{ "meta": { "finished_at": "2026-10-15T08:38:07+0900", "trials": 5, "plan": { "order": [ "mcp-1", "angelo-1", "mcp-2", "angelo-2", "mcp-3", "angelo-3", "mcp-4", "angelo-4", "mcp-5", "angelo-5" ], "runtime": "claude", "model": "claude-opus-5-5", "effort": "low", "fast": false, "task_set": "oneshot-mini-camera", "gate_version": null, "input_isolation": "on", "after_failure": "skip", "refusal_fallback": "disabled", "rate_limits": "a provider rate or usage limit makes its trial invalid (never a task failure) and stops the run; invalid trials are left out of every pass rate and counted", "angelo": { "observation": "turn-context", "trusted_scripts": true, "approval": "default", "always_allow_scripts": false, "context_rebase_tokens": 0, "cache_ttl": "product default", "stable_workspace": true, "returning_user": false, "opening_wait_s": 0 }, "mcp_tasks": [ "brief" ], "angelo_tasks": [ "brief" ], "scenario": { "scenario": "oneshot_mini_camera", "title": "A new scene camera aimed at the cube", "turns": [ "brief" ], "checks": 4, "oneshot": { "messages_per_side": 1, "limits": { "wall_s": 300, "job_wait_s": 0, "cost_usd": 3 }, "final_still": null, "ground_truth": "stand-in" }, "references_per_turn": {}, "reference_digest": "22ea44460463babd", "references": "", "scored_views": [ "front", "side", "top", "hero" ], "frame": 1 }, "python_approvals": "approve", "demo_record": false, "stage_copies": false, "render_device": "factory preference", "oneshot": { "messages_per_side": 1, "follow_ups": "none: the harness never types again", "limits": { "wall_s": 300, "job_wait_s": 0, "cost_usd": 3 }, "other_side_client": [ "every tool of the Blender MCP server (model-written Python included, unrestricted)", "Claude Code's Read tool on image files anywhere, and its read-only file tools in its working folder and in the trial folder (stock Blender's working and temporary folder): it can open the images it rendered", "the reference images also as files in its working folder", "anything else that would prompt (shell, edits, web) is denied and counted: nobody answers prompts" ], "grading": "closed checks and pixel statistics only; no model-judged score is counted" }, "billed_prompts": { "mcp": 5, "angelo": 5 }, "billing": "Each prompt is one billed message on the selected Claude account; a message can contain several model calls. Failed tasks are recorded, never retried." }, "method": [ "Same Claude account on both sides: the Angelo connection's native Claude Code profile (CLAUDE_CONFIG_DIR).", "Same pinned model, --effort low and Fast off on both sides; the model each side actually used is read back from Claude Code's own output (MCP: init/result; Angelo: its session transcript).", "Same five prompts from eval_live_tasks.build_tasks(), one continuing chat per trial, factory default-cube scene.", "MCP side: stock Blender with a private copy of Blender Lab's MCP add-on (default 0.25 s active / 1 s idle socket polling), Blender Lab's installed MCP server, headless Claude Code with its default system prompt and built-in tools; --allowedTools mcp__blender (all of that server's tools), --permission-prompts none (any other prompt is denied and reported), --strict-mcp-config, --no-session-persistence.", "Angelo side: the unchanged live-eval harness (approvals on Ask; its test-only rule approves typed bounded scene tools and declines reviewed Python, file writes and the rest).", "Wall time = prompt sent -> final result (MCP: Claude Code's result line; Angelo: the request receipt's total_ms). The first prompt of each trial includes the runtime launch on both sides: by default Angelo readies nothing before it (ANGELO_EVAL_OPENING_WAIT 0; each trial's report records its opening).", "Before every MCP prompt the driver waits 6 s after its own last add-on request, so the add-on is on its default idle polling, as after a human pause; Angelo sends its next prompt immediately (it does not poll).", "Correctness: after each turn the driver (not the model) grades the scene with the Angelo harness's own predicates (verify() + eval_live_tasks checks + the session's scene_digest/bind_checks/smooth_shaded); MCP grading runs over the add-on socket. Failed checks count as failures; their time is still reported.", "Both sides start from a minimal environment (PATH, HOME, USER, locale; DISABLE_AUTOUPDATER=1) so no inherited ANTHROPIC_*/CLAUDE_CODE_* routing or session variable reaches either runtime.", "Trials alternate MCP, Angelo, MCP, Angelo...; each trial is a fresh process pair, scene and session; nothing is retried.", "Scenario set: one continuing chat of scripted turns with the same prompts and byte-identical reference images on both sides (Angelo: chat attachments; Claude Code: base64 image blocks; Codex: local images); each turn is graded by closed host checks (scenario_checks) and the scene is digested after it.", "Scenario approvals: Angelo's tester also approves reviewed Python (--python-approvals approve; review, then approve, as a person clicks twice) and counts each approval; the MCP side has no approval gate.", "At the end both sides save a copy (no thumbnail), report scene hygiene and run the Undo probe; the saved scene is captured and scored against the ground truth in stock Blender (scenario-scorecard.json, SCENARIO-REPORT.md, scenario-report.html)." ], "threats": [ "Safety asymmetry: the MCP side executes model-written raw Python in the live scene with no staging, verification gate, approval or single-step Undo; Angelo stages typed operations, verifies and commits one Undo step. Equal latency would not mean equal risk, and Angelo's checks cost time the MCP side does not spend.", "Prompt wording: the prompts were written for Angelo's harness (explicit names, exact numbers). Other phrasings, vaguer requests or larger scenes may favour either side; five short tasks are not a general workload.", "Small sample: n = 5 trials per side and task; medians and p75 are reported with min/max, but differences of a few seconds are within run-to-run spread. No significance test is claimed.", "Network and service variance: both sides call the same hosted model over the internet; queueing, retries and time-of-day load move wall times. Alternation spreads drift across sides but does not remove it.", "Shared machine: other agents' builds/tests ran concurrently; load averages are recorded per trial.", "Context asymmetry by design: Angelo attaches a bounded scene observation to each prompt; the MCP side must call tools to see the scene. Angelo's Claude Code runs with a restricted tool set and structured output, the MCP side with Claude Code's full default tool set and system prompt. This is each product as shipped.", "Stock Blender 5.2.0 (MCP) vs the Angelo 5.2.2 build: different binaries; Blender-side execution is milliseconds on both, so this does not drive end-to-end time, but window focus behaviour differs (stock Blender may come to the front).", "The 6 s settle and the add-on's 1 s idle polling are the MCP add-on's defaults; a user who lowers the idle interval would see somewhat faster MCP tool round trips.", "Cost figures are Claude Code's own estimates, not a bill; this is a subscription account.", "Prompt cache: the MCP side reuses one empty working directory, so after its first trial its system prompt is served from the prompt cache. Each Angelo trial has a new workspace path (Claude's working directory) unless the run used --angelo-stable-workspace (see plan.angelo); see the first-message cache writes under Observations. Without it this mainly lowers the MCP cost estimate.", "Round-count variance: Angelo's number of model rounds depends on the model's own verify choice per plan, and Claude Code defers MCP tools behind ToolSearch on first use; both are counted as they happened, and n = 5 may under-sample how often each occurs.", "Wall-clock boundaries differ slightly: the MCP clock ends when the driver reads Claude Code's result line; Angelo's receipt total_ms ends inside the controller (the harness's send-to-idle time is also recorded)." ], "versions": { "runtime": "claude", "claude_code": "2.1.293 (Claude Code)", "codex": null, "codex_account": null, "models_reported": { "mcp": [ "claude-opus-5-5" ], "angelo": [ "claude-opus-5-5" ] }, "requested_model": "claude-opus-5-5", "effort": "low", "fast_mode": false, "mcp_init_model": [ "claude-opus-5-5" ], "mcp_server": { "extension_manifest_version": null, "path": "" }, "mcp_addon": "1.0.0", "stock_blender": { "blender_version": "5.2.0 LTS", "build_hash": "fbe6228777e7" }, "angelo_binary": [ "Blender 5.2.2 LTS", "build date: 2026-10-14", "build hash: 3092c5ccba45", "build branch: angelo-v1.0-release", "build type: Release" ], "angelo_checkout_commit": "3092c5ccba457d380943ce52221596b3f19434c7", "angelo_checkout_branch": "", "angelo_checkout_dirty_files": 0, "angelo_python_source": "the add-on and harness Python load from this checkout ()", "machine": { "platform": "macOS-26.3-arm64-arm-64bit-Mach-O", "cpu": "Apple M4 Pro", "cores": 14, "memory_bytes": "51539607552" } }, "trial_meta": [ { "side": "mcp", "trial": 1, "port": 39076, "success": true, "gate_version": null, "load_average_start": [ 3.421875, 3.0810546875, 3.00634765625 ], "started_at": "2026-10-15T08:34:57+0900", "blender": { "ok": true, "port": 39076, "outcome": [ "FINISHED" ], "blender_version": "5.2.0 LTS", "build_hash": "fbe6228777e7", "online_access": true, "addon_module": "bl_ext.user_default.mcp", "addon_version": "1.0.0" }, "mcp_server": { "server_info": { "name": "blender-mcp", "version": "1.27.0" }, "initialize_ms": 254.85, "tools": [ "execute_blender_code", "execute_blender_code_for_cli", "get_blendfile_summary_datablocks", "get_blendfile_summary_datablocks_for_cli", "get_blendfile_summary_missing_files", "get_blendfile_summary_missing_files_for_cli", "get_blendfile_summary_of_linked_libraries", "get_blendfile_summary_of_linked_libraries_for_cli", "get_blendfile_summary_path_info", "get_blendfile_summary_path_info_for_cli", "get_blendfile_summary_usage_guess", "get_blendfile_summary_usage_guess_for_cli", "get_object_detail_summary", "get_objects_summary", "get_python_api_docs", "get_screenshot_of_area_as_image", "get_screenshot_of_window_as_image", "get_screenshot_of_window_as_json", "jump_to_tab_by_name", "jump_to_tab_by_space_type", "jump_to_view3d_object_by_name", "jump_to_view3d_object_data_by_name", "render_thumbnail_to_path", "render_viewport_to_path", "search_api_docs", "search_manual_docs" ] }, "claude_command": [ "", "-p", "--input-format", "stream-json", "--output-format", "stream-json", "--verbose", "--include-partial-messages", "--model", "claude-opus-5-5", "--effort", "low", "--settings", "{\"fastMode\": false}", "--mcp-config", "/h2h-opus-oneshot-mini-camera/mcp-trial-01/mcp-config.json", "--strict-mcp-config", "--allowedTools", "mcp__blender,mcp__blender__execute_blender_code,mcp__blender__execute_blender_code_for_cli,mcp__blender__get_blendfile_summary_datablocks,mcp__blender__get_blendfile_summary_datablocks_for_cli,mcp__blender__get_blendfile_summary_missing_files,mcp__blender__get_blendfile_summary_missing_files_for_cli,mcp__blender__get_blendfile_summary_of_linked_libraries,mcp__blender__get_blendfile_summary_of_linked_libraries_for_cli,mcp__blender__get_blendfile_summary_path_info,mcp__blender__get_blendfile_summary_path_info_for_cli,mcp__blender__get_blendfile_summary_usage_guess,mcp__blender__get_blendfile_summary_usage_guess_for_cli,mcp__blender__get_object_detail_summary,mcp__blender__get_objects_summary,mcp__blender__get_python_api_docs,mcp__blender__get_screenshot_of_area_as_image,mcp__blender__get_screenshot_of_window_as_image,mcp__blender__get_screenshot_of_window_as_json,mcp__blender__jump_to_tab_by_name,mcp__blender__jump_to_tab_by_space_type,mcp__blender__jump_to_view3d_object_by_name,mcp__blender__jump_to_view3d_object_data_by_name,mcp__blender__render_thumbnail_to_path,mcp__blender__render_viewport_to_path,mcp__blender__search_api_docs,mcp__blender__search_manual_docs,Read(//**/*.png),Read(//**/*.jpg),Read(//**/*.jpeg),Read(//**/*.exr),Read(//**/*.webp),Read(//**/*.tif),Read(//**/*.tiff),Read(//**/*.bmp)", "--permission-prompts", "none", "--no-session-persistence", "--add-dir", "/h2h-opus-oneshot-mini-camera/mcp-trial-01" ], "oneshot": { "messages": 1, "refused_sends": 0, "stop": null, "sent_wall": 1792020905.4564931, "cost_cap_usd": 3, "watch": { "usage": { "input_tokens": 110614, "cached_input_tokens": 85768, "cache_write_input_tokens": 24840, "output_tokens": 88, "requests": 3 }, "cost_usd": 0.1431 }, "allowances": [ "every tool of the Blender MCP server (model-written Python included, unrestricted)", "Claude Code's Read tool on image files anywhere, and its read-only file tools in its working folder and in the trial folder (stock Blender's working and temporary folder): it can open the images it rendered", "the reference images also as files in its working folder", "anything else that would prompt (shell, edits, web) is denied and counted: nobody answers prompts" ] }, "scenario": { "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb", "saved": "/h2h-opus-oneshot-mini-camera/mcp-trial-01/final.blend", "undo": { "undo_steps": 0, "stopped": "nothing left to undo", "landed": [], "reaches_start": false, "first_undo_returns_previous_turn": false, "first_undo_lands_on": null, "turn_boundaries_reachable": 0, "turns": 1, "steps_per_turn": {} } }, "load_average_end": [ 3.48291015625, 3.12158203125, 3.021484375 ] }, { "side": "angelo", "trial": 1, "harness_command": [ "/Applications/Blender.app/Contents/Resources/5.2/python/bin/python3.13", "", "--authorize-billed-runs", "--connection-id", "", "--source-connections-dir", "", "--blender", "", "--effort", "low", "--model", "claude-opus-5-5", "--trials", "1", "--task-timeout", "300", "--output", "/h2h-opus-oneshot-mini-camera/angelo-trial-01", "--task-set", "oneshot-mini-camera", "--long-tasks", "40", "--observation", "turn-context", "--context-rebase-tokens", "0", "--input-isolation", "on", "--after-failure", "skip", "--python-approvals", "approve", "--trusted-scripts", "--stable-state-dir", "/h2h-opus-oneshot-mini-camera/angelo-state" ], "load_average_start": [ 3.48291015625, 3.12158203125, 3.021484375 ], "started_at": "2026-10-15T08:35:20+0900", "exit_code": 0, "elapsed_s": 8.255, "load_average_end": [ 3.44384765625, 3.119140625, 3.02099609375 ], "version": "5.2.2 LTS", "binary": "", "aborted": null, "success": true, "invalid": null, "scenario": { "python_approvals": "approve", "oneshot": { "limits": { "wall_s": 300, "job_wait_s": 0, "cost_usd": 3 }, "cost_cap_usd": 3 }, "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98", "saved": "/h2h-opus-oneshot-mini-camera/angelo-trial-01/trial-01/qa/final.blend", "undo": { "undo_steps": 1, "stopped": "reached the start", "landed": [ 0 ], "reaches_start": true, "first_undo_returns_previous_turn": true, "first_undo_lands_on": 0, "turn_boundaries_reachable": 1, "turns": 1, "steps_per_turn": { "1-1": 1 } } } }, { "side": "mcp", "trial": 2, "port": 39145, "success": true, "gate_version": null, "load_average_start": [ 3.61572265625, 3.16455078125, 3.03759765625 ], "started_at": "2026-10-15T08:35:36+0900", "blender": { "ok": true, "port": 39145, "outcome": [ "FINISHED" ], "blender_version": "5.2.0 LTS", "build_hash": "fbe6228777e7", "online_access": true, "addon_module": "bl_ext.user_default.mcp", "addon_version": "1.0.0" }, "mcp_server": { "server_info": { "name": "blender-mcp", "version": "1.27.0" }, "initialize_ms": 251.451, "tools": [ "execute_blender_code", "execute_blender_code_for_cli", "get_blendfile_summary_datablocks", "get_blendfile_summary_datablocks_for_cli", "get_blendfile_summary_missing_files", "get_blendfile_summary_missing_files_for_cli", "get_blendfile_summary_of_linked_libraries", "get_blendfile_summary_of_linked_libraries_for_cli", "get_blendfile_summary_path_info", "get_blendfile_summary_path_info_for_cli", "get_blendfile_summary_usage_guess", "get_blendfile_summary_usage_guess_for_cli", "get_object_detail_summary", "get_objects_summary", "get_python_api_docs", "get_screenshot_of_area_as_image", "get_screenshot_of_window_as_image", "get_screenshot_of_window_as_json", "jump_to_tab_by_name", "jump_to_tab_by_space_type", "jump_to_view3d_object_by_name", "jump_to_view3d_object_data_by_name", "render_thumbnail_to_path", "render_viewport_to_path", "search_api_docs", "search_manual_docs" ] }, "claude_command": [ "", "-p", "--input-format", "stream-json", "--output-format", "stream-json", "--verbose", "--include-partial-messages", "--model", "claude-opus-5-5", "--effort", "low", "--settings", "{\"fastMode\": false}", "--mcp-config", "/h2h-opus-oneshot-mini-camera/mcp-trial-02/mcp-config.json", "--strict-mcp-config", "--allowedTools", "mcp__blender,mcp__blender__execute_blender_code,mcp__blender__execute_blender_code_for_cli,mcp__blender__get_blendfile_summary_datablocks,mcp__blender__get_blendfile_summary_datablocks_for_cli,mcp__blender__get_blendfile_summary_missing_files,mcp__blender__get_blendfile_summary_missing_files_for_cli,mcp__blender__get_blendfile_summary_of_linked_libraries,mcp__blender__get_blendfile_summary_of_linked_libraries_for_cli,mcp__blender__get_blendfile_summary_path_info,mcp__blender__get_blendfile_summary_path_info_for_cli,mcp__blender__get_blendfile_summary_usage_guess,mcp__blender__get_blendfile_summary_usage_guess_for_cli,mcp__blender__get_object_detail_summary,mcp__blender__get_objects_summary,mcp__blender__get_python_api_docs,mcp__blender__get_screenshot_of_area_as_image,mcp__blender__get_screenshot_of_window_as_image,mcp__blender__get_screenshot_of_window_as_json,mcp__blender__jump_to_tab_by_name,mcp__blender__jump_to_tab_by_space_type,mcp__blender__jump_to_view3d_object_by_name,mcp__blender__jump_to_view3d_object_data_by_name,mcp__blender__render_thumbnail_to_path,mcp__blender__render_viewport_to_path,mcp__blender__search_api_docs,mcp__blender__search_manual_docs,Read(//**/*.png),Read(//**/*.jpg),Read(//**/*.jpeg),Read(//**/*.exr),Read(//**/*.webp),Read(//**/*.tif),Read(//**/*.tiff),Read(//**/*.bmp)", "--permission-prompts", "none", "--no-session-persistence", "--add-dir", "/h2h-opus-oneshot-mini-camera/mcp-trial-02" ], "oneshot": { "messages": 1, "refused_sends": 0, "stop": null, "sent_wall": 1792020944.319686, "cost_cap_usd": 3, "watch": { "usage": { "input_tokens": 110618, "cached_input_tokens": 101936, "cache_write_input_tokens": 8676, "output_tokens": 88, "requests": 3 }, "cost_usd": 0.0656 }, "allowances": [ "every tool of the Blender MCP server (model-written Python included, unrestricted)", "Claude Code's Read tool on image files anywhere, and its read-only file tools in its working folder and in the trial folder (stock Blender's working and temporary folder): it can open the images it rendered", "the reference images also as files in its working folder", "anything else that would prompt (shell, edits, web) is denied and counted: nobody answers prompts" ] }, "scenario": { "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "17a8d4e2efb700562c05d4dca1d2adae7816dcc5570eac9f52c5397f0e294571" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 0, "issues": 3 }, "final_digest": "17a8d4e2efb700562c05d4dca1d2adae7816dcc5570eac9f52c5397f0e294571", "saved": "/h2h-opus-oneshot-mini-camera/mcp-trial-02/final.blend", "undo": { "undo_steps": 0, "stopped": "nothing left to undo", "landed": [], "reaches_start": false, "first_undo_returns_previous_turn": false, "first_undo_lands_on": null, "turn_boundaries_reachable": 0, "turns": 1, "steps_per_turn": {} } }, "load_average_end": [ 3.736328125, 3.21826171875, 3.05908203125 ] }, { "side": "angelo", "trial": 2, "harness_command": [ "/Applications/Blender.app/Contents/Resources/5.2/python/bin/python3.13", "", "--authorize-billed-runs", "--connection-id", "", "--source-connections-dir", "", "--blender", "", "--effort", "low", "--model", "claude-opus-5-5", "--trials", "1", "--task-timeout", "300", "--output", "/h2h-opus-oneshot-mini-camera/angelo-trial-02", "--task-set", "oneshot-mini-camera", "--long-tasks", "40", "--observation", "turn-context", "--context-rebase-tokens", "0", "--input-isolation", "on", "--after-failure", "skip", "--python-approvals", "approve", "--trusted-scripts", "--stable-state-dir", "/h2h-opus-oneshot-mini-camera/angelo-state" ], "load_average_start": [ 3.736328125, 3.21826171875, 3.05908203125 ], "started_at": "2026-10-15T08:35:58+0900", "exit_code": 0, "elapsed_s": 7.345, "load_average_end": [ 3.62939453125, 3.2109375, 3.05810546875 ], "version": "5.2.2 LTS", "binary": "", "aborted": null, "success": true, "invalid": null, "scenario": { "python_approvals": "approve", "oneshot": { "limits": { "wall_s": 300, "job_wait_s": 0, "cost_usd": 3 }, "cost_cap_usd": 3 }, "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98", "saved": "/h2h-opus-oneshot-mini-camera/angelo-trial-02/trial-01/qa/final.blend", "undo": { "undo_steps": 1, "stopped": "reached the start", "landed": [ 0 ], "reaches_start": true, "first_undo_returns_previous_turn": true, "first_undo_lands_on": 0, "turn_boundaries_reachable": 1, "turns": 1, "steps_per_turn": { "1-1": 1 } } } }, { "side": "mcp", "trial": 3, "port": 39464, "success": true, "gate_version": null, "load_average_start": [ 3.73876953125, 3.240234375, 3.0693359375 ], "started_at": "2026-10-15T08:36:13+0900", "blender": { "ok": true, "port": 39464, "outcome": [ "FINISHED" ], "blender_version": "5.2.0 LTS", "build_hash": "fbe6228777e7", "online_access": true, "addon_module": "bl_ext.user_default.mcp", "addon_version": "1.0.0" }, "mcp_server": { "server_info": { "name": "blender-mcp", "version": "1.27.0" }, "initialize_ms": 251.923, "tools": [ "execute_blender_code", "execute_blender_code_for_cli", "get_blendfile_summary_datablocks", "get_blendfile_summary_datablocks_for_cli", "get_blendfile_summary_missing_files", "get_blendfile_summary_missing_files_for_cli", "get_blendfile_summary_of_linked_libraries", "get_blendfile_summary_of_linked_libraries_for_cli", "get_blendfile_summary_path_info", "get_blendfile_summary_path_info_for_cli", "get_blendfile_summary_usage_guess", "get_blendfile_summary_usage_guess_for_cli", "get_object_detail_summary", "get_objects_summary", "get_python_api_docs", "get_screenshot_of_area_as_image", "get_screenshot_of_window_as_image", "get_screenshot_of_window_as_json", "jump_to_tab_by_name", "jump_to_tab_by_space_type", "jump_to_view3d_object_by_name", "jump_to_view3d_object_data_by_name", "render_thumbnail_to_path", "render_viewport_to_path", "search_api_docs", "search_manual_docs" ] }, "claude_command": [ "", "-p", "--input-format", "stream-json", "--output-format", "stream-json", "--verbose", "--include-partial-messages", "--model", "claude-opus-5-5", "--effort", "low", "--settings", "{\"fastMode\": false}", "--mcp-config", "/h2h-opus-oneshot-mini-camera/mcp-trial-03/mcp-config.json", "--strict-mcp-config", "--allowedTools", "mcp__blender,mcp__blender__execute_blender_code,mcp__blender__execute_blender_code_for_cli,mcp__blender__get_blendfile_summary_datablocks,mcp__blender__get_blendfile_summary_datablocks_for_cli,mcp__blender__get_blendfile_summary_missing_files,mcp__blender__get_blendfile_summary_missing_files_for_cli,mcp__blender__get_blendfile_summary_of_linked_libraries,mcp__blender__get_blendfile_summary_of_linked_libraries_for_cli,mcp__blender__get_blendfile_summary_path_info,mcp__blender__get_blendfile_summary_path_info_for_cli,mcp__blender__get_blendfile_summary_usage_guess,mcp__blender__get_blendfile_summary_usage_guess_for_cli,mcp__blender__get_object_detail_summary,mcp__blender__get_objects_summary,mcp__blender__get_python_api_docs,mcp__blender__get_screenshot_of_area_as_image,mcp__blender__get_screenshot_of_window_as_image,mcp__blender__get_screenshot_of_window_as_json,mcp__blender__jump_to_tab_by_name,mcp__blender__jump_to_tab_by_space_type,mcp__blender__jump_to_view3d_object_by_name,mcp__blender__jump_to_view3d_object_data_by_name,mcp__blender__render_thumbnail_to_path,mcp__blender__render_viewport_to_path,mcp__blender__search_api_docs,mcp__blender__search_manual_docs,Read(//**/*.png),Read(//**/*.jpg),Read(//**/*.jpeg),Read(//**/*.exr),Read(//**/*.webp),Read(//**/*.tif),Read(//**/*.tiff),Read(//**/*.bmp)", "--permission-prompts", "none", "--no-session-persistence", "--add-dir", "/h2h-opus-oneshot-mini-camera/mcp-trial-03" ], "oneshot": { "messages": 1, "refused_sends": 0, "stop": null, "sent_wall": 1792020981.523048, "cost_cap_usd": 3, "watch": { "usage": { "input_tokens": 110614, "cached_input_tokens": 101936, "cache_write_input_tokens": 8672, "output_tokens": 61, "requests": 3 }, "cost_usd": 0.065 }, "allowances": [ "every tool of the Blender MCP server (model-written Python included, unrestricted)", "Claude Code's Read tool on image files anywhere, and its read-only file tools in its working folder and in the trial folder (stock Blender's working and temporary folder): it can open the images it rendered", "the reference images also as files in its working folder", "anything else that would prompt (shell, edits, web) is denied and counted: nobody answers prompts" ] }, "scenario": { "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "17a8d4e2efb700562c05d4dca1d2adae7816dcc5570eac9f52c5397f0e294571" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 0, "issues": 3 }, "final_digest": "17a8d4e2efb700562c05d4dca1d2adae7816dcc5570eac9f52c5397f0e294571", "saved": "/h2h-opus-oneshot-mini-camera/mcp-trial-03/final.blend", "undo": { "undo_steps": 0, "stopped": "nothing left to undo", "landed": [], "reaches_start": false, "first_undo_returns_previous_turn": false, "first_undo_lands_on": null, "turn_boundaries_reachable": 0, "turns": 1, "steps_per_turn": {} } }, "load_average_end": [ 3.06982421875, 3.123046875, 3.03173828125 ] }, { "side": "angelo", "trial": 3, "harness_command": [ "/Applications/Blender.app/Contents/Resources/5.2/python/bin/python3.13", "", "--authorize-billed-runs", "--connection-id", "", "--source-connections-dir", "", "--blender", "", "--effort", "low", "--model", "claude-opus-5-5", "--trials", "1", "--task-timeout", "300", "--output", "/h2h-opus-oneshot-mini-camera/angelo-trial-03", "--task-set", "oneshot-mini-camera", "--long-tasks", "40", "--observation", "turn-context", "--context-rebase-tokens", "0", "--input-isolation", "on", "--after-failure", "skip", "--python-approvals", "approve", "--trusted-scripts", "--stable-state-dir", "/h2h-opus-oneshot-mini-camera/angelo-state" ], "load_average_start": [ 3.06982421875, 3.123046875, 3.03173828125 ], "started_at": "2026-10-15T08:36:39+0900", "exit_code": 0, "elapsed_s": 8.976, "load_average_end": [ 2.98486328125, 3.10205078125, 3.02490234375 ], "version": "5.2.2 LTS", "binary": "", "aborted": null, "success": true, "invalid": null, "scenario": { "python_approvals": "approve", "oneshot": { "limits": { "wall_s": 300, "job_wait_s": 0, "cost_usd": 3 }, "cost_cap_usd": 3 }, "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98", "saved": "/h2h-opus-oneshot-mini-camera/angelo-trial-03/trial-01/qa/final.blend", "undo": { "undo_steps": 1, "stopped": "reached the start", "landed": [ 0 ], "reaches_start": true, "first_undo_returns_previous_turn": true, "first_undo_lands_on": 0, "turn_boundaries_reachable": 1, "turns": 1, "steps_per_turn": { "1-1": 1 } } } }, { "side": "mcp", "trial": 4, "port": 39295, "success": true, "gate_version": null, "load_average_start": [ 2.8330078125, 3.0654296875, 3.01220703125 ], "started_at": "2026-10-15T08:36:55+0900", "blender": { "ok": true, "port": 39295, "outcome": [ "FINISHED" ], "blender_version": "5.2.0 LTS", "build_hash": "fbe6228777e7", "online_access": true, "addon_module": "bl_ext.user_default.mcp", "addon_version": "1.0.0" }, "mcp_server": { "server_info": { "name": "blender-mcp", "version": "1.27.0" }, "initialize_ms": 253.859, "tools": [ "execute_blender_code", "execute_blender_code_for_cli", "get_blendfile_summary_datablocks", "get_blendfile_summary_datablocks_for_cli", "get_blendfile_summary_missing_files", "get_blendfile_summary_missing_files_for_cli", "get_blendfile_summary_of_linked_libraries", "get_blendfile_summary_of_linked_libraries_for_cli", "get_blendfile_summary_path_info", "get_blendfile_summary_path_info_for_cli", "get_blendfile_summary_usage_guess", "get_blendfile_summary_usage_guess_for_cli", "get_object_detail_summary", "get_objects_summary", "get_python_api_docs", "get_screenshot_of_area_as_image", "get_screenshot_of_window_as_image", "get_screenshot_of_window_as_json", "jump_to_tab_by_name", "jump_to_tab_by_space_type", "jump_to_view3d_object_by_name", "jump_to_view3d_object_data_by_name", "render_thumbnail_to_path", "render_viewport_to_path", "search_api_docs", "search_manual_docs" ] }, "claude_command": [ "", "-p", "--input-format", "stream-json", "--output-format", "stream-json", "--verbose", "--include-partial-messages", "--model", "claude-opus-5-5", "--effort", "low", "--settings", "{\"fastMode\": false}", "--mcp-config", "/h2h-opus-oneshot-mini-camera/mcp-trial-04/mcp-config.json", "--strict-mcp-config", "--allowedTools", "mcp__blender,mcp__blender__execute_blender_code,mcp__blender__execute_blender_code_for_cli,mcp__blender__get_blendfile_summary_datablocks,mcp__blender__get_blendfile_summary_datablocks_for_cli,mcp__blender__get_blendfile_summary_missing_files,mcp__blender__get_blendfile_summary_missing_files_for_cli,mcp__blender__get_blendfile_summary_of_linked_libraries,mcp__blender__get_blendfile_summary_of_linked_libraries_for_cli,mcp__blender__get_blendfile_summary_path_info,mcp__blender__get_blendfile_summary_path_info_for_cli,mcp__blender__get_blendfile_summary_usage_guess,mcp__blender__get_blendfile_summary_usage_guess_for_cli,mcp__blender__get_object_detail_summary,mcp__blender__get_objects_summary,mcp__blender__get_python_api_docs,mcp__blender__get_screenshot_of_area_as_image,mcp__blender__get_screenshot_of_window_as_image,mcp__blender__get_screenshot_of_window_as_json,mcp__blender__jump_to_tab_by_name,mcp__blender__jump_to_tab_by_space_type,mcp__blender__jump_to_view3d_object_by_name,mcp__blender__jump_to_view3d_object_data_by_name,mcp__blender__render_thumbnail_to_path,mcp__blender__render_viewport_to_path,mcp__blender__search_api_docs,mcp__blender__search_manual_docs,Read(//**/*.png),Read(//**/*.jpg),Read(//**/*.jpeg),Read(//**/*.exr),Read(//**/*.webp),Read(//**/*.tif),Read(//**/*.tiff),Read(//**/*.bmp)", "--permission-prompts", "none", "--no-session-persistence", "--add-dir", "/h2h-opus-oneshot-mini-camera/mcp-trial-04" ], "oneshot": { "messages": 1, "refused_sends": 0, "stop": null, "sent_wall": 1792021023.419028, "cost_cap_usd": 3, "watch": { "usage": { "input_tokens": 110614, "cached_input_tokens": 101936, "cache_write_input_tokens": 8672, "output_tokens": 88, "requests": 3 }, "cost_usd": 0.0655 }, "allowances": [ "every tool of the Blender MCP server (model-written Python included, unrestricted)", "Claude Code's Read tool on image files anywhere, and its read-only file tools in its working folder and in the trial folder (stock Blender's working and temporary folder): it can open the images it rendered", "the reference images also as files in its working folder", "anything else that would prompt (shell, edits, web) is denied and counted: nobody answers prompts" ] }, "scenario": { "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb", "saved": "/h2h-opus-oneshot-mini-camera/mcp-trial-04/final.blend", "undo": { "undo_steps": 0, "stopped": "nothing left to undo", "landed": [], "reaches_start": false, "first_undo_returns_previous_turn": false, "first_undo_lands_on": null, "turn_boundaries_reachable": 0, "turns": 1, "steps_per_turn": {} } }, "load_average_end": [ 2.7548828125, 3.0283203125, 3 ] }, { "side": "angelo", "trial": 4, "harness_command": [ "/Applications/Blender.app/Contents/Resources/5.2/python/bin/python3.13", "", "--authorize-billed-runs", "--connection-id", "", "--source-connections-dir", "", "--blender", "", "--effort", "low", "--model", "claude-opus-5-5", "--trials", "1", "--task-timeout", "300", "--output", "/h2h-opus-oneshot-mini-camera/angelo-trial-04", "--task-set", "oneshot-mini-camera", "--long-tasks", "40", "--observation", "turn-context", "--context-rebase-tokens", "0", "--input-isolation", "on", "--after-failure", "skip", "--python-approvals", "approve", "--trusted-scripts", "--stable-state-dir", "/h2h-opus-oneshot-mini-camera/angelo-state" ], "load_average_start": [ 2.7548828125, 3.0283203125, 3 ], "started_at": "2026-10-15T08:37:19+0900", "exit_code": 0, "elapsed_s": 7.767, "load_average_end": [ 3.09326171875, 3.0927734375, 3.02294921875 ], "version": "5.2.2 LTS", "binary": "", "aborted": null, "success": true, "invalid": null, "scenario": { "python_approvals": "approve", "oneshot": { "limits": { "wall_s": 300, "job_wait_s": 0, "cost_usd": 3 }, "cost_cap_usd": 3 }, "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98", "saved": "/h2h-opus-oneshot-mini-camera/angelo-trial-04/trial-01/qa/final.blend", "undo": { "undo_steps": 1, "stopped": "reached the start", "landed": [ 0 ], "reaches_start": true, "first_undo_returns_previous_turn": true, "first_undo_lands_on": 0, "turn_boundaries_reachable": 1, "turns": 1, "steps_per_turn": { "1-1": 1 } } } }, { "side": "mcp", "trial": 5, "port": 39892, "success": true, "gate_version": null, "load_average_start": [ 2.99169921875, 3.072265625, 3.01611328125 ], "started_at": "2026-10-15T08:37:35+0900", "blender": { "ok": true, "port": 39892, "outcome": [ "FINISHED" ], "blender_version": "5.2.0 LTS", "build_hash": "fbe6228777e7", "online_access": true, "addon_module": "bl_ext.user_default.mcp", "addon_version": "1.0.0" }, "mcp_server": { "server_info": { "name": "blender-mcp", "version": "1.27.0" }, "initialize_ms": 252.353, "tools": [ "execute_blender_code", "execute_blender_code_for_cli", "get_blendfile_summary_datablocks", "get_blendfile_summary_datablocks_for_cli", "get_blendfile_summary_missing_files", "get_blendfile_summary_missing_files_for_cli", "get_blendfile_summary_of_linked_libraries", "get_blendfile_summary_of_linked_libraries_for_cli", "get_blendfile_summary_path_info", "get_blendfile_summary_path_info_for_cli", "get_blendfile_summary_usage_guess", "get_blendfile_summary_usage_guess_for_cli", "get_object_detail_summary", "get_objects_summary", "get_python_api_docs", "get_screenshot_of_area_as_image", "get_screenshot_of_window_as_image", "get_screenshot_of_window_as_json", "jump_to_tab_by_name", "jump_to_tab_by_space_type", "jump_to_view3d_object_by_name", "jump_to_view3d_object_data_by_name", "render_thumbnail_to_path", "render_viewport_to_path", "search_api_docs", "search_manual_docs" ] }, "claude_command": [ "", "-p", "--input-format", "stream-json", "--output-format", "stream-json", "--verbose", "--include-partial-messages", "--model", "claude-opus-5-5", "--effort", "low", "--settings", "{\"fastMode\": false}", "--mcp-config", "/h2h-opus-oneshot-mini-camera/mcp-trial-05/mcp-config.json", "--strict-mcp-config", "--allowedTools", "mcp__blender,mcp__blender__execute_blender_code,mcp__blender__execute_blender_code_for_cli,mcp__blender__get_blendfile_summary_datablocks,mcp__blender__get_blendfile_summary_datablocks_for_cli,mcp__blender__get_blendfile_summary_missing_files,mcp__blender__get_blendfile_summary_missing_files_for_cli,mcp__blender__get_blendfile_summary_of_linked_libraries,mcp__blender__get_blendfile_summary_of_linked_libraries_for_cli,mcp__blender__get_blendfile_summary_path_info,mcp__blender__get_blendfile_summary_path_info_for_cli,mcp__blender__get_blendfile_summary_usage_guess,mcp__blender__get_blendfile_summary_usage_guess_for_cli,mcp__blender__get_object_detail_summary,mcp__blender__get_objects_summary,mcp__blender__get_python_api_docs,mcp__blender__get_screenshot_of_area_as_image,mcp__blender__get_screenshot_of_window_as_image,mcp__blender__get_screenshot_of_window_as_json,mcp__blender__jump_to_tab_by_name,mcp__blender__jump_to_tab_by_space_type,mcp__blender__jump_to_view3d_object_by_name,mcp__blender__jump_to_view3d_object_data_by_name,mcp__blender__render_thumbnail_to_path,mcp__blender__render_viewport_to_path,mcp__blender__search_api_docs,mcp__blender__search_manual_docs,Read(//**/*.png),Read(//**/*.jpg),Read(//**/*.jpeg),Read(//**/*.exr),Read(//**/*.webp),Read(//**/*.tif),Read(//**/*.tiff),Read(//**/*.bmp)", "--permission-prompts", "none", "--no-session-persistence", "--add-dir", "/h2h-opus-oneshot-mini-camera/mcp-trial-05" ], "oneshot": { "messages": 1, "refused_sends": 0, "stop": null, "sent_wall": 1792021063.0223029, "cost_cap_usd": 3, "watch": { "usage": { "input_tokens": 110614, "cached_input_tokens": 101936, "cache_write_input_tokens": 8672, "output_tokens": 88, "requests": 3 }, "cost_usd": 0.0655 }, "allowances": [ "every tool of the Blender MCP server (model-written Python included, unrestricted)", "Claude Code's Read tool on image files anywhere, and its read-only file tools in its working folder and in the trial folder (stock Blender's working and temporary folder): it can open the images it rendered", "the reference images also as files in its working folder", "anything else that would prompt (shell, edits, web) is denied and counted: nobody answers prompts" ] }, "scenario": { "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb", "saved": "/h2h-opus-oneshot-mini-camera/mcp-trial-05/final.blend", "undo": { "undo_steps": 0, "stopped": "nothing left to undo", "landed": [], "reaches_start": false, "first_undo_returns_previous_turn": false, "first_undo_lands_on": null, "turn_boundaries_reachable": 0, "turns": 1, "steps_per_turn": {} } }, "load_average_end": [ 3.265625, 3.130859375, 3.03759765625 ] }, { "side": "angelo", "trial": 5, "harness_command": [ "/Applications/Blender.app/Contents/Resources/5.2/python/bin/python3.13", "", "--authorize-billed-runs", "--connection-id", "", "--source-connections-dir", "", "--blender", "", "--effort", "low", "--model", "claude-opus-5-5", "--trials", "1", "--task-timeout", "300", "--output", "/h2h-opus-oneshot-mini-camera/angelo-trial-05", "--task-set", "oneshot-mini-camera", "--long-tasks", "40", "--observation", "turn-context", "--context-rebase-tokens", "0", "--input-isolation", "on", "--after-failure", "skip", "--python-approvals", "approve", "--trusted-scripts", "--stable-state-dir", "/h2h-opus-oneshot-mini-camera/angelo-state" ], "load_average_start": [ 3.265625, 3.130859375, 3.03759765625 ], "started_at": "2026-10-15T08:37:58+0900", "exit_code": 0, "elapsed_s": 9.662, "load_average_end": [ 3.22412109375, 3.1259765625, 3.03662109375 ], "version": "5.2.2 LTS", "binary": "", "aborted": null, "success": true, "invalid": null, "scenario": { "python_approvals": "approve", "oneshot": { "limits": { "wall_s": 300, "job_wait_s": 0, "cost_usd": 3 }, "cost_cap_usd": 3 }, "digests": [ "cf5a1d3ec5fa84a6c89b54e67d8dc74e36e577e00d2ad485a71a5de4624ee578", "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98" ], "baseline_orphans": { "materials": [ "Dots Stroke" ] }, "hygiene": { "objects": 3, "unexpected_objects": [], "default_named": [], "leftover_defaults": [ "Cube", "Camera", "Light" ], "new_orphans": {}, "new_orphan_count": 0, "duplicate_materials": [], "empty_material_slots": [], "meshes_without_material": [], "negative_scale": [], "mode": "OBJECT", "collections": 1, "objects_in_scene_root": 1, "issues": 3 }, "final_digest": "e7fd5efb7a4d70c16ea55850475988facb0bbce03cc2677a91897b6b39379b98", "saved": "/h2h-opus-oneshot-mini-camera/angelo-trial-05/trial-01/qa/final.blend", "undo": { "undo_steps": 1, "stopped": "reached the start", "landed": [ 0 ], "reaches_start": true, "first_undo_returns_previous_turn": true, "first_undo_lands_on": 0, "turn_boundaries_reachable": 1, "turns": 1, "steps_per_turn": { "1-1": 1 } } } } ], "stopped_sides": [], "rate_limited": null, "load_averages": [ { "side": "mcp", "trial": 1, "start": [ 3.421875, 3.0810546875, 3.00634765625 ], "end": [ 3.48291015625, 3.12158203125, 3.021484375 ] }, { "side": "angelo", "trial": 1, "start": [ 3.48291015625, 3.12158203125, 3.021484375 ], "end": [ 3.44384765625, 3.119140625, 3.02099609375 ] }, { "side": "mcp", "trial": 2, "start": [ 3.61572265625, 3.16455078125, 3.03759765625 ], "end": [ 3.736328125, 3.21826171875, 3.05908203125 ] }, { "side": "angelo", "trial": 2, "start": [ 3.736328125, 3.21826171875, 3.05908203125 ], "end": [ 3.62939453125, 3.2109375, 3.05810546875 ] }, { "side": "mcp", "trial": 3, "start": [ 3.73876953125, 3.240234375, 3.0693359375 ], "end": [ 3.06982421875, 3.123046875, 3.03173828125 ] }, { "side": "angelo", "trial": 3, "start": [ 3.06982421875, 3.123046875, 3.03173828125 ], "end": [ 2.98486328125, 3.10205078125, 3.02490234375 ] }, { "side": "mcp", "trial": 4, "start": [ 2.8330078125, 3.0654296875, 3.01220703125 ], "end": [ 2.7548828125, 3.0283203125, 3 ] }, { "side": "angelo", "trial": 4, "start": [ 2.7548828125, 3.0283203125, 3 ], "end": [ 3.09326171875, 3.0927734375, 3.02294921875 ] }, { "side": "mcp", "trial": 5, "start": [ 2.99169921875, 3.072265625, 3.01611328125 ], "end": [ 3.265625, 3.130859375, 3.03759765625 ] }, { "side": "angelo", "trial": 5, "start": [ 3.265625, 3.130859375, 3.03759765625 ], "end": [ 3.22412109375, 3.1259765625, 3.03662109375 ] } ] }, "summary": { "tasks": { "brief": { "mcp": { "runs": 5, "skipped": 0, "passed": 5, "completed": 5, "pass_rate": 1, "graded": { "passed": 5, "failed": 0, "ungraded": 0, "skipped": 0, "total": 5 }, "answer_unchecked": false, "wall_ms": { "n": 5, "median": 13437.693, "p75": 14239.073, "min": 12699.815, "max": 15786.396 }, "wall_ms_passed_only": { "n": 5, "median": 13437.693, "p75": 14239.073, "min": 12699.815, "max": 15786.396 }, "first_text_ms": { "n": 5, "median": 13431.8, "p75": 14233.2, "min": 12693.9, "max": 15767.1 }, "first_feedback_ms": { "n": 5, "median": 2879.5, "p75": 3006.4, "min": 2399, "max": 5131.4 }, "input_tokens_per_call": { "n": 5, "median": 37089.5, "p75": 37093.5, "min": 37079.25, "max": 37099.75 }, "api_calls": { "n": 5, "median": 4, "p75": 4, "min": 4, "max": 4 }, "tool_calls": { "n": 5, "median": 3, "p75": 3, "min": 3, "max": 3 }, "input_tokens": { "n": 5, "median": 148358, "p75": 148374, "min": 148317, "max": 148399 }, "cached_input_tokens": { "n": 5, "median": 139105, "p75": 139105, "min": 122937, "max": 139109 }, "cache_write_input_tokens": { "n": 5, "median": 9257, "p75": 9286, "min": 9204, "max": 25374 }, "output_tokens": { "n": 5, "median": 851, "p75": 889, "min": 823, "max": 901 }, "api_ms": { "n": 5, "median": 11839, "p75": 11893, "min": 11395, "max": 12866 }, "cost_usd": { "n": 5, "median": 0.12, "p75": 0.12, "min": 0.118, "max": 0.244 } }, "angelo": { "runs": 5, "skipped": 0, "passed": 5, "completed": 5, "pass_rate": 1, "graded": { "passed": 5, "failed": 0, "ungraded": 0, "skipped": 0, "total": 5 }, "answer_unchecked": false, "wall_ms": { "n": 5, "median": 3979.395, "p75": 4922.697, "min": 3127.647, "max": 5683.351 }, "wall_ms_passed_only": { "n": 5, "median": 3979.395, "p75": 4922.697, "min": 3127.647, "max": 5683.351 }, "first_text_ms": { "n": 0 }, "first_feedback_ms": { "n": 0 }, "input_tokens_per_call": { "n": 5, "median": 36192, "p75": 36194, "min": 35865, "max": 36195 }, "api_calls": { "n": 5, "median": 1, "p75": 1, "min": 1, "max": 1 }, "tool_calls": { "n": 5, "median": 1, "p75": 1, "min": 1, "max": 1 }, "input_tokens": { "n": 5, "median": 36192, "p75": 36194, "min": 35865, "max": 36195 }, "cached_input_tokens": { "n": 5, "median": 34543, "p75": 34543, "min": 34543, "max": 34543 }, "cache_write_input_tokens": { "n": 5, "median": 1647, "p75": 1649, "min": 1320, "max": 1650 }, "output_tokens": { "n": 5, "median": 227, "p75": 227, "min": 226, "max": 227 }, "api_ms": { "n": 5, "median": 3005, "p75": 3709, "min": 2131, "max": 4676 }, "cost_usd": { "n": 5, "median": 0.025, "p75": 0.025, "min": 0.022, "max": 0.025 } } } }, "totals": { "mcp": { "trials": 5, "trials_with_all_five_timed": 5, "trials_all_five_passed": 5, "wall_ms": { "n": 5, "median": 13437.693, "p75": 14239.073, "min": 12699.815, "max": 15786.396 }, "api_calls": { "n": 5, "median": 4, "p75": 4, "min": 4, "max": 4 }, "cost_usd": { "n": 5, "median": 0.12, "p75": 0.12, "min": 0.118, "max": 0.244 }, "head_to_head_pass_rate": 1, "graded": { "passed": 5, "failed": 0, "ungraded": 0, "skipped": 0, "total": 5, "not_run": 0 }, "answers": { "read": 0, "parsed_leniently": 0 }, "answer_unchecked_tasks": [] }, "angelo": { "trials": 5, "trials_with_all_five_timed": 5, "trials_all_five_passed": 5, "wall_ms": { "n": 5, "median": 3979.395, "p75": 4922.697, "min": 3127.647, "max": 5683.351 }, "api_calls": { "n": 5, "median": 1, "p75": 1, "min": 1, "max": 1 }, "cost_usd": { "n": 5, "median": 0.025, "p75": 0.025, "min": 0.022, "max": 0.025 }, "head_to_head_pass_rate": 1, "graded": { "passed": 5, "failed": 0, "ungraded": 0, "skipped": 0, "total": 5, "not_run": 0 }, "answers": { "read": 0, "parsed_leniently": 0 }, "answer_unchecked_tasks": [] } }, "invalid_rate_limited": {} }, "observations": [ "Model API time as a share of head-to-head wall time: MCP 86%, Angelo 75%.", "Model API calls per trial (1 tasks): MCP [4, 4, 4, 4, 4]; Angelo [1, 1, 1, 1, 1].", "Wall s per trial (1 tasks): MCP [13.4, 12.7, 15.8, 14.2, 13.4]; Angelo [4.0, 3.1, 4.9, 3.5, 5.7].", "MCP add-on socket round trips per trial, s (includes its 0.25/1 s polling): [0.8, 0.48, 0.68, 0.63, 1.02].", "Angelo opening before the first prompt, per trial: ['none', 'none', 'none', 'none', 'none'].", "MCP cold start (process spawn to Claude Code init), ms: [733, 811, 3219, 748, 726].", "MCP first-message prompt-cache writes per trial (tokens): [25374, 9257, 9286, 9245, 9204].", "MCP permission denials 0, API retries 0, add-on script errors 0; Angelo declined tools 0, Python fallbacks 0." ], "tool_usage": { "mcp": { "brief": { "ToolSearch": 5, "mcp__blender__execute_blender_code": 10 } }, "angelo": { "brief": { "angelo_scene_program": 5 } } }, "session_curve": {}, "records": [ { "side": "mcp", "trial": 1, "task": "brief", "kind": "edit", "source": "scenario oneshot_mini_camera turn 1: A new scene camera aimed at the cube", "sent_at_s": 0, "runtime_start": "cold", "wall_ms": 13365.845, "init": { "model": "claude-opus-5-5", "claude_code_version": "2.1.293", "permissionMode": "default", "mcp_servers": [ { "name": "blender", "status": "connected", "source": "dynamic" } ], "received_ms": 732.9 }, "first_text_ms": 13344.5, "first_tool_call_ms": 2879.5, "api_calls": 4, "tool_calls": [ { "name": "ToolSearch", "issued_ms": 3275.3, "input_chars": 72, "returned_ms": 3279.2, "round_trip_ms": 3.9, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 5414.4, "input_chars": 160, "returned_ms": 5989.1, "round_trip_ms": 574.7, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 10264.4, "input_chars": 770, "returned_ms": 10488.7, "round_trip_ms": 224.3, "is_error": false } ], "tool_call_count": 3, "tool_names": [ "ToolSearch", "mcp__blender__execute_blender_code", "mcp__blender__execute_blender_code" ], "tool_errors": 0, "api_retries": [], "tool_round_trip_ms_total": 802.9, "result": { "subtype": "success", "is_error": false, "num_turns": 4, "duration_ms": 12654, "duration_api_ms": 11839, "total_cost_usd": 0.24407140000000002, "stop_reason": "end_turn", "models": [ "claude-opus-5-5" ], "usage": { "input_tokens": 8, "cache_creation_input_tokens": 25374, "cache_read_input_tokens": 122937, "output_tokens": 823, "output_tokens_details": { "thinking_tokens": 0 }, "server_tool_use": { "web_search_requests": 0, "web_fetch_requests": 0 }, "service_tier": "standard", "cache_creation": { "ephemeral_1h_input_tokens": 25374, "ephemeral_5m_input_tokens": 0 }, "inference_geo": "global", "iterations": [ { "input_tokens": 2, "output_tokens": 183, "cache_read_input_tokens": 37169, "cache_creation_input_tokens": 534, "cache_creation": { "ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 534 }, "type": "message" } ], "speed": "standard", "fallback_credit": null }, "permission_denials": [], "reply": "I added a camera at (6, -8, 4) pointed at the cube's centre (0, 0, 0) with a 50 mm lens, and made it the scene camera. I also deleted the old camera and its unused camera data. A check of the scene afterwards confirmed the new camera's position, lens, and scene-camera setting, and that it's the only camera left.\n\nI named the new camera and its data \"Camera\", the same name the old one used, so anything that looks for \"Camera\" will still find it. Its rotation is about (68.2°, 0°, 36.9°), with no roll." }, "provider_usage": { "input_tokens": 148319, "cached_input_tokens": 122937, "cache_write_input_tokens": 25374, "output_tokens": 823, "reasoning_output_tokens": 0, "api_ms": 11839, "cost_usd": 0.24407140000000002, "models": [ "claude-opus-5-5" ], "turn_ms": 12654 }, "host_check": { "success": true, "passed": 4, "total": 4, "results": [ { "index": 0, "kind": "object", "label": "the default cube is still there (the factory scene is kept)", "passed": true, "details": { "found": [ "Cube" ], "types": [ "MESH" ] } }, { "index": 1, "kind": "bbox", "label": "the cube is where and as large as it was", "passed": true, "details": { "min": [ -1, -1, -1 ], "max": [ 1, 1, 1 ], "size": [ 2, 2, 2 ], "off": [] } }, { "index": 2, "kind": "count", "label": "one camera remains: the old one is gone", "passed": true, "details": { "count": 1, "names": [ "Camera" ] } }, { "index": 3, "kind": "camera", "label": "the scene camera is at (6, -8, 4), aimed at the cube, 50 mm", "passed": true, "details": { "name": "Camera", "location": [ 6, -8, 4 ], "lens": 50, "active": true, "type": "PERSP", "distance": 0, "aim_dot": 1, "problems": [] } } ] }, "host_check_passed": true, "state_digest": "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb", "context_mode": "OBJECT", "objects": [ "Camera", "Cube", "Light" ], "status": "completed", "grading_contract": 4, "uncached_input_tokens": 25382, "list_price_usd": 0.1679, "list_price_basis": "model_profiles list prices, 5m cache writes" }, { "side": "angelo", "trial": 1, "task": "brief", "kind": "edit", "status": "completed", "runtime_start": "cold", "wall_ms": 3979.395, "send_to_idle_ms": 3980.902, "first_text_ms": null, "first_verified_result_ms": null, "model_rounds": 1, "provider_turn_requests": 1, "native_tool_invocations": 1, "python_fallbacks": 0, "api_calls": 1, "tool_names": [ "angelo_scene_program" ], "plans": [ [ "angelo_scene_program" ] ], "models": [ "claude-opus-5-5" ], "tool_call_count": 1, "approved": [], "declined": [], "grade": "passed", "gate_version": null, "answer": null, "approval": "AUTO_BOUNDED", "always_allow_scripts": false, "grading_contract": 4, "provider_usage": { "input_tokens": 36192, "cached_input_tokens": 34543, "cache_write_input_tokens": 1647, "output_tokens": 227, "reasoning_output_tokens": 0, "api_ms": 3005, "cost_usd": 0.024633, "turn_ms": 3029, "requests": 1, "turns": 1 }, "host_check_passed": true, "warm_ups": 0, "warm_up_usage": null, "opening": { "prepared": false, "runtime_live": false, "bases": [], "waited_s": 0.097, "wait_limit_s": 0 }, "environment": { "frame_before": 1, "input_isolated": true, "frame_after": 1, "playback": false, "external_input_suspected": false }, "harness_reset": null, "receipt_counts": null, "accept": null, "uncached_input_tokens": 1649, "list_price_usd": 0.0246, "list_price_basis": "model_profiles list prices, 1h cache writes" }, { "side": "mcp", "trial": 2, "task": "brief", "kind": "edit", "source": "scenario oneshot_mini_camera turn 1: A new scene camera aimed at the cube", "sent_at_s": 0, "runtime_start": "cold", "wall_ms": 12699.815, "init": { "model": "claude-opus-5-5", "claude_code_version": "2.1.293", "permissionMode": "default", "mcp_servers": [ { "name": "blender", "status": "connected", "source": "dynamic" } ], "received_ms": 811.2 }, "first_text_ms": 12693.9, "first_tool_call_ms": 2399, "api_calls": 4, "tool_calls": [ { "name": "ToolSearch", "issued_ms": 2780.8, "input_chars": 72, "returned_ms": 2784.7, "round_trip_ms": 3.9, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 4530, "input_chars": 164, "returned_ms": 4989.2, "round_trip_ms": 459.2, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 9464.6, "input_chars": 896, "returned_ms": 9489.2, "round_trip_ms": 24.6, "is_error": false } ], "tool_call_count": 3, "tool_names": [ "ToolSearch", "mcp__blender__execute_blender_code", "mcp__blender__execute_blender_code" ], "tool_errors": 0, "api_retries": [], "tool_round_trip_ms_total": 487.7, "result": { "subtype": "success", "is_error": false, "num_turns": 4, "duration_ms": 11910, "duration_api_ms": 11395, "total_cost_usd": 0.1199298, "stop_reason": "end_turn", "models": [ "claude-opus-5-5" ], "usage": { "input_tokens": 8, "cache_creation_input_tokens": 9257, "cache_read_input_tokens": 139109, "output_tokens": 901, "output_tokens_details": { "thinking_tokens": 22 }, "server_tool_use": { "web_search_requests": 0, "web_fetch_requests": 0 }, "service_tier": "standard", "cache_creation": { "ephemeral_1h_input_tokens": 9257, "ephemeral_5m_input_tokens": 0 }, "inference_geo": "global", "iterations": [ { "input_tokens": 2, "output_tokens": 207, "cache_read_input_tokens": 37173, "cache_creation_input_tokens": 581, "cache_creation": { "ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 581 }, "type": "message" } ], "speed": "standard", "fallback_credit": null }, "permission_denials": [], "reply": "I added the new camera, made it the scene camera, and deleted the old one.\n\n- **New camera:** at (6, -8, 4) with a 50 mm lens. It's aimed straight at the cube's centre at the origin; I checked the aim and it's exact.\n- **Scene camera:** set to the new camera.\n- **Old camera:** deleted, along with its camera data, which nothing else was using.\n\nI named the new camera \"Camera\", the same as the old one, so anything that looks for \"Camera\" by name still finds it. The scene now holds only Cube, Light and Camera." }, "provider_usage": { "input_tokens": 148374, "cached_input_tokens": 139109, "cache_write_input_tokens": 9257, "output_tokens": 901, "reasoning_output_tokens": 22, "api_ms": 11395, "cost_usd": 0.1199298, "models": [ "claude-opus-5-5" ], "turn_ms": 11910 }, "host_check": { "success": true, "passed": 4, "total": 4, "results": [ { "index": 0, "kind": "object", "label": "the default cube is still there (the factory scene is kept)", "passed": true, "details": { "found": [ "Cube" ], "types": [ "MESH" ] } }, { "index": 1, "kind": "bbox", "label": "the cube is where and as large as it was", "passed": true, "details": { "min": [ -1, -1, -1 ], "max": [ 1, 1, 1 ], "size": [ 2, 2, 2 ], "off": [] } }, { "index": 2, "kind": "count", "label": "one camera remains: the old one is gone", "passed": true, "details": { "count": 1, "names": [ "Camera" ] } }, { "index": 3, "kind": "camera", "label": "the scene camera is at (6, -8, 4), aimed at the cube, 50 mm", "passed": true, "details": { "name": "Camera", "location": [ 6, -8, 4 ], "lens": 50, "active": true, "type": "PERSP", "distance": 0, "aim_dot": 1, "problems": [] } } ] }, "host_check_passed": true, "state_digest": "17a8d4e2efb700562c05d4dca1d2adae7816dcc5570eac9f52c5397f0e294571", "context_mode": "OBJECT", "objects": [ "Camera", "Cube", "Light" ], "status": "completed", "grading_contract": 4, "uncached_input_tokens": 9265, "list_price_usd": 0.0922, "list_price_basis": "model_profiles list prices, 5m cache writes" }, { "side": "angelo", "trial": 2, "task": "brief", "kind": "edit", "status": "completed", "runtime_start": "cold", "wall_ms": 3127.647, "send_to_idle_ms": 3129.097, "first_text_ms": null, "first_verified_result_ms": null, "model_rounds": 1, "provider_turn_requests": 1, "native_tool_invocations": 1, "python_fallbacks": 0, "api_calls": 1, "tool_names": [ "angelo_scene_program" ], "plans": [ [ "angelo_scene_program" ] ], "models": [ "claude-opus-5-5" ], "tool_call_count": 1, "approved": [], "declined": [], "grade": "passed", "gate_version": null, "answer": null, "approval": "AUTO_BOUNDED", "always_allow_scripts": false, "grading_contract": 4, "provider_usage": { "input_tokens": 36194, "cached_input_tokens": 34543, "cache_write_input_tokens": 1649, "output_tokens": 226, "reasoning_output_tokens": 0, "api_ms": 2131, "cost_usd": 0.024629, "turn_ms": 2151, "requests": 1, "turns": 1 }, "host_check_passed": true, "warm_ups": 0, "warm_up_usage": null, "opening": { "prepared": false, "runtime_live": false, "bases": [], "waited_s": 0.097, "wait_limit_s": 0 }, "environment": { "frame_before": 1, "input_isolated": true, "frame_after": 1, "playback": false, "external_input_suspected": false }, "harness_reset": null, "receipt_counts": null, "accept": null, "uncached_input_tokens": 1651, "list_price_usd": 0.0246, "list_price_basis": "model_profiles list prices, 1h cache writes" }, { "side": "mcp", "trial": 3, "task": "brief", "kind": "edit", "source": "scenario oneshot_mini_camera turn 1: A new scene camera aimed at the cube", "sent_at_s": 0, "runtime_start": "cold", "wall_ms": 15786.396, "init": { "model": "claude-opus-5-5", "claude_code_version": "2.1.293", "permissionMode": "default", "mcp_servers": [ { "name": "blender", "status": "connected", "source": "dynamic" } ], "received_ms": 3219.1 }, "first_text_ms": 15767.1, "first_tool_call_ms": 5131.4, "api_calls": 4, "tool_calls": [ { "name": "ToolSearch", "issued_ms": 5488.3, "input_chars": 72, "returned_ms": 5492, "round_trip_ms": 3.7, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 7495.8, "input_chars": 160, "returned_ms": 7989, "round_trip_ms": 493.2, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 12803.5, "input_chars": 863, "returned_ms": 12988.6, "round_trip_ms": 185.1, "is_error": false } ], "tool_call_count": 3, "tool_names": [ "ToolSearch", "mcp__blender__execute_blender_code", "mcp__blender__execute_blender_code" ], "tool_errors": 0, "api_retries": [], "tool_round_trip_ms_total": 682, "result": { "subtype": "success", "is_error": false, "num_turns": 4, "duration_ms": 12579, "duration_api_ms": 11893, "total_cost_usd": 0.119921, "stop_reason": "end_turn", "models": [ "claude-opus-5-5" ], "usage": { "input_tokens": 8, "cache_creation_input_tokens": 9286, "cache_read_input_tokens": 139105, "output_tokens": 889, "output_tokens_details": { "thinking_tokens": 44 }, "server_tool_use": { "web_search_requests": 0, "web_fetch_requests": 0 }, "service_tier": "standard", "cache_creation": { "ephemeral_1h_input_tokens": 9286, "ephemeral_5m_input_tokens": 0 }, "inference_geo": "global", "iterations": [ { "input_tokens": 2, "output_tokens": 169, "cache_read_input_tokens": 37169, "cache_creation_input_tokens": 614, "cache_creation": { "ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 614 }, "type": "message" } ], "speed": "standard", "fallback_credit": null }, "permission_denials": [], "reply": "I added a new camera at (6, -8, 4) with a 50 mm lens, made it the scene camera, and deleted the old one.\n\n- **Aim:** it points straight at the cube's centre at the origin. I checked this after the move.\n- **Name:** the new camera now has the old one's name, \"Camera\". The old camera's data was removed too, since nothing else used it.\n- **Fixed rotation:** I set the angle directly rather than adding a \"Track To\" constraint. It won't follow the cube if you move the cube later." }, "provider_usage": { "input_tokens": 148399, "cached_input_tokens": 139105, "cache_write_input_tokens": 9286, "output_tokens": 889, "reasoning_output_tokens": 44, "api_ms": 11893, "cost_usd": 0.119921, "models": [ "claude-opus-5-5" ], "turn_ms": 12579 }, "host_check": { "success": true, "passed": 4, "total": 4, "results": [ { "index": 0, "kind": "object", "label": "the default cube is still there (the factory scene is kept)", "passed": true, "details": { "found": [ "Cube" ], "types": [ "MESH" ] } }, { "index": 1, "kind": "bbox", "label": "the cube is where and as large as it was", "passed": true, "details": { "min": [ -1, -1, -1 ], "max": [ 1, 1, 1 ], "size": [ 2, 2, 2 ], "off": [] } }, { "index": 2, "kind": "count", "label": "one camera remains: the old one is gone", "passed": true, "details": { "count": 1, "names": [ "Camera" ] } }, { "index": 3, "kind": "camera", "label": "the scene camera is at (6, -8, 4), aimed at the cube, 50 mm", "passed": true, "details": { "name": "Camera", "location": [ 6, -8, 4 ], "lens": 50, "active": true, "type": "PERSP", "distance": 0, "aim_dot": 1, "problems": [] } } ] }, "host_check_passed": true, "state_digest": "17a8d4e2efb700562c05d4dca1d2adae7816dcc5570eac9f52c5397f0e294571", "context_mode": "OBJECT", "objects": [ "Camera", "Cube", "Light" ], "status": "completed", "grading_contract": 4, "uncached_input_tokens": 9294, "list_price_usd": 0.0921, "list_price_basis": "model_profiles list prices, 5m cache writes" }, { "side": "angelo", "trial": 3, "task": "brief", "kind": "edit", "status": "completed", "runtime_start": "cold", "wall_ms": 4922.697, "send_to_idle_ms": 4924.21, "first_text_ms": null, "first_verified_result_ms": null, "model_rounds": 1, "provider_turn_requests": 1, "native_tool_invocations": 1, "python_fallbacks": 0, "api_calls": 1, "tool_names": [ "angelo_scene_program" ], "plans": [ [ "angelo_scene_program" ] ], "models": [ "claude-opus-5-5" ], "tool_call_count": 1, "approved": [], "declined": [], "grade": "passed", "gate_version": null, "answer": null, "approval": "AUTO_BOUNDED", "always_allow_scripts": false, "grading_contract": 4, "provider_usage": { "input_tokens": 36191, "cached_input_tokens": 34543, "cache_write_input_tokens": 1646, "output_tokens": 227, "reasoning_output_tokens": 0, "api_ms": 3709, "cost_usd": 0.024625, "turn_ms": 3728, "requests": 1, "turns": 1 }, "host_check_passed": true, "warm_ups": 0, "warm_up_usage": null, "opening": { "prepared": false, "runtime_live": false, "bases": [], "waited_s": 0.097, "wait_limit_s": 0 }, "environment": { "frame_before": 1, "input_isolated": true, "frame_after": 1, "playback": false, "external_input_suspected": false }, "harness_reset": null, "receipt_counts": null, "accept": null, "uncached_input_tokens": 1648, "list_price_usd": 0.0246, "list_price_basis": "model_profiles list prices, 1h cache writes" }, { "side": "mcp", "trial": 4, "task": "brief", "kind": "edit", "source": "scenario oneshot_mini_camera turn 1: A new scene camera aimed at the cube", "sent_at_s": 0, "runtime_start": "cold", "wall_ms": 14239.073, "init": { "model": "claude-opus-5-5", "claude_code_version": "2.1.293", "permissionMode": "default", "mcp_servers": [ { "name": "blender", "status": "connected", "source": "dynamic" } ], "received_ms": 748.5 }, "first_text_ms": 14233.2, "first_tool_call_ms": 3006.4, "api_calls": 4, "tool_calls": [ { "name": "ToolSearch", "issued_ms": 3403.7, "input_chars": 72, "returned_ms": 3407.4, "round_trip_ms": 3.7, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 5539.9, "input_chars": 160, "returned_ms": 5988.6, "round_trip_ms": 448.7, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 10805, "input_chars": 835, "returned_ms": 10988.9, "round_trip_ms": 183.9, "is_error": false } ], "tool_call_count": 3, "tool_names": [ "ToolSearch", "mcp__blender__execute_blender_code", "mcp__blender__execute_blender_code" ], "tool_errors": 0, "api_retries": [], "tool_round_trip_ms_total": 636.3, "result": { "subtype": "success", "is_error": false, "num_turns": 4, "duration_ms": 13514, "duration_api_ms": 12866, "total_cost_usd": 0.118833, "stop_reason": "end_turn", "models": [ "claude-opus-5-5" ], "usage": { "input_tokens": 8, "cache_creation_input_tokens": 9245, "cache_read_input_tokens": 139105, "output_tokens": 851, "output_tokens_details": { "thinking_tokens": 0 }, "server_tool_use": { "web_search_requests": 0, "web_fetch_requests": 0 }, "service_tier": "standard", "cache_creation": { "ephemeral_1h_input_tokens": 9245, "ephemeral_5m_input_tokens": 0 }, "inference_geo": "global", "iterations": [ { "input_tokens": 2, "output_tokens": 171, "cache_read_input_tokens": 37169, "cache_creation_input_tokens": 573, "cache_creation": { "ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 573 }, "type": "message" } ], "speed": "standard", "fallback_credit": null }, "permission_denials": [], "reply": "I added a new camera at (6, -8, 4) with a 50 mm lens, aimed it at the cube at the origin, and made it the scene camera. Then I deleted the old camera and its unused camera data.\n\n- **Name:** I renamed the new camera to \"Camera\" to match the old one's name, so it's still the only camera in the scene.\n- **Checks:** a readback confirmed the position, the lens and that it's the scene camera. It points straight at the cube's center and its up direction follows world Z, so the horizon stays level." }, "provider_usage": { "input_tokens": 148358, "cached_input_tokens": 139105, "cache_write_input_tokens": 9245, "output_tokens": 851, "reasoning_output_tokens": 0, "api_ms": 12866, "cost_usd": 0.118833, "models": [ "claude-opus-5-5" ], "turn_ms": 13514 }, "host_check": { "success": true, "passed": 4, "total": 4, "results": [ { "index": 0, "kind": "object", "label": "the default cube is still there (the factory scene is kept)", "passed": true, "details": { "found": [ "Cube" ], "types": [ "MESH" ] } }, { "index": 1, "kind": "bbox", "label": "the cube is where and as large as it was", "passed": true, "details": { "min": [ -1, -1, -1 ], "max": [ 1, 1, 1 ], "size": [ 2, 2, 2 ], "off": [] } }, { "index": 2, "kind": "count", "label": "one camera remains: the old one is gone", "passed": true, "details": { "count": 1, "names": [ "Camera" ] } }, { "index": 3, "kind": "camera", "label": "the scene camera is at (6, -8, 4), aimed at the cube, 50 mm", "passed": true, "details": { "name": "Camera", "location": [ 6, -8, 4 ], "lens": 50, "active": true, "type": "PERSP", "distance": 0, "aim_dot": 1, "problems": [] } } ] }, "host_check_passed": true, "state_digest": "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb", "context_mode": "OBJECT", "objects": [ "Camera", "Cube", "Light" ], "status": "completed", "grading_contract": 4, "uncached_input_tokens": 9253, "list_price_usd": 0.0911, "list_price_basis": "model_profiles list prices, 5m cache writes" }, { "side": "angelo", "trial": 4, "task": "brief", "kind": "edit", "status": "completed", "runtime_start": "cold", "wall_ms": 3543.914, "send_to_idle_ms": 3545.404, "first_text_ms": null, "first_verified_result_ms": null, "model_rounds": 1, "provider_turn_requests": 1, "native_tool_invocations": 1, "python_fallbacks": 0, "api_calls": 1, "tool_names": [ "angelo_scene_program" ], "plans": [ [ "angelo_scene_program" ] ], "models": [ "claude-opus-5-5" ], "tool_call_count": 1, "approved": [], "declined": [], "grade": "passed", "gate_version": null, "answer": null, "approval": "AUTO_BOUNDED", "always_allow_scripts": false, "grading_contract": 4, "provider_usage": { "input_tokens": 36195, "cached_input_tokens": 34543, "cache_write_input_tokens": 1650, "output_tokens": 227, "reasoning_output_tokens": 0, "api_ms": 2384, "cost_usd": 0.024657, "turn_ms": 2403, "requests": 1, "turns": 1 }, "host_check_passed": true, "warm_ups": 0, "warm_up_usage": null, "opening": { "prepared": false, "runtime_live": false, "bases": [], "waited_s": 0.097, "wait_limit_s": 0 }, "environment": { "frame_before": 1, "input_isolated": true, "frame_after": 1, "playback": false, "external_input_suspected": false }, "harness_reset": null, "receipt_counts": null, "accept": null, "uncached_input_tokens": 1652, "list_price_usd": 0.0247, "list_price_basis": "model_profiles list prices, 1h cache writes" }, { "side": "mcp", "trial": 5, "task": "brief", "kind": "edit", "source": "scenario oneshot_mini_camera turn 1: A new scene camera aimed at the cube", "sent_at_s": 0, "runtime_start": "cold", "wall_ms": 13437.693, "init": { "model": "claude-opus-5-5", "claude_code_version": "2.1.293", "permissionMode": "default", "mcp_servers": [ { "name": "blender", "status": "connected", "source": "dynamic" } ], "received_ms": 726.1 }, "first_text_ms": 13431.8, "first_tool_call_ms": 2789, "api_calls": 4, "tool_calls": [ { "name": "ToolSearch", "issued_ms": 3144.8, "input_chars": 72, "returned_ms": 3148.7, "round_trip_ms": 3.9, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 5143.3, "input_chars": 160, "returned_ms": 5989, "round_trip_ms": 845.7, "is_error": false }, { "name": "mcp__blender__execute_blender_code", "issued_ms": 10059.8, "input_chars": 775, "returned_ms": 10238.8, "round_trip_ms": 179, "is_error": false } ], "tool_call_count": 3, "tool_names": [ "ToolSearch", "mcp__blender__execute_blender_code", "mcp__blender__execute_blender_code" ], "tool_errors": 0, "api_retries": [], "tool_round_trip_ms_total": 1028.6, "result": { "subtype": "success", "is_error": false, "num_turns": 4, "duration_ms": 12723, "duration_api_ms": 11673, "total_cost_usd": 0.11848500000000001, "stop_reason": "end_turn", "models": [ "claude-opus-5-5" ], "usage": { "input_tokens": 8, "cache_creation_input_tokens": 9204, "cache_read_input_tokens": 139105, "output_tokens": 850, "output_tokens_details": { "thinking_tokens": 0 }, "server_tool_use": { "web_search_requests": 0, "web_fetch_requests": 0 }, "service_tier": "standard", "cache_creation": { "ephemeral_1h_input_tokens": 9204, "ephemeral_5m_input_tokens": 0 }, "inference_geo": "global", "iterations": [ { "input_tokens": 2, "output_tokens": 208, "cache_read_input_tokens": 37169, "cache_creation_input_tokens": 532, "cache_creation": { "ephemeral_5m_input_tokens": 0, "ephemeral_1h_input_tokens": 532 }, "type": "message" } ], "speed": "standard", "fallback_credit": null }, "permission_denials": [], "reply": "I added the new camera, made it the scene camera and deleted the old one.\n\n- **Position:** (6, -8, 4), with a 50 mm lens.\n- **Aim:** it points at the Cube's origin at (0, 0, 0). A check of its viewing direction confirmed it lines up with the cube.\n- **Old camera:** removed, along with its camera data. The new camera takes over the name \"Camera\", so the scene now holds just Camera, Cube and Light.\n\nThe camera's angle is set once to face the cube; it won't follow the cube if you move it. If you want it to keep tracking the cube, I can add a Track To constraint instead." }, "provider_usage": { "input_tokens": 148317, "cached_input_tokens": 139105, "cache_write_input_tokens": 9204, "output_tokens": 850, "reasoning_output_tokens": 0, "api_ms": 11673, "cost_usd": 0.11848500000000001, "models": [ "claude-opus-5-5" ], "turn_ms": 12723 }, "host_check": { "success": true, "passed": 4, "total": 4, "results": [ { "index": 0, "kind": "object", "label": "the default cube is still there (the factory scene is kept)", "passed": true, "details": { "found": [ "Cube" ], "types": [ "MESH" ] } }, { "index": 1, "kind": "bbox", "label": "the cube is where and as large as it was", "passed": true, "details": { "min": [ -1, -1, -1 ], "max": [ 1, 1, 1 ], "size": [ 2, 2, 2 ], "off": [] } }, { "index": 2, "kind": "count", "label": "one camera remains: the old one is gone", "passed": true, "details": { "count": 1, "names": [ "Camera" ] } }, { "index": 3, "kind": "camera", "label": "the scene camera is at (6, -8, 4), aimed at the cube, 50 mm", "passed": true, "details": { "name": "Camera", "location": [ 6, -8, 4 ], "lens": 50, "active": true, "type": "PERSP", "distance": 0, "aim_dot": 1, "problems": [] } } ] }, "host_check_passed": true, "state_digest": "74c57894dfddf0f0b2432bb64d9c50e37fa8ed56472ea64f75d8105b1e517afb", "context_mode": "OBJECT", "objects": [ "Camera", "Cube", "Light" ], "status": "completed", "grading_contract": 4, "uncached_input_tokens": 9212, "list_price_usd": 0.0909, "list_price_basis": "model_profiles list prices, 5m cache writes" }, { "side": "angelo", "trial": 5, "task": "brief", "kind": "edit", "status": "completed", "runtime_start": "cold", "wall_ms": 5683.351, "send_to_idle_ms": 5684.894, "first_text_ms": null, "first_verified_result_ms": null, "model_rounds": 1, "provider_turn_requests": 1, "native_tool_invocations": 1, "python_fallbacks": 0, "api_calls": 1, "tool_names": [ "angelo_scene_program" ], "plans": [ [ "angelo_scene_program" ] ], "models": [ "claude-opus-5-5" ], "tool_call_count": 1, "approved": [], "declined": [], "grade": "passed", "gate_version": null, "answer": null, "approval": "AUTO_BOUNDED", "always_allow_scripts": false, "grading_contract": 4, "provider_usage": { "input_tokens": 35865, "cached_input_tokens": 34543, "cache_write_input_tokens": 1320, "output_tokens": 227, "reasoning_output_tokens": 0, "api_ms": 4676, "cost_usd": 0.022017, "turn_ms": 4697, "requests": 1, "turns": 1 }, "host_check_passed": true, "warm_ups": 0, "warm_up_usage": null, "opening": { "prepared": false, "runtime_live": false, "bases": [], "waited_s": 0.097, "wait_limit_s": 0 }, "environment": { "frame_before": 1, "input_isolated": true, "frame_after": 1, "playback": false, "external_input_suspected": false }, "harness_reset": null, "receipt_counts": null, "accept": null, "uncached_input_tokens": 1322, "list_price_usd": 0.022, "list_price_basis": "model_profiles list prices, 1h cache writes" } ] }