Add local machine context snapshots

.gitignore

@@ -30,6 +30,7 @@ build/
 
 # MCP files
 openspace/config/config_mcp.json
+.mcp.json
 
 # Logs
 logs/
@@ -46,6 +47,7 @@ showcase/.openspace/*
 # GDPVal benchmark cache
 gdpval_bench/.openspace/*
 !gdpval_bench/.openspace/*.db
+gdpval_bench/results/
 
 # Embedding cache
 embedding_cache/
@@ -73,3 +75,6 @@ openspace/skills/*
 node_modules/
 # Frontend local dependency link
 frontend/node_modules
+
+# Local scratch
+tmp/

context/local-machine/admin-macos/README.md

@@ -0,0 +1,28 @@
+## Admin macOS Local Context
+
+This directory keeps machine-specific snapshots that are useful for future
+deployment, migration, and debugging on other machines, while avoiding noise in
+the repo root.
+
+Included here:
+
+- `mcp/repo-local.mcp.json`
+  - snapshot of the repo-local MCP wiring that was used during local debugging
+- `gdpval_bench/...`
+  - selected benchmark result snapshots that were useful during local call-rate
+    and provider-path investigation
+
+Intentional choices:
+
+- absolute local paths are preserved because they are part of the context
+- localhost API base values are preserved because they document the local stack
+- secrets are not preserved
+  - any benchmark config copied here has API keys redacted
+
+Intentionally omitted from this snapshot:
+
+- SQLite/WAL benchmark databases
+- raw recording directories
+- ad hoc `tmp/` scratch files
+
+Those source locations remain local-only and are ignored via `.gitignore`.

context/local-machine/admin-macos/gdpval_bench/codex_callrate_smoke/config.json

@@ -0,0 +1,19 @@
+{
+  "clawwork_root": "/tmp/openspace-bench-mp4o3U",
+  "gdpval_path": null,
+  "model": "gpt-5.4",
+  "max_iterations": 20,
+  "backend_scope": [
+    "shell"
+  ],
+  "use_clawwork_productivity": false,
+  "run_name": "codex_callrate_smoke",
+  "max_tasks": 3,
+  "per_occupation": null,
+  "sectors": null,
+  "occupations": null,
+  "task_ids": null,
+  "record_call_details": true,
+  "enable_evaluation": false,
+  "concurrency": 1
+}

context/local-machine/admin-macos/gdpval_bench/codex_callrate_smoke_noref/config.json

@@ -0,0 +1,19 @@
+{
+  "clawwork_root": "/tmp/openspace-bench-noref-JwyV4l",
+  "gdpval_path": null,
+  "model": "gpt-5.4",
+  "max_iterations": 20,
+  "backend_scope": [
+    "shell"
+  ],
+  "use_clawwork_productivity": false,
+  "run_name": "codex_callrate_smoke_noref",
+  "max_tasks": 3,
+  "per_occupation": null,
+  "sectors": null,
+  "occupations": null,
+  "task_ids": null,
+  "record_call_details": true,
+  "enable_evaluation": false,
+  "concurrency": 1
+}

context/local-machine/admin-macos/gdpval_bench/codex_callrate_smoke_noref/phase1_results.jsonl

@@ -0,0 +1,6 @@
+{"task_id": "0112fc9b-c3b2-4084-8993-5a4abb1f54f1", "phase": "phase1", "occupation": "Nurse Practitioners", "sector": "Health Care and Social Assistance", "task_value_usd": 0.0, "status": "error", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 0.01, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 0, "tool_calls": 0, "time_sec": 0.01}, "skills": {"before": 2, "after": 2, "new_this_task": 0, "evolved": [], "used": []}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:37:44.273256"}
+{"task_id": "02314fc6-a24e-42f4-a8cd-362cae0f0ec1", "phase": "phase1", "occupation": "General and Operations Managers", "sector": "Retail Trade", "task_value_usd": 0.0, "status": "error", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 0.0, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 0, "tool_calls": 0, "time_sec": 0.0}, "skills": {"before": 2, "after": 2, "new_this_task": 0, "evolved": [], "used": []}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:37:44.278249"}
+{"task_id": "02aa1805-c658-4069-8a6a-02dec146063a", "phase": "phase1", "occupation": "Project Management Specialists", "sector": "Professional, Scientific, and Technical Services", "task_value_usd": 0.0, "status": "error", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 0.0, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 0, "tool_calls": 0, "time_sec": 0.0}, "skills": {"before": 2, "after": 2, "new_this_task": 0, "evolved": [], "used": []}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:37:44.282846"}
+{"task_id": "0112fc9b-c3b2-4084-8993-5a4abb1f54f1", "phase": "phase1", "occupation": "Nurse Practitioners", "sector": "Health Care and Social Assistance", "task_value_usd": 0.0, "status": "error", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 0.0, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 0, "tool_calls": 0, "time_sec": 0.0}, "skills": {"before": 2, "after": 2, "new_this_task": 0, "evolved": [], "used": []}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:38:19.523517"}
+{"task_id": "02314fc6-a24e-42f4-a8cd-362cae0f0ec1", "phase": "phase1", "occupation": "General and Operations Managers", "sector": "Retail Trade", "task_value_usd": 0.0, "status": "error", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 0.0, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 0, "tool_calls": 0, "time_sec": 0.0}, "skills": {"before": 2, "after": 2, "new_this_task": 0, "evolved": [], "used": []}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:38:19.526964"}
+{"task_id": "02aa1805-c658-4069-8a6a-02dec146063a", "phase": "phase1", "occupation": "Project Management Specialists", "sector": "Professional, Scientific, and Technical Services", "task_value_usd": 0.0, "status": "error", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 0.0, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 0, "tool_calls": 0, "time_sec": 0.0}, "skills": {"before": 2, "after": 2, "new_this_task": 0, "evolved": [], "used": []}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:38:19.529894"}

context/local-machine/admin-macos/gdpval_bench/codex_callrate_smoke_noref_fix/config.json

@@ -0,0 +1,23 @@
+{
+  "clawwork_root": "/tmp/openspace-bench-noref-JwyV4l",
+  "gdpval_path": null,
+  "model": "gpt-5.4",
+  "max_iterations": 20,
+  "backend_scope": [
+    "shell"
+  ],
+  "use_clawwork_productivity": false,
+  "run_name": "codex_callrate_smoke_noref_fix",
+  "max_tasks": 2,
+  "per_occupation": null,
+  "sectors": null,
+  "occupations": null,
+  "task_ids": null,
+  "record_call_details": true,
+  "llm_kwargs": {
+    "api_key": "REDACTED",
+    "api_base": "http://127.0.0.1:8080/v1"
+  },
+  "enable_evaluation": false,
+  "concurrency": 1
+}

context/local-machine/admin-macos/gdpval_bench/codex_callrate_smoke_noref_fix/phase1_results.jsonl

@@ -0,0 +1,2 @@
+{"task_id": "0112fc9b-c3b2-4084-8993-5a4abb1f54f1", "phase": "phase1", "occupation": "Nurse Practitioners", "sector": "Health Care and Social Assistance", "task_value_usd": 0.0, "status": "success", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 93.54, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 1, "tool_calls": 0, "time_sec": 93.54}, "skills": {"before": 3, "after": 4, "new_this_task": 1, "evolved": [{"name": "clinical-soap-note-from-narrative-enhanced", "origin": "derived", "path": "/Users/admin/PycharmProjects/openspace/openspace/skills/clinical-soap-note-from-narrative-enhanced/SKILL.md"}], "used": ["clinical-soap-note-from-narrative__v0_9ff620af"]}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:50:48.493407"}
+{"task_id": "02314fc6-a24e-42f4-a8cd-362cae0f0ec1", "phase": "phase1", "occupation": "General and Operations Managers", "sector": "Retail Trade", "task_value_usd": 0.0, "status": "success", "tokens": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0, "llm_calls": 0, "cost_usd": 0.0, "wall_time_sec": 269.94, "agent_prompt_tokens": 0, "agent_completion_tokens": 0, "agent_total_tokens": 0, "agent_llm_calls": 0, "call_details": []}, "execution": {"iterations": 10, "tool_calls": 10, "time_sec": 269.94}, "skills": {"before": 4, "after": 6, "new_this_task": 2, "evolved": [{"name": "adaptive-document-output", "origin": "captured", "path": "/Users/admin/PycharmProjects/openspace/openspace/skills/adaptive-document-output/SKILL.md"}, {"name": "verify-build-artifacts", "origin": "captured", "path": "/Users/admin/PycharmProjects/openspace/openspace/skills/verify-build-artifacts/SKILL.md"}], "used": []}, "evaluation": {"has_evaluation": false}, "timestamp": "2026-04-12T02:55:18.445041"}

context/local-machine/admin-macos/mcp/repo-local.mcp.json

@@ -0,0 +1,23 @@
+{
+  "mcpServers": {
+    "openspace_evolution": {
+      "type": "stdio",
+      "command": "/Users/admin/PycharmProjects/openspace/.venv/bin/python",
+      "args": [
+        "-m",
+        "openspace.evolution_mcp_server",
+        "--transport",
+        "stdio"
+      ],
+      "env": {
+        "OPENSPACE_WORKSPACE": "/Users/admin/PycharmProjects/openspace",
+        "OPENSPACE_HOST_SKILL_DIRS": "/Users/admin/.claude/projects/openspace/skills,/Users/admin/.claude/skills",
+        "OPENSPACE_MODEL": "gpt-5.4",
+        "OPENSPACE_LLM_API_BASE": "http://127.0.0.1:8080/v1",
+        "OPENSPACE_LLM_OPENAI_STREAM_COMPAT": "true",
+        "OPENSPACE_ENABLE_RECORDING": "false",
+        "OPENSPACE_BACKEND_SCOPE": "shell,system"
+      }
+    }
+  }
+}

gdpval_bench/run_benchmark.py

@@ -388,6 +388,7 @@ def _make_config(cfg: Dict, phase: str, worker_id: int = 0):
 
     return OpenSpaceConfig(
         llm_model=cfg["model"],
+        llm_kwargs=cfg.get("llm_kwargs", {}),
         workspace_dir=str(rd / "workspace" / phase),
         recording_log_dir=rec_dir,
         recording_backends=cfg.get("backend_scope", ["shell", "web"]),
@@ -1164,6 +1165,16 @@ async def main(args: argparse.Namespace) -> None:
     if getattr(args, "use_clawwork_productivity", False):
         cfg["use_clawwork_productivity"] = True
 
+    # Match CLI / MCP entrypoints: resolve provider credentials once and pass
+    # them into OpenSpaceConfig.llm_kwargs so benchmark runs work with custom
+    # OPENSPACE_LLM_* endpoints too.
+    from openspace.host_detection import build_llm_kwargs, load_runtime_env
+
+    load_runtime_env()
+    resolved_model, llm_kwargs = build_llm_kwargs(cfg.get("model", ""))
+    cfg["model"] = resolved_model
+    cfg["llm_kwargs"] = llm_kwargs
+
     # When using ClawWork productivity tools, ensure livebench is importable
     # before OpenSpace.initialize() (ShellSession loads productivity_tools which imports livebench)
     if cfg.get("use_clawwork_productivity"):

openspace/skill_engine/evolver.py

@@ -43,6 +43,7 @@ from .patch import (
     create_skill,
     fix_skill,
     derive_skill,
+    parse_multi_file_full,
     SKILL_FILENAME,
 )
 from .skill_utils import (
@@ -71,6 +72,7 @@ EVOLUTION_FAILED = SkillEnginePrompts.EVOLUTION_FAILED
 
 _SKILL_CONTENT_MAX_CHARS = 12_000   # Max chars of SKILL.md in evolution prompt
 _MAX_SKILL_NAME_LENGTH = 50         # Max chars for a skill name (directory name)
+_MULTI_FILE_FULL_HEADER_RE = re.compile(r"(?m)^\*\*\*\s*File:\s*")
 
 
 def _sanitize_skill_name(name: str) -> str:
@@ -95,6 +97,88 @@ def _sanitize_skill_name(name: str) -> str:
         truncated = truncated[:last_hyphen]
     return truncated.strip("-")
 
+
+def _extract_generated_frontmatter_field(content: str, field_name: str) -> Optional[str]:
+    """Extract frontmatter fields from single-file or multi-file FULL outputs."""
+    direct = _extract_frontmatter_field(content, field_name)
+    if direct:
+        return direct
+
+    if _MULTI_FILE_FULL_HEADER_RE.search(content):
+        try:
+            skill_md = parse_multi_file_full(content).get(SKILL_FILENAME, "")
+        except Exception:
+            skill_md = ""
+        if skill_md:
+            nested = _extract_frontmatter_field(skill_md, field_name)
+            if nested:
+                return nested
+
+    # Fall back to the top of the generated text to tolerate minor format drift.
+    head = "\n".join(content.splitlines()[:40])
+    match = re.search(
+        rf"(?im)^\s*{re.escape(field_name)}\s*:\s*(.+?)\s*$",
+        head,
+    )
+    if not match:
+        return None
+
+    value = match.group(1).strip().strip("\"'")
+    return value or None
+
+
+def _serialize_multi_file_full(files: Dict[str, str]) -> str:
+    """Serialize parsed FULL files back into the ``*** Begin Files`` format."""
+    ordered_paths = [SKILL_FILENAME]
+    ordered_paths.extend(path for path in sorted(files) if path != SKILL_FILENAME)
+
+    lines = ["*** Begin Files"]
+    for path in ordered_paths:
+        if path not in files:
+            continue
+        lines.append(f"*** File: {path}")
+        body = files[path].rstrip("\n")
+        if body:
+            lines.append(body)
+    lines.append("*** End Files")
+    return "\n".join(lines).strip() + "\n"
+
+
+def _set_generated_frontmatter_field(content: str, field_name: str, value: str) -> str:
+    """Inject or update frontmatter fields in single-file or multi-file FULL outputs."""
+    if _MULTI_FILE_FULL_HEADER_RE.search(content):
+        files = parse_multi_file_full(content)
+        files[SKILL_FILENAME] = _set_frontmatter_field(
+            files.get(SKILL_FILENAME, ""),
+            field_name,
+            value,
+        )
+        return _serialize_multi_file_full(files)
+
+    return _set_frontmatter_field(content, field_name, value)
+
+
+def _fallback_captured_skill_name(
+    direction: str,
+    category: Optional[SkillCategory],
+) -> str:
+    """Generate a deterministic fallback skill name from the capture direction."""
+    hint = (direction or "").strip()
+    hint = re.split(r"[:.;\n]", hint, maxsplit=1)[0].strip()
+    hint = re.sub(r"(?i)^capture\s+", "", hint)
+    hint = re.sub(r"(?i)^a\s+reusable\s+", "", hint)
+    hint = re.sub(r"(?i)^reusable\s+", "", hint)
+    hint = re.sub(r"(?i)^(workflow|reference|tool[- ]guide)\s+for\s+", "", hint)
+    hint = re.sub(r"(?i)^(adding|creating|capturing|building)\s+", "", hint)
+    hint = re.sub(r"(?i)^(a|an|the)\s+", "", hint)
+
+    fallback = _sanitize_skill_name(hint)
+    if fallback:
+        return fallback
+
+    prefix = (category.value if category else "captured-skill").replace("_", "-")
+    return f"{prefix}-{uuid.uuid4().hex[:6]}"
+
 _ANALYSIS_CONTEXT_MAX = 5           # Max recent analyses to include in prompt
 _ANALYSIS_NOTE_MAX_CHARS = 500      # Per-analysis note truncation
 
@@ -937,15 +1021,26 @@ class SkillEvolver:
         new_content, change_summary = _extract_change_summary(new_content)
 
         # Extract name/description from the generated content
-        new_name = _extract_frontmatter_field(new_content, "name")
-        new_desc = _extract_frontmatter_field(new_content, "description")
+        new_name = _extract_generated_frontmatter_field(new_content, "name")
+        new_desc = _extract_generated_frontmatter_field(new_content, "description")
         if not new_name:
-            logger.warning("CAPTURED: LLM did not produce a valid skill name")
-            return None
+            new_name = _fallback_captured_skill_name(
+                ctx.suggestion.direction,
+                ctx.suggestion.category,
+            )
+            logger.warning(
+                "CAPTURED: LLM did not produce a valid skill name; using fallback '%s'",
+                new_name,
+            )
 
         # Sanitize name (enforce length limit + valid chars)
         new_name = _sanitize_skill_name(new_name)
-        new_content = _set_frontmatter_field(new_content, "name", new_name)
+        if not new_name:
+            new_name = _fallback_captured_skill_name(
+                ctx.suggestion.direction,
+                ctx.suggestion.category,
+            )
+        new_content = _set_generated_frontmatter_field(new_content, "name", new_name)
 
         # Create new skill directory via create_skill (handles multi-file FULL)
         skill_dirs = self._registry._skill_dirs
@@ -1528,4 +1623,4 @@ class SkillEvolver:
                 f"better error handling or alternative approaches."
             )
 
-        return None, ""
+        return None, ""

openspace/tool_layer.py

@@ -357,6 +357,7 @@ class OpenSpace:
         self._task_done.clear()
         self._last_evolved_skills = []  # Reset per-execution tracking
         start_time = asyncio.get_event_loop().time()
+        execution_time = 0.0
         # Use external task_id if provided, otherwise generate one
         if task_id is None:
             task_id = f"task_{uuid.uuid4().hex[:12]}"
@@ -555,6 +556,9 @@ class OpenSpace:
             }
         
         finally:
+            if execution_time <= 0:
+                execution_time = asyncio.get_event_loop().time() - start_time
+
             recording_dir = None
             if self._recording_manager and self._recording_manager.recording_status:
                 recording_dir = self._recording_manager.trajectory_dir

tests/test_evolver_captured_names.py

@@ -0,0 +1,79 @@
+from openspace.skill_engine.evolver import (
+    _extract_generated_frontmatter_field,
+    _fallback_captured_skill_name,
+    _set_generated_frontmatter_field,
+)
+
+
+def test_extract_generated_frontmatter_field_from_single_file() -> None:
+    content = """---
+name: safe-file-write
+description: Write files with validation.
+---
+
+# Safe File Write
+"""
+
+    assert _extract_generated_frontmatter_field(content, "name") == "safe-file-write"
+    assert (
+        _extract_generated_frontmatter_field(content, "description")
+        == "Write files with validation."
+    )
+
+
+def test_extract_generated_frontmatter_field_from_multi_file_full() -> None:
+    content = """*** Begin Files
+*** File: SKILL.md
+---
+name: local-acceptance-entry
+description: Run a repo's backend and frontend together for review.
+---
+
+# Local Acceptance Entry
+*** File: examples/start.sh
+#!/usr/bin/env bash
+echo start
+*** End Files
+"""
+
+    assert (
+        _extract_generated_frontmatter_field(content, "name")
+        == "local-acceptance-entry"
+    )
+    assert (
+        _extract_generated_frontmatter_field(content, "description")
+        == "Run a repo's backend and frontend together for review."
+    )
+
+
+def test_set_generated_frontmatter_field_updates_skill_md_in_multi_file_full() -> None:
+    content = """*** Begin Files
+*** File: SKILL.md
+---
+description: Run a repo's backend and frontend together for review.
+---
+
+# Local Acceptance Entry
+*** File: examples/start.sh
+#!/usr/bin/env bash
+echo start
+*** End Files
+"""
+
+    updated = _set_generated_frontmatter_field(content, "name", "local-acceptance-entry")
+
+    assert "*** File: SKILL.md" in updated
+    assert "name: local-acceptance-entry" in updated
+    assert "*** File: examples/start.sh" in updated
+
+
+def test_fallback_captured_skill_name_uses_direction_signal() -> None:
+    direction = (
+        "Capture a reusable workflow for adding a local acceptance entry in "
+        "split backend/frontend repos: create one canonical foreground acceptance script."
+    )
+
+    fallback = _fallback_captured_skill_name(direction, None)
+
+    assert fallback.startswith("local-acceptance-entry-in-split-backend")
+    assert len(fallback) <= 50