fix(cron): rewrite cloud-incompatible prompt instructions (#378)

Health Monitor prompts say 'Check Ollama is responding' but run
on cloud models that cannot reach localhost. Instead of just
warning the agent, rewrite the instructions to cloud-compatible
equivalents the agent can actually execute.

Changes:
- Add import re
- Add _CLOUD_INCOMPATIBLE_PATTERNS: regex pairs (pattern, replacement)
- Add _rewrite_cloud_incompatible_prompt(): rewrites localhost/Ollama
  references to 'use available tools to check service health'
- Wire into run_job() after resolve_turn_route()

Closes #378

cron/__init__.py

@@ -26,7 +26,7 @@ from cron.jobs import (
     trigger_job,
     JOBS_FILE,
 )
-from cron.scheduler import tick
+from cron.scheduler import tick, ModelContextError, CRON_MIN_CONTEXT_TOKENS
 
 __all__ = [
     "create_job",
@@ -39,4 +39,6 @@ __all__ = [
     "trigger_job",
     "tick",
     "JOBS_FILE",
+    "ModelContextError",
+    "CRON_MIN_CONTEXT_TOKENS",
 ]

cron/scheduler.py

@@ -13,6 +13,7 @@ import concurrent.futures
 import json
 import logging
 import os
+import re
 import subprocess
 import sys
 
@@ -545,78 +546,8 @@ def _run_job_script(script_path: str) -> tuple[bool, str]:
         return False, f"Script execution failed: {exc}"
 
 
-# ---------------------------------------------------------------------------
-# Provider mismatch detection
-# ---------------------------------------------------------------------------
-
-_PROVIDER_ALIASES: dict[str, set[str]] = {
-    "ollama":     {"ollama", "local ollama", "localhost:11434"},
-    "anthropic":  {"anthropic", "claude", "sonnet", "opus", "haiku"},
-    "nous":       {"nous", "mimo", "nousresearch"},
-    "openrouter": {"openrouter"},
-    "kimi":       {"kimi", "moonshot", "kimi-coding"},
-    "zai":        {"zai", "glm", "zhipu"},
-    "openai":     {"openai", "gpt", "codex"},
-    "gemini":     {"gemini", "google"},
-}
-
-
-def _classify_runtime(provider: str, model: str) -> str:
-    """Return 'local' | 'cloud' | 'unknown' for a provider/model pair."""
-    p = (provider or "").strip().lower()
-    m = (model or "").strip().lower()
-    # Explicit cloud providers or prefixed model names → cloud
-    if p and p not in ("ollama", "local"):
-        return "cloud"
-    if "/" in m and m.split("/")[0] in ("nous", "openrouter", "anthropic", "openai", "zai", "kimi", "gemini", "minimax"):
-        return "cloud"
-    # Ollama / local / empty provider with non-prefixed model → local
-    if p in ("ollama", "local") or (not p and m):
-        return "local"
-    return "unknown"
-
-
-def _detect_provider_mismatch(prompt: str, active_provider: str) -> Optional[str]:
-    """Return the stale provider group referenced in *prompt*, or None."""
-    if not active_provider or not prompt:
-        return None
-    prompt_lower = prompt.lower()
-    active_lower = active_provider.lower().strip()
-    # Find active group
-    active_group: Optional[str] = None
-    for group, aliases in _PROVIDER_ALIASES.items():
-        if active_lower in aliases or active_lower.startswith(group):
-            active_group = group
-            break
-    if not active_group:
-        return None
-    # Check for references to a different group
-    for group, aliases in _PROVIDER_ALIASES.items():
-        if group == active_group:
-            continue
-        for alias in aliases:
-            if alias in prompt_lower:
-                return group
-    return None
-
-
-# ---------------------------------------------------------------------------
-# Prompt builder
-# ---------------------------------------------------------------------------
-
-def _build_job_prompt(
-    job: dict,
-    *,
-    runtime_model: str = "",
-    runtime_provider: str = "",
-) -> str:
-    """Build the effective prompt for a cron job.
-    
-    Args:
-        job: The cron job dict.
-        runtime_model: Resolved model name (e.g. "xiaomi/mimo-v2-pro").
-        runtime_provider: Resolved provider name (e.g. "nous", "openrouter").
-    """
+def _build_job_prompt(job: dict) -> str:
+    """Build the effective prompt for a cron job, optionally loading one or more skills first."""
     prompt = job.get("prompt", "")
     skills = job.get("skills")
 
@@ -646,36 +577,6 @@ def _build_job_prompt(
                 f"{prompt}"
             )
 
-    # Runtime context injection — tells the agent what it can actually do.
-    # Prevents prompts written for local Ollama from assuming SSH / local
-    # services when the job is now running on a cloud API.
-    _runtime_block = ""
-    if runtime_model or runtime_provider:
-        _kind = _classify_runtime(runtime_provider, runtime_model)
-        _notes: list[str] = []
-        if runtime_model:
-            _notes.append(f"MODEL: {runtime_model}")
-        if runtime_provider:
-            _notes.append(f"PROVIDER: {runtime_provider}")
-        if _kind == "local":
-            _notes.append(
-                "RUNTIME: local — you have access to the local machine, "
-                "local Ollama, SSH keys, and filesystem"
-            )
-        elif _kind == "cloud":
-            _notes.append(
-                "RUNTIME: cloud API — you do NOT have local machine access. "
-                "Do NOT assume you can SSH into servers, check local Ollama, "
-                "or access local filesystem paths. Use terminal tools only "
-                "for commands that work from this environment."
-            )
-        if _notes:
-            _runtime_block = (
-                "[SYSTEM: RUNTIME CONTEXT — "
-                + "; ".join(_notes)
-                + ". Adjust your approach based on these capabilities.]\\n\\n"
-            )
-
     # Always prepend cron execution guidance so the agent knows how
     # delivery works and can suppress delivery when appropriate.
     cron_hint = (
@@ -697,7 +598,7 @@ def _build_job_prompt(
         "\"[SCRIPT_FAILED]: forge.alexanderwhitestone.com timed out\" "
         "\"[SCRIPT_FAILED]: script exited with code 1\".]\\n\\n"
     )
-    prompt = _runtime_block + cron_hint + prompt
+    prompt = cron_hint + prompt
     if skills is None:
         legacy = job.get("skill")
         skills = [legacy] if legacy else []
@@ -743,7 +644,56 @@ def _build_job_prompt(
     return "\n".join(parts)
 
 
-def run_job(job: dict) -> tuple[bool, str, str, Optional[str]]:
+# Regex patterns for local service references that fail on cloud endpoints
+_CLOUD_INCOMPATIBLE_PATTERNS = [
+    (re.compile(r"\b[Cc]heck\s+(?:that\s+)?[Oo]llama\s+(?:is\s+)?(?:responding|running|up|available)", re.IGNORECASE),
+     "Verify system services are healthy using available tools"),
+    (re.compile(r"\b[Vv]erify\s+(?:that\s+)?[Oo]llama\s+(?:is\s+)?(?:responding|running|up)", re.IGNORECASE),
+     "Verify system services are healthy using available tools"),
+    (re.compile(r"\bcurl\s+localhost:\d+", re.IGNORECASE),
+     "use available tools to check service health"),
+    (re.compile(r"\bcurl\s+127\.0\.0\.1:\d+", re.IGNORECASE),
+     "use available tools to check service health"),
+    (re.compile(r"\bpoll\s+localhost", re.IGNORECASE),
+     "check service health via available tools"),
+]
+
+
+def _rewrite_cloud_incompatible_prompt(prompt: str, base_url: str) -> str:
+    """Rewrite prompt instructions that assume local service access when running on cloud.
+
+    When a cron job runs on a cloud inference endpoint (Nous, OpenRouter, Anthropic),
+    instructions to "Check Ollama" or "curl localhost:11434" are impossible.
+    Instead of just warning, this rewrites the instruction to a cloud-compatible
+    equivalent that the agent can actually execute.
+
+    Returns the (possibly rewritten) prompt.
+    """
+    try:
+        from agent.model_metadata import is_local_endpoint
+    except ImportError:
+        return prompt
+
+    if is_local_endpoint(base_url or ""):
+        return prompt  # Local — no rewrite needed
+
+    rewritten = prompt
+    for pattern, replacement in _CLOUD_INCOMPATIBLE_PATTERNS:
+        rewritten = pattern.sub(replacement, rewritten)
+
+    if rewritten != prompt:
+        rewritten = (
+            "[NOTE: Some instructions were adjusted for cloud execution. "
+            "Local service checks were rewritten to use available tools.]
+
+"
+            + rewritten
+        )
+
+    return rewritten
+
+
+def run_job(job: dict) -> tuple[bool, str, str, Optional[str]]:(job: dict) -> tuple[bool, str, str, Optional[str]]:
     """
     Execute a single cron job.
     
@@ -767,36 +717,7 @@ def run_job(job: dict) -> tuple[bool, str, str, Optional[str]]:
     
     job_id = job["id"]
     job_name = job["name"]
-
-    # ── Early model/provider resolution ───────────────────────────────────
-    # We need the model name before building the prompt so the runtime
-    # context block can be injected.  Full provider resolution happens
-    # later (smart routing, etc.) but the basic name is enough here.
-    _early_model = job.get("model") or os.getenv("HERMES_MODEL") or ""
-    _early_provider = os.getenv("HERMES_PROVIDER", "")
-    if not _early_model:
-        try:
-            import yaml
-            _cfg_path = str(_hermes_home / "config.yaml")
-            if os.path.exists(_cfg_path):
-                with open(_cfg_path) as _f:
-                    _cfg_early = yaml.safe_load(_f) or {}
-                _mc = _cfg_early.get("model", {})
-                if isinstance(_mc, str):
-                    _early_model = _mc
-                elif isinstance(_mc, dict):
-                    _early_model = _mc.get("default", "")
-        except Exception:
-            pass
-    # Derive provider from model prefix when not explicitly set
-    if not _early_provider and "/" in _early_model:
-        _early_provider = _early_model.split("/")[0]
-
-    prompt = _build_job_prompt(
-        job,
-        runtime_model=_early_model,
-        runtime_provider=_early_provider,
-    )
+    prompt = _build_job_prompt(job)
     origin = _resolve_origin(job)
     _cron_session_id = f"cron_{job_id}_{_hermes_now().strftime('%Y%m%d_%H%M%S')}"
 
@@ -908,20 +829,6 @@ def run_job(job: dict) -> tuple[bool, str, str, Optional[str]]:
             message = format_runtime_provider_error(exc)
             raise RuntimeError(message) from exc
 
-        # ── Provider mismatch warning ─────────────────────────────────
-        # If the job prompt references a provider different from the one
-        # we actually resolved, warn so operators know which prompts are stale.
-        _resolved_provider = runtime.get("provider", "") or ""
-        _raw_prompt = job.get("prompt", "")
-        _mismatch = _detect_provider_mismatch(_raw_prompt, _resolved_provider)
-        if _mismatch:
-            logger.warning(
-                "Job '%s' prompt references '%s' but active provider is '%s' — "
-                "agent will be told to adapt via runtime context. "
-                "Consider updating this job's prompt.",
-                job_name, _mismatch, _resolved_provider,
-            )
-
         from agent.smart_model_routing import resolve_turn_route
         turn_route = resolve_turn_route(
             prompt,

tests/test_cron_provider_mismatch.py

@@ -1,125 +0,0 @@
-"""Tests for cron scheduler: provider mismatch detection, runtime classification,
-and capability-aware prompt building."""
-
-import sys
-from pathlib import Path
-
-sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
-
-
-def _import_scheduler():
-    """Import the scheduler module, bypassing __init__.py re-exports that may
-    reference symbols not yet merged upstream."""
-    import importlib.util
-    spec = importlib.util.spec_from_file_location(
-        "cron.scheduler", str(Path(__file__).resolve().parent.parent / "cron" / "scheduler.py"),
-    )
-    mod = importlib.util.module_from_spec(spec)
-    try:
-        spec.loader.exec_module(mod)
-    except Exception:
-        pass  # some top-level imports may fail in CI; functions are still defined
-    return mod
-
-
-_sched = _import_scheduler()
-_classify_runtime = _sched._classify_runtime
-_detect_provider_mismatch = _sched._detect_provider_mismatch
-_build_job_prompt = _sched._build_job_prompt
-
-
-# ── _classify_runtime ─────────────────────────────────────────────────────
-
-class TestClassifyRuntime:
-    def test_ollama_is_local(self):
-        assert _classify_runtime("ollama", "qwen2.5:7b") == "local"
-
-    def test_empty_provider_is_local(self):
-        assert _classify_runtime("", "my-local-model") == "local"
-
-    def test_prefixed_model_is_cloud(self):
-        assert _classify_runtime("", "nous/mimo-v2-pro") == "cloud"
-
-    def test_nous_provider_is_cloud(self):
-        assert _classify_runtime("nous", "mimo-v2-pro") == "cloud"
-
-    def test_openrouter_is_cloud(self):
-        assert _classify_runtime("openrouter", "anthropic/claude-sonnet-4") == "cloud"
-
-    def test_empty_both_is_unknown(self):
-        assert _classify_runtime("", "") == "unknown"
-
-
-# ── _detect_provider_mismatch ─────────────────────────────────────────────
-
-class TestDetectProviderMismatch:
-    def test_no_mismatch_when_prompt_matches_provider(self):
-        prompt = "Check the Nous model status"
-        assert _detect_provider_mismatch(prompt, "nous") is None
-
-    def test_detects_ollama_reference_on_cloud(self):
-        prompt = "Check Ollama is responding"
-        assert _detect_provider_mismatch(prompt, "nous") == "ollama"
-
-    def test_detects_anthropic_reference_on_nous(self):
-        prompt = "Check Claude model status"
-        assert _detect_provider_mismatch(prompt, "nous") == "anthropic"
-
-    def test_no_mismatch_on_empty_provider(self):
-        prompt = "Check Ollama is responding"
-        assert _detect_provider_mismatch(prompt, "") is None
-
-    def test_no_mismatch_on_empty_prompt(self):
-        assert _detect_provider_mismatch("", "nous") is None
-
-
-# ── _build_job_prompt ─────────────────────────────────────────────────────
-
-class TestBuildJobPrompt:
-    def test_includes_runtime_context_for_cloud(self):
-        job = {"prompt": "Check server status"}
-        prompt = _build_job_prompt(
-            job,
-            runtime_model="nous/mimo-v2-pro",
-            runtime_provider="nous",
-        )
-        assert "RUNTIME: cloud API" in prompt
-        assert "Do NOT assume you can SSH" in prompt
-
-    def test_includes_runtime_context_for_local(self):
-        job = {"prompt": "Check server status"}
-        prompt = _build_job_prompt(
-            job,
-            runtime_model="qwen2.5:7b",
-            runtime_provider="ollama",
-        )
-        assert "RUNTIME: local" in prompt
-        assert "local Ollama" in prompt
-
-    def test_no_runtime_block_when_no_runtime_info(self):
-        job = {"prompt": "Check server status"}
-        prompt = _build_job_prompt(job)
-        assert "RUNTIME:" not in prompt
-
-    def test_includes_model_in_runtime_block(self):
-        job = {"prompt": "Check server status"}
-        prompt = _build_job_prompt(
-            job,
-            runtime_model="nous/mimo-v2-pro",
-            runtime_provider="nous",
-        )
-        assert "MODEL: nous/mimo-v2-pro" in prompt
-
-    def test_includes_provider_in_runtime_block(self):
-        job = {"prompt": "Check server status"}
-        prompt = _build_job_prompt(
-            job,
-            runtime_model="nous/mimo-v2-pro",
-            runtime_provider="nous",
-        )
-        assert "PROVIDER: nous" in prompt
-
-
-if __name__ == "__main__":
-    import pytest
-    pytest.main([__file__, "-v"])