp:
+                show_warning_message(
+                    "mcp_server provided but mcp=off. MCP server filter will be ignored."
+                )
+                mcp_server = None
+            if mode != "harness" and harness_strategy != "tool_call":
+                show_warning_message(
+                    "strategy is only used for mode=harness. Resetting to tool_call."
+                )
+                harness_strategy = "tool_call"
+
             if list_only:
                 try:
                     rows = self.rlm_runner.benchmark_presets(pack_paths=pack_paths_override)
@@ -2681,6 +2750,11 @@ class SlashCommandHandler:
             if timeout is not None:
                 console.print(f"  Override timeout: [cyan]{timeout}s[/cyan]")
             console.print(f"  Branch width: [cyan]{branch_width}[/cyan]")
+            if mode == "harness":
+                console.print(f"  Harness strategy: [cyan]{harness_strategy}[/cyan]")
+                console.print(f"  Harness MCP: [cyan]{'on' if include_mcp else 'off'}[/cyan]")
+                if include_mcp and mcp_server:
+                    console.print(f"  Harness MCP server: [cyan]{mcp_server}[/cyan]")
             if pack_paths_override:
                 console.print(f"  Benchmark packs: [cyan]{', '.join(pack_paths_override)}[/cyan]")
             if environment:
@@ -2704,6 +2778,9 @@ class SlashCommandHandler:
                     branch_width=branch_width,
                     sub_model=sub_model,
                     sub_provider=sub_provider,
+                    include_mcp=include_mcp,
+                    mcp_server=mcp_server,
+                    harness_strategy=harness_strategy,
                     pack_paths=pack_paths_override,
                 )
             except ValueError as exc:
@@ -4413,7 +4490,7 @@ class SlashCommandHandler:
 
 [bold magenta]RLM Workflows:[/bold magenta]
   [yellow]/rlm run[/yellow] <task> [steps=N] [timeout=N] [branch=N] [depth=N] [children=N] [parallel=N] [budget=N] [framework=native|dspy-rlm|adk-rlm|pydantic-ai|google-adk|deepagents] [env=generic|dspy|pure_rlm] [sub=provider/model] - Run an RLM coding episode
-  [yellow]/rlm bench[/yellow] [list|preset=name] [mode=native|harness|direct-llm] [pack=path[,path2]] [limit=N] [steps=N] [timeout=N] [branch=N] [framework=native|dspy-rlm|adk-rlm|pydantic-ai|google-adk|deepagents] [env=generic|dspy|pure_rlm] [sub=provider/model] - Run benchmark preset
+  [yellow]/rlm bench[/yellow] [list|preset=name] [mode=native|harness|direct-llm] [strategy=tool_call|codemode] [mcp=on|off] [mcp_server=name] [pack=path[,path2]] [limit=N] [steps=N] [timeout=N] [branch=N] [framework=native|dspy-rlm|adk-rlm|pydantic-ai|google-adk|deepagents] [env=generic|dspy|pure_rlm] [sub=provider/model] - Run benchmark preset
   [yellow]/rlm bench compare[/yellow] [candidate=<id|path|latest>] [baseline=<id|path|previous>] [min_reward_delta=N] [min_completion_delta=N] [max_steps_increase=N] - Gate regressions
   [yellow]/rlm bench validate[/yellow] [candidate=<id|path|latest>] [baseline=<id|path|previous>] [min_reward_delta=N] [min_completion_delta=N] [max_steps_increase=N] [--json] - CI-style gate output
   [yellow]/rlm bench report[/yellow] [candidate=<id|path|latest>] [baseline=<id|path|previous>] [format=markdown|csv|json] [output=path] - Export compare report
@@ -4431,7 +4508,7 @@ class SlashCommandHandler:
   [yellow]/rlm observability[/yellow]                   - Show local/MLflow observability sink status
   [yellow]/harness tools[/yellow] [mcp=on|off]         - List coding harness tools (local + MCP)
   [yellow]/harness doctor[/yellow]                     - Show harness tool coverage report
-  [yellow]/harness run[/yellow] <task> [steps=N] [mcp=on|off] [tools=name[,name2]] - Run tool-using coding harness
+  [yellow]/harness run[/yellow] <task> [steps=N] [mcp=on|off] [mcp_server=name] [strategy=tool_call|codemode] [tools=name[,name2]] - Run tool-using coding harness
 
 [bold magenta]Optimization (GEPA):[/bold magenta]
   [yellow]/optimize-start[/yellow] [budget]           - Start GEPA optimization workflow

{rlm_code-0.1.5 → rlm_code-0.1.6}/rlm_code/core/config.py

@@ -102,7 +102,7 @@ class SandboxAppleContainerConfig:
 class SandboxConfig:
     """Execution sandbox runtime configuration."""
 
-    runtime: str = "docker"  # local | docker | apple-container | daytona | e2b
+    runtime: str = "docker"  # local | monty | docker | apple-container | daytona | e2b
     default_timeout_seconds: int = 30
     memory_limit_mb: int = 512
     allowed_mount_roots: list[str] = field(