@gamaze/hicortex 0.7.0 → 0.10.0

package/hermes-plugin/hicortex/provider.py

@@ -0,0 +1,432 @@
+"""Hicortex MemoryProvider for Hermes — recall-only.
+
+Recall:   prefetch()          -> GET /search   (relevant memories before each turn)
+          queue_prefetch()    -> GET /search   (background recall for the next turn)
+          tools               -> hicortex_search / hicortex_recall_recent
+          system_prompt_block -> lessons + memory index injected into the prompt
+
+Capture is NOT the plugin's job. A nightly reader on the Hicortex server
+distills each agent's own session store (Hermes: ~/.hermes/profiles/<agent>/
+state.db) centrally — see specs/2026-07-01-memory-capture-architecture.md. This
+plugin has no local LLM, no spool, no timer, and no capture path.
+"""
+
+from __future__ import annotations
+
+import hashlib
+import json
+import logging
+import threading
+from typing import Any, Dict, List, Optional
+
+from agent.memory_provider import MemoryProvider
+
+from .client import HicortexClient
+from .config import CONFIG_SCHEMA, load_config
+
+logger = logging.getLogger(__name__)
+
+_INJECT_CONTENT_CAP = 500
+
+
+class HicortexProvider(MemoryProvider):
+    """Hicortex long-term memory backend for Hermes (recall-only)."""
+
+    def __init__(self):
+        self._client: Optional[HicortexClient] = None
+        self._project: Optional[str] = None
+        self._recall_limit: int = 5
+        self._privacy: Optional[str] = "WORK,PERSONAL"
+        self._prefetch_cache: Dict[str, str] = {}
+        self._bg_threads: List[threading.Thread] = []
+
+    @property
+    def name(self) -> str:
+        return "hicortex"
+
+    # ------------------------------------------------------------------ config
+    def _build_client(self) -> Optional[HicortexClient]:
+        cfg = load_config()
+        url = cfg.get("hicortex_url")
+        if not url:
+            return None
+        token = cfg.get("hicortex_auth_token")
+        return HicortexClient(url, auth_token=token or None)
+
+    def _client_or_none(self) -> Optional[HicortexClient]:
+        if self._client is None:
+            try:
+                self._client = self._build_client()
+            except Exception as e:
+                logger.warning("hicortex: failed to build client: %s", e)
+        return self._client
+
+    def is_available(self) -> bool:
+        """Configured and ready — NO network call (per MemoryProvider contract).
+
+        ``is_available`` runs at agent init to decide whether to activate this
+        provider. Pinging the server here would mean a slow or momentarily-down
+        server silently disables memory for the whole session. Per the contract
+        ("should not make network calls — just check config and installed deps")
+        we only verify a server URL is configured; per-request failures are
+        handled at use time.
+        """
+        return self._build_client() is not None
+
+    def initialize(self, session_id: str, **kwargs) -> None:
+        cfg = load_config()
+        self._project = cfg.get("default_project") or None
+        try:
+            self._recall_limit = int(cfg.get("recall_limit", 5))
+        except (TypeError, ValueError):
+            self._recall_limit = 5
+        self._privacy = cfg.get("privacy_filter", "WORK,PERSONAL")
+        try:
+            self._client = self._build_client()
+        except Exception as e:
+            logger.warning("hicortex: init client build failed: %s", e)
+
+    # ------------------------------------------------------------------- recall
+    def _format_hits(self, hits: list[dict]) -> str:
+        if not hits:
+            return ""
+        lines = [
+            "Relevant prior context from your long-term memory "
+            "(verify before relying on these — each shows date and project):"
+        ]
+        for h in hits[: self._recall_limit]:
+            date = (h.get("created_at") or "")[:10]
+            proj = h.get("project") or "global"
+            content = (h.get("content") or "").strip().replace("\n", " ")
+            if len(content) > _INJECT_CONTENT_CAP:
+                content = content[:_INJECT_CONTENT_CAP] + "…"
+            lines.append(f"- [{date}, {proj}] {content}")
+        return "\n".join(lines)
+
+    def prefetch(self, query: str, *, session_id: str = "") -> str:
+        key = hashlib.sha1(query.encode("utf-8")).hexdigest()
+        cached = self._prefetch_cache.pop(key, None)
+        if cached is not None:
+            return cached
+        client = self._client_or_none()
+        if client is None:
+            return ""
+        try:
+            hits = client.search(
+                query, limit=self._recall_limit, project=self._project, privacy=self._privacy
+            )
+            return self._format_hits(hits)
+        except Exception as e:
+            logger.debug("hicortex prefetch failed: %s", e)
+            return ""
+
+    def queue_prefetch(self, query: str, *, session_id: str = "") -> None:
+        client = self._client_or_none()
+        if client is None:
+            return
+        key = hashlib.sha1(query.encode("utf-8")).hexdigest()
+
+        def _bg() -> None:
+            try:
+                hits = client.search(
+                    query, limit=self._recall_limit, project=self._project, privacy=self._privacy
+                )
+                self._prefetch_cache[key] = self._format_hits(hits)
+            except Exception as e:
+                logger.debug("hicortex queue_prefetch failed: %s", e)
+
+        self._spawn(_bg)
+
+    # ------------------------------------------------------ system prompt/tools
+    def system_prompt_block(self) -> str:
+        client = self._client_or_none()
+        if client is None:
+            return ""
+        try:
+            data = client.lessons()
+        except Exception as e:
+            logger.debug("hicortex lessons fetch failed: %s", e)
+            return ""
+        lessons = (data.get("lessons") or [])[:8]
+        idx = data.get("index") or {}
+        lines = [
+            "## Hicortex long-term memory",
+            "You have shared long-term memory across sessions. Use `hicortex_search` "
+            "for specific recall and `hicortex_recall_recent` for recent context.",
+        ]
+        if lessons:
+            lines.append("Lessons:")
+            for l in lessons:
+                c = (l.get("content") or "").strip().replace("\n", " ")
+                lines.append(f"- {c[:200]}")
+        if idx.get("total"):
+            lines.append(
+                f"({idx.get('total')} memories, {idx.get('lessonCount')} lessons "
+                f"across {idx.get('sourceCount')} agents)"
+            )
+        return "\n".join(lines)
+
+    def get_tool_schemas(self) -> List[Dict[str, Any]]:
+        return [
+            {
+                "name": "hicortex_search",
+                "description": (
+                    "Search long-term memory using semantic similarity. Returns the most "
+                    "relevant memories from past sessions."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "query": {"type": "string", "description": "Search query text"},
+                        "limit": {
+                            "type": "number",
+                            "description": "Max results (default 5)",
+                        },
+                        "project": {"type": "string", "description": "Filter by project name"},
+                    },
+                    "required": ["query"],
+                },
+            },
+            {
+                "name": "hicortex_recall_recent",
+                "description": "Recall recent context memories, optionally filtered by project.",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "project": {"type": "string"},
+                        "limit": {
+                            "type": "number",
+                            "description": "Max results (default 10)",
+                        },
+                    },
+                },
+            },
+            {
+                "name": "hicortex_context",
+                "description": (
+                    "Get recent context memories, optionally filtered by project. "
+                    "Useful to recall what happened recently."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "project": {"type": "string", "description": "Filter by project name"},
+                        "limit": {"type": "number", "description": "Max results (default 10)"},
+                    },
+                },
+            },
+            {
+                "name": "hicortex_ingest",
+                "description": (
+                    "Store a new memory in long-term storage. "
+                    "Use for important facts, decisions, or lessons."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "content": {"type": "string", "description": "Memory content to store"},
+                        "project": {"type": "string", "description": "Project this memory belongs to"},
+                        "memory_type": {
+                            "type": "string",
+                            "enum": ["episode", "lesson", "fact", "decision"],
+                            "description": "Type of memory (default: episode)",
+                        },
+                    },
+                    "required": ["content"],
+                },
+            },
+            {
+                "name": "hicortex_lessons",
+                "description": (
+                    "Get actionable lessons learned from past sessions. "
+                    "Auto-generated insights about mistakes to avoid."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "project": {"type": "string", "description": "Filter by project name"},
+                    },
+                },
+            },
+            {
+                "name": "hicortex_index",
+                "description": (
+                    "Get the knowledge domain index — shows what topics and projects "
+                    "are stored in memory, grouped by domain."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {},
+                },
+            },
+            {
+                "name": "hicortex_graph",
+                "description": (
+                    "Query the memory knowledge graph — find connected memories, "
+                    "hub nodes, or paths between memories."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "operation": {
+                            "type": "string",
+                            "enum": ["neighbors", "hubs", "path"],
+                            "description": "Graph operation to perform",
+                        },
+                        "id": {"type": "string", "description": "Memory ID (required for neighbors and path operations)"},
+                        "target_id": {"type": "string", "description": "Target memory ID (required for path operation)"},
+                        "limit": {"type": "number", "description": "Max results (default 10)"},
+                        "domain": {"type": "string", "description": "Filter hubs by domain"},
+                        "relationship": {
+                            "type": "string",
+                            "description": "Filter neighbors by relationship type (e.g., CONTRADICTS, SUPERSEDES, derives)",
+                        },
+                    },
+                    "required": ["operation"],
+                },
+            },
+            {
+                "name": "hicortex_update",
+                "description": (
+                    "Update an existing memory. Use after searching to fix incorrect information. "
+                    "If content changes, the embedding is re-computed."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "id": {"type": "string", "description": "Memory ID (from search results, first 8 chars or full UUID)"},
+                        "content": {"type": "string", "description": "New content text"},
+                        "project": {"type": "string", "description": "New project name"},
+                        "memory_type": {
+                            "type": "string",
+                            "enum": ["episode", "lesson", "fact", "decision"],
+                            "description": "New memory type",
+                        },
+                    },
+                    "required": ["id"],
+                },
+            },
+            {
+                "name": "hicortex_delete",
+                "description": (
+                    "Permanently delete a memory and its links. "
+                    "Use when a memory is incorrect and should be removed entirely."
+                ),
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "id": {"type": "string", "description": "Memory ID (from search results, first 8 chars or full UUID)"},
+                    },
+                    "required": ["id"],
+                },
+            },
+        ]
+
+    def handle_tool_call(self, tool_name: str, args: Dict[str, Any], **kwargs) -> str:
+        client = self._client_or_none()
+        if client is None:
+            return json.dumps({"error": "hicortex server not configured"})
+        try:
+            if tool_name == "hicortex_search":
+                hits = client.search(
+                    args.get("query", ""),
+                    limit=int(args.get("limit", 5)),
+                    project=args.get("project") or self._project,
+                )
+                return json.dumps(hits)
+
+            elif tool_name in ("hicortex_recall_recent", "hicortex_context"):
+                hits = client.context(
+                    project=args.get("project") or self._project,
+                    limit=int(args.get("limit", 10)),
+                )
+                return json.dumps(hits)
+
+            elif tool_name == "hicortex_ingest":
+                content = args.get("content", "")
+                if not content:
+                    return json.dumps({"error": "content is required"})
+                status, resp = client.ingest(
+                    content=content,
+                    source_agent="hermes/manual",
+                    project=args.get("project") or self._project,
+                    memory_type=args.get("memory_type", "episode"),
+                )
+                if status not in (200, 201):
+                    return json.dumps({"error": resp.get("error", f"HTTP {status}")})
+                id_val = resp.get("id") or ""
+                return json.dumps({"id": id_val, "message": f"Memory stored (id: {id_val[:8]})"})
+
+            elif tool_name == "hicortex_lessons":
+                data = client.lessons()
+                lessons = (data.get("lessons") or [])
+                if not lessons:
+                    return json.dumps({"message": "No lessons found."})
+                return json.dumps([{"content": l.get("content", "")[:500]} for l in lessons])
+
+            elif tool_name == "hicortex_index":
+                return json.dumps(client.index())
+
+            elif tool_name == "hicortex_graph":
+                op = args.get("operation", "")
+                result = client.graph(
+                    op=op,
+                    id=args.get("id"),
+                    target_id=args.get("target_id"),
+                    limit=args.get("limit"),
+                    domain=args.get("domain"),
+                    relationship=args.get("relationship"),
+                )
+                return json.dumps(result)
+
+            elif tool_name == "hicortex_update":
+                id_val = args.get("id", "")
+                if not id_val:
+                    return json.dumps({"error": "id is required"})
+                status, resp = client.update(
+                    id=id_val,
+                    content=args.get("content"),
+                    project=args.get("project"),
+                    memory_type=args.get("memory_type"),
+                )
+                if status == 404:
+                    return json.dumps({"error": f"Memory not found: {id_val}"})
+                if status not in (200, 201):
+                    return json.dumps({"error": resp.get("error", f"HTTP {status}")})
+                return json.dumps({"updated": True, "id": resp.get("id", id_val)})
+
+            elif tool_name == "hicortex_delete":
+                id_val = args.get("id", "")
+                if not id_val:
+                    return json.dumps({"error": "id is required"})
+                status, resp = client.delete(id=id_val)
+                if status == 404:
+                    return json.dumps({"error": f"Memory not found: {id_val}"})
+                if status not in (200, 201):
+                    return json.dumps({"error": resp.get("error", f"HTTP {status}")})
+                return json.dumps({"deleted": True, "id": resp.get("id", id_val)})
+
+            else:
+                return json.dumps({"error": f"unknown tool: {tool_name}"})
+
+        except Exception as e:
+            return json.dumps({"error": str(e)})
+
+    # ---------------------------------------------------------------- lifecycle
+    def _spawn(self, fn) -> None:
+        self._bg_threads = [t for t in self._bg_threads if t.is_alive()]
+        t = threading.Thread(target=fn, daemon=True)
+        t.start()
+        self._bg_threads.append(t)
+
+    def shutdown(self) -> None:
+        for t in self._bg_threads:
+            t.join(timeout=2.0)
+
+    def get_config_schema(self) -> List[Dict[str, Any]]:
+        return CONFIG_SCHEMA
+
+    def save_config(self, values: Dict[str, Any], hermes_home: str) -> None:
+        from .config import save_config as _save
+
+        _save(values, hermes_home)

package/openclaw.plugin.json

@@ -2,69 +2,42 @@
   "id": "hicortex",
   "name": "Hicortex — Long-term Memory That Learns",
   "description": "Your agents remember past decisions, avoid repeated mistakes, and get smarter every day. Nightly reflection generates actionable lessons that automatically update agent behavior.",
-  "version": "0.7.0",
+  "version": "0.10.0",
   "kind": "lifecycle",
   "skills": ["./skills/hicortex-memory", "./skills/hicortex-learn", "./skills/hicortex-activate"],
   "configSchema": {
     "type": "object",
     "properties": {
-      "licenseKey": {
-        "type": "string",
-        "description": "Hicortex license key (hctx-...). Leave empty for free tier (250 memory cap)."
-      },
-      "llmBaseUrl": {
-        "type": "string",
-        "description": "LLM API base URL. Auto-detected from OpenClaw config or env vars if omitted."
-      },
-      "llmApiKey": {
-        "type": "string",
-        "description": "LLM API key. Auto-detected from OpenClaw config or env vars if omitted."
-      },
-      "llmModel": {
+      "serverUrl": {
         "type": "string",
-        "default": "qwen3.5:4b",
-        "description": "Model for importance scoring and distillation (local recommended)"
+        "default": "http://127.0.0.1:8787",
+        "description": "Hicortex server URL. Defaults to localhost (co-located server). For multi-machine setups, point this at the remote server (e.g. http://bedrock:8787 or a Tailscale HTTPS URL)."
       },
-      "reflectModel": {
+      "authToken": {
         "type": "string",
-        "default": "qwen3.5:cloud",
-        "description": "Model for nightly reflection (cloud recommended)"
+        "description": "Bearer token for the Hicortex server. Localhost bypasses auth automatically. Required for remote servers — use the authToken from ~/.hicortex/config.json on the server."
       },
-      "consolidateHour": {
-        "type": "number",
-        "default": 2,
-        "description": "Hour (0-23, local time) to run nightly consolidation"
-      },
-      "dbPath": {
+      "licenseKey": {
         "type": "string",
-        "description": "Custom path for the SQLite database file"
+        "description": "Hicortex license key (hctx-...). Leave empty for free tier (250 memory cap)."
       }
     },
     "required": []
   },
   "uiHints": {
+    "serverUrl": {
+      "label": "Server URL",
+      "placeholder": "http://127.0.0.1:8787"
+    },
+    "authToken": {
+      "label": "Auth Token",
+      "sensitive": true,
+      "placeholder": "Required for remote servers"
+    },
     "licenseKey": {
       "label": "License Key",
       "placeholder": "hctx-... (optional, free tier without key)",
       "sensitive": true
-    },
-    "llmBaseUrl": {
-      "label": "LLM API URL",
-      "placeholder": "Auto-detected if empty"
-    },
-    "llmApiKey": {
-      "label": "LLM API Key",
-      "sensitive": true,
-      "placeholder": "Auto-detected if empty"
-    },
-    "llmModel": {
-      "label": "Scoring Model"
-    },
-    "reflectModel": {
-      "label": "Reflection Model"
-    },
-    "consolidateHour": {
-      "label": "Consolidation Hour"
     }
   }
 }

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@gamaze/hicortex",
-  "version": "0.7.0",
-  "description": "Human-like memory for self-improving AI agents. Automatic capturing, nightly reflection, and cross-agent learning. Works with Claude Code and OpenClaw.",
+  "version": "0.10.0",
+  "description": "Self-learning memory for AI agents \u2014 experience captured automatically, distilled into lessons overnight, shared across your whole fleet. Works with Hermes, OpenClaw, Claude Code, and Pi.",
   "main": "dist/index.js",
   "bin": {
     "hicortex": "dist/cli.js"
@@ -26,14 +26,16 @@
   "files": [
     "dist/",
     "skills/",
+    "hermes-plugin/",
     "openclaw.plugin.json",
     "README.md"
   ],
   "scripts": {
-    "build": "tsc",
+    "build": "rm -rf dist && tsc",
     "test": "vitest run",
     "test:watch": "vitest",
-    "prepublishOnly": "npm run build"
+    "prepack": "npm run build && rm -rf ./hermes-plugin && mkdir -p ./hermes-plugin && cp -r ../../hermes-plugin/hicortex ./hermes-plugin/ && find ./hermes-plugin -name __pycache__ -type d -exec rm -rf {} + 2>/dev/null || true",
+    "prepublishOnly": "npm run build && rm -rf ./hermes-plugin && mkdir -p ./hermes-plugin && cp -r ../../hermes-plugin/hicortex ./hermes-plugin/ && find ./hermes-plugin -name __pycache__ -type d -exec rm -rf {} + 2>/dev/null || true"
   },
   "devDependencies": {
     "@types/better-sqlite3": "^7.6.0",
@@ -45,7 +47,7 @@
   "engines": {
     "node": ">=18"
   },
-  "license": "MIT",
+  "license": "PolyForm-Noncommercial-1.0.0",
   "homepage": "https://hicortex.gamaze.com",
   "repository": {
     "type": "git",

package/dist/pro-loader.d.ts

@@ -1,33 +0,0 @@
-/**
- * Pro extension loader.
- *
- * Called from features.ts at boot when a valid paid license is detected.
- * Responsibilities:
- *   1. Check ~/.hicortex/pro/installed.json for the currently-installed
- *      Pro version (if any).
- *   2. Fetch GET /api/pro/meta from hicortex.gamaze.com to discover the
- *      latest available version for the caller's license tier.
- *   3. If the installed version is older (or missing), download the
- *      tarball, verify the sha256 sidecar, and extract to ~/.hicortex/pro/.
- *   4. Dynamic-import ~/.hicortex/pro/package/index.js and call activate()
- *      on its default export with a ProActivationContext.
- *
- * Failure modes (all soft — OSS host keeps running with defaults):
- *   - Network to /api/pro/meta fails → use whatever is already installed
- *   - No cached Pro and network fails → Pro not activated, OSS defaults apply
- *   - Downloaded tarball fails sha256 → abort download, keep old version
- *   - import() of activated module throws → log warning, keep defaults
- *
- * The loader is strictly best-effort. It must NEVER crash the OSS host.
- */
-/**
- * Entry point called from features.ts at boot.
- *
- * @param licenseKey   The Pro license key (hctx-...) from config
- * @param stateDir     Usually ~/.hicortex/
- * @param hostVersion  The version of the OSS host (from package.json) — passed
- *                     to the Pro activate() for compatibility gating
- * @param serverUrl    Override for the Pro meta/download endpoint (defaults
- *                     to https://hicortex.gamaze.com). Useful for testing.
- */
-export declare function loadPro(licenseKey: string, stateDir: string, hostVersion: string, serverUrl?: string): Promise<void>;