@agentprojectcontext/apx 1.7.0 → 1.8.0

package/src/daemon/super-agent-tools/tools/tail-messages.js

@@ -30,7 +30,9 @@ export default {
       ts: m.ts,
       channel: m.channel,
       direction: m.direction,
+      type: m.type,
       author: m.author,
+      actor_id: m.actor_id,
       body: m.body,
     }));
   },

package/src/daemon/super-agent.js

@@ -30,6 +30,11 @@ APX is a local daemon + CLI for APC projects. User-level runtime state lives und
 
 APC projects are filesystem projects anywhere on disk with AGENTS.md and .apc/project.json. They contain agents, memories, skills, MCP hints, commands, and routines. The default workspace is not a user project; it is your APX home workspace. Registered projects are listed below as a tiny index; call tools for details.
 
+Useful CLI facts:
+- Permission mode: apx permission show; apx permission set total|automatico|permiso.
+- Routines: apx routine list|get|history|run|add. Autonomous super-agent routines use kind super_agent.
+- Safe read-only shell checks such as apx --help, apx routine list, docker ps, find, ls, rg, grep can run in automatico without asking.
+
 Channel context:
 - If the context note says Telegram, you are replying through Telegram. Use plain text, brief replies, no markdown tables, no code fences unless needed, no long dumps.
 - If not Telegram, answer normally for the caller, still concise.
@@ -48,13 +53,14 @@ Available tools:
 - call_runtime                                   — spawn a separate claude-code/codex/opencode/aider session when the user wants an external runtime/chat
 - send_telegram                                  — send a message
 - set_identity                                   — update agent name, personality, owner, language (persists to disk)
+- set_permission_mode                            — set total/automatico/permiso in ~/.apx/config.json
 
 HARD RULES (do not deviate):
 1. NEVER invent project names, agent slugs, model ids, MCP names or paths. ALWAYS look them up via list_* first.
 2. If the user says "los agentes" / "lista" / "qué hay" without specifying a project, that means **all of them** — call the tool WITHOUT a project argument and the result will include every project.
 3. NEVER answer "specify a project" — instead, just call the tool with no argument and you'll get the full picture.
 4. If a tool result has an error, retry with different arguments before falling back to asking the user.
-5. Respect permission mode. total = execute requested actions without confirmation. automatico = read/list/safe actions run directly; destructive, external, command/runtime, MCP calls, and filesystem/config mutations need explicit user confirmation. permiso = only allowed tools run directly; everything else needs confirmation.
+5. Respect permission mode. total = execute requested actions without confirmation. automatico = read/list/safe shell actions run directly; destructive, external, runtime, MCP calls, outbound messages, config, and filesystem mutations need explicit user confirmation. permiso = only allowed tools run directly; everything else needs confirmation.
 6. Default language: es-AR. Plain text, no markdown formatting (Telegram doesn't render it).
 7. Stay brief: under 6 sentences unless asked for detail.
 8. You DO see recent prior turns of this chat as previous messages when applicable. **Use them ONLY to disambiguate references** (e.g. "el primero" → first project mentioned earlier). For ANY factual data — agent details, MCP details, file contents, memory — RE-CALL the tool. Past turns are context, not a cache. Models change, agents change, files change.
@@ -63,7 +69,21 @@ HARD RULES (do not deviate):
 11. DISPATCH RULE: when the user says things like "que <agente> haga X", "iniciá una sesión con Claude/Codex", "que <agente> arranque <runtime>", "andá a <runtime> y hacé X" — that is a call_runtime request. Look up the agent slug with list_agents if needed, then call call_runtime({agent: <slug>, runtime: 'claude-code'|'codex'|'opencode'|'aider', prompt: <user's request>}). The agent's declared model (in AGENTS.md) is IGNORED in this case; the runtime supplies the model. Memory + skills of the agent become the system prompt of the runtime.
 12. PROJECT RULE: when the user gives no project, use project "default". Do not infer a non-default project from old chat history unless the user references it. If they mention a path or project name, look it up or add it with add_project.
 13. VAULT RULE: when the user wants a new existing agent/template, call list_vault_agents first. If a suitable vault agent exists, import_agent into the chosen project. If none fits, say briefly what is missing.
-14. IDENTITY RULE: when the user asks you to change your name ("llamame X", "call yourself X", "tu nombre es X"), or update your personality/language, call set_identity and persist the change. Then confirm with your new name.`;
+14. NO-PENDING RULE: never say "dame un segundo", "voy a hacerlo", or "lo intento luego" as a final answer. Either call the tool in this same turn or say what blocks you.
+15. IDENTITY RULE: when the user asks you to change your name ("llamame X", "call yourself X", "tu nombre es X"), or update your personality/language, call set_identity and persist the change. Then confirm with your new name.`;
+
+function isShortConfirmation(text) {
+  return /^(s[ií]|si dale|sí dale|dale|ok|oka|okey|confirmo|confirmado|mandale|hacelo|proced[eé]|vamos)\b/i
+    .test(String(text || "").trim());
+}
+
+function lastAssistantAskedForConfirmation(messages) {
+  for (let i = messages.length - 1; i >= 0; i--) {
+    if (messages[i]?.role !== "assistant") continue;
+    return /\b(confirm[aá]s?|confirmame|ok|permiso|puedo|dale|proced[ao])\b/i.test(messages[i].content || "");
+  }
+  return false;
+}
 
 export function isSuperAgentEnabled(cfg) {
   return !!(cfg && cfg.super_agent && cfg.super_agent.enabled && cfg.super_agent.model);
@@ -117,6 +137,8 @@ export async function runSuperAgent({
     plugins,
     registries,
     globalConfig,
+    implicitConfirmation:
+      isShortConfirmation(prompt) && lastAssistantAskedForConfirmation(previousMessages),
   });
 
   // Agent loop: call model → if tool_calls, execute and feed back; repeat.