@misterhuydo/sentinel 1.5.42 → 1.5.44

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@misterhuydo/sentinel",
-  "version": "1.5.42",
+  "version": "1.5.44",
   "description": "Sentinel — Autonomous DevOps Agent installer and manager",
   "bin": {
     "sentinel": "./bin/sentinel.js"

package/python/sentinel/__init__.py

@@ -1 +1 @@
-__version__ = "1.5.42"
+__version__ = "1.5.44"

package/python/sentinel/config_loader.py

@@ -105,6 +105,7 @@ class LogSourceConfig:
     cf_url: str = ""
     cf_token: str = ""
     sync_enabled: bool = True
+    health_url: str = ""   # optional HTTP health endpoint for this log source's service
 
 
 @dataclass
@@ -308,6 +309,7 @@ class ConfigLoader:
             s.cf_token = d.get("CF_TOKEN", "")
             s.target_repo = d.get("TARGET_REPO", "auto")
             s.sync_enabled = d.get("SYNC_ENABLED", "true").lower() != "false"
+            s.health_url = d.get("HEALTH_URL", "")
             self.log_sources[s.name] = s
 
     def _load_repos(self):

package/python/sentinel/sentinel_boss.py

@@ -157,7 +157,7 @@ COMPLETE TOOL REFERENCE
                         custom fetch, and filter_logs searches them automatically.
                         Use grep_filter for INFO-level or feature-specific patterns that the default
                         WARN/ERROR filter would miss.
-                        "fetch logs", "fetch SSOLWA with filter provision/phone", "fetch without filter"
+                        "fetch logs", "fetch logs for <source> with filter <pattern>", "fetch without filter"
 
  7. search_logs         Live SSH grep on production servers using GREP_FILTER.
                         Falls back to cached files if SSH unavailable.
@@ -171,7 +171,11 @@ COMPLETE TOOL REFERENCE
  9. tail_log            Last N lines of a log source live, no filter.
                         "show recent SSOLWA logs", "tail STS", "last 200 lines from 1881"
 
-10. ask_logs            Ask Claude Code to read and reason over log history.
+10. check_health        Call HEALTH_URL for a log source or repo — returns live version + status.
+                        Use whenever deployment status needs verification. NEVER guess from logs.
+                        "is version X deployed?", "what version is <service> running?", "is the service up?"
+
+11. ask_logs            Ask Claude Code to read and reason over log history.
                         Use for summarisation, pattern detection, trend analysis.
                         "what caused 400s in 1881 logs?", "summarise last week of STS logs"
 
@@ -576,9 +580,10 @@ CORRECT:
   "No '<feature log pattern>' lines yet — the code path hasn't been triggered since the last
   fetch. This says nothing about whether the release is deployed."
 
-To verify if a release is live, offer to search for startup log lines:
-  search_logs with query "Starting|started in|version|initialized"
-NEVER repeat a deployment assertion from a prior turn without doing this check first.
+To verify if a release is live:
+  1. FIRST: call check_health for the relevant source/repo — it returns the live version directly.
+  2. If no HEALTH_URL is configured: search_logs with query "Starting|started in|version|initialized".
+NEVER state deployment status without calling check_health first.
 - If a tool call will take a moment (search, fetch, pull), prefix your reply with a brief "working" line ending in "..." before the results, e.g. "Searching SSOLWA for TryDig activity..." then the actual output.
   Never just say a working line and stop — always follow it with the results in the same message.
 
@@ -1443,6 +1448,30 @@ _TOOLS = [
             "required": ["question"],
         },
     },
+    {
+        "name": "check_health",
+        "description": (
+            "Call the HEALTH_URL for a log source or repo and return the raw JSON response. "
+            "Use this to verify whether a service is running and what version it reports. "
+            "ALWAYS call this instead of guessing deployment status from log evidence. "
+            "Use for: 'is version X deployed?', 'what version is the service running?', "
+            "'is the service up?', 'verify the release is live'."
+        ),
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "source": {
+                    "type": "string",
+                    "description": "Log source name (e.g. 'SSOLWA') or repo name to check. "
+                                   "Checks HEALTH_URL from that source/repo config.",
+                },
+                "url": {
+                    "type": "string",
+                    "description": "Direct health URL to call (overrides config). Use when the user provides a URL.",
+                },
+            },
+        },
+    },
     {
         "name": "post_file",
         "description": (
@@ -2040,6 +2069,36 @@ def _filter_log_sources(props_files: list, source_hint: str) -> list:
     return matched
 
 
+def _auto_health_check(source_hint: str, cfg_loader) -> dict | None:
+    """
+    If any log source or repo matching source_hint has a HEALTH_URL, call it and return
+    the result so it can be injected into fetch/filter responses automatically.
+    Returns None if no health URL is configured or the call fails silently.
+    """
+    try:
+        import requests as _req
+        url = ""
+        for src_name, src in cfg_loader.log_sources.items():
+            if source_hint.lower() in src_name.lower() and getattr(src, "health_url", ""):
+                url = src.health_url
+                break
+        if not url:
+            for repo_name, repo in cfg_loader.repos.items():
+                if source_hint.lower() in repo_name.lower() and repo.health_url:
+                    url = repo.health_url
+                    break
+        if not url:
+            return None
+        resp = _req.get(url, timeout=8)
+        try:
+            data = resp.json()
+        except Exception:
+            data = resp.text[:500]
+        return {"health_url": url, "status_code": resp.status_code, "health": data}
+    except Exception:
+        return None
+
+
 # ── Tool execution ────────────────────────────────────────────────────────────
 
 async def _run_tool(name: str, inputs: dict, cfg_loader, store, slack_client=None, user_id: str = "", channel: str = "", is_admin: bool = False) -> str:
@@ -2900,6 +2959,7 @@ async def _run_tool(name: str, inputs: dict, cfg_loader, store, slack_client=Non
         sources_searched = [d.name + (" [temp]" if is_temp else "") for d, is_temp in search_pairs]
         if total == 0:
             has_temp = bool(temp_dirs)
+            health = _auto_health_check(source_f, cfg_loader) if source_f else None
             return json.dumps({
                 "query": query_f,
                 "total_matches": 0,
@@ -2910,6 +2970,7 @@ async def _run_tool(name: str, inputs: dict, cfg_loader, store, slack_client=Non
                     + "If searching for a specific log line from a new feature, use fetch_logs "
                     "with a matching grep_filter first — the default filter only captures WARN/ERROR."
                 ),
+                **({"service_health": health} if health else {}),
             })
 
         # Pattern grouping: count occurrences of each error signature
@@ -2951,6 +3012,7 @@ async def _run_tool(name: str, inputs: dict, cfg_loader, store, slack_client=Non
         except Exception:
             pass
 
+        health = _auto_health_check(source_f, cfg_loader) if source_f else None
         return json.dumps({
             "query":            query_f,
             "total_matches":    total,
@@ -2960,6 +3022,7 @@ async def _run_tool(name: str, inputs: dict, cfg_loader, store, slack_client=Non
             "sample_lines":     sample_lines,
             "time_span":        time_span,
             "capped":           total >= max_matches,
+            **({"service_health": health} if health else {}),
         })
 
     if name == "trigger_poll":
@@ -3112,7 +3175,12 @@ async def _run_tool(name: str, inputs: dict, cfg_loader, store, slack_client=Non
             except Exception as e:
                 results.append({"source": props.stem, "error": str(e)})
 
-        return json.dumps({"fetched": len(results), "results": results})
+        health = _auto_health_check(source_filter, cfg_loader) if source_filter else None
+        return json.dumps({
+            "fetched": len(results),
+            "results": results,
+            **({"service_health": health} if health else {}),
+        })
 
     if name == "watch_bot":
         if not is_admin:
@@ -3552,6 +3620,50 @@ async def _run_tool(name: str, inputs: dict, cfg_loader, store, slack_client=Non
                 results.append({"project": d.name, "status": "error", "detail": str(e)})
         return json.dumps({"results": results})
 
+    if name == "check_health":
+        import requests as _requests
+        url = inputs.get("url", "").strip()
+        source_hint = inputs.get("source", "").strip().lower()
+
+        # Resolve URL from config if not provided directly
+        if not url:
+            # Check log sources first, then repos
+            for src_name, src in cfg_loader.log_sources.items():
+                if source_hint in src_name.lower() and getattr(src, "health_url", ""):
+                    url = src.health_url
+                    break
+            if not url:
+                for repo_name, repo in cfg_loader.repos.items():
+                    if source_hint in repo_name.lower() and repo.health_url:
+                        url = repo.health_url
+                        break
+
+        if not url:
+            configured = {
+                **{n: s.health_url for n, s in cfg_loader.log_sources.items() if getattr(s, "health_url", "")},
+                **{n: r.health_url for n, r in cfg_loader.repos.items() if r.health_url},
+            }
+            return json.dumps({
+                "error": f"No HEALTH_URL found for '{source_hint}'.",
+                "configured_health_urls": configured or "none",
+                "hint": "Add HEALTH_URL=https://... to the log-source or repo .properties file, "
+                        "or pass url= directly.",
+            })
+
+        try:
+            resp = _requests.get(url, timeout=10)
+            try:
+                data = resp.json()
+            except Exception:
+                data = resp.text[:2000]
+            return json.dumps({
+                "url":         url,
+                "status_code": resp.status_code,
+                "response":    data,
+            })
+        except Exception as e:
+            return json.dumps({"url": url, "error": str(e)})
+
     if name == "tail_log":
         source      = inputs.get("source", "").lower()
         lines       = int(inputs.get("lines", 100))