pi-open-agents 0.1.16 → 0.1.18

package/README.md

@@ -13,6 +13,8 @@ Unified agent and subagent management for [pi](https://pi.dev), with [OpenCode](
 Replaces `pi-agent-mode` + `@johnnywu/pi-subagents` with one coherent plugin.
 
 > **Why pi?** Pi's minimalist core keeps system prompts under 1,000 tokens — making it exceptionally fast on local models and cheap on cloud APIs. pi-open-agents is built to leverage that minimalism. See [Pi vs OpenCode: Performance & Architecture](./PI-VS-OPencode.md) for a detailed comparison.
+>
+> 🐑 **Herd your agents.** pi-open-agents + [Herdr](https://herdr.dev) runs a full team of coding agents — persistent workspaces, live agent awareness, control from one screen. **[Set up the stack in 5 commands →](./PI-WITH-HERDR.md)**
 
 ---
 
@@ -82,6 +84,22 @@ have no permission system. `pi-open-agents` replaces both:
 
 ---
 
+## Popular requests, already covered
+
+Some of the most upvoted agent/subagent asks across the ecosystem are already
+here — or native in pi itself:
+
+| Asked elsewhere (👍) | Status in pi + pi-open-agents |
+|---|---|
+| `AGENTS.md` adoption — anthropics/claude-code#31005 👍336, openai/codex#12115 👍109 | ✅ Native in pi: `AGENTS.md`/`CLAUDE.md` discovery with `AGENTS.override.md` layering |
+| Agent Skills standard (`.agents/skills/`) — agentskills/agentskills#15 👍183 | ✅ Native in pi; shared agents from `.agents/` (#7 aligns to the dotagents layout) |
+| Tool restrictions that actually reach subagents — anthropics/claude-code#18950 👍70 | ✅ Since v0.1.14, `tools:`/`permission:` restrict the child toolset via `--tools` (full parent inheritance → #8) |
+| Reasoning effort per agent — Kilo-Org/kilocode#13105 👍8 | ✅ `thinking: off…xhigh` per agent, forwarded to the child process |
+| Delegation guards — Kilo-Org/kilocode#9985 | ✅ `allowedAgents` allowlist + `maxDepth` recursion limit |
+| Agent management UI — openai/codex#22321 👍41 | ✅ `/agent`, `/agents`, `/agent-search`, `Ctrl+Shift+M`, programmatic `set_agent` |
+
+---
+
 ## Features
 
 ### Per-agent model, thinking, and permissions
@@ -290,6 +308,24 @@ Optional cleanup:
 
 ---
 
+## Run a fleet 🐑
+
+Pi stays minimal on purpose — one agent, done well. When you need **many**,
+add [Herdr](https://herdr.dev): persistent workspaces, mouse-first UX, live
+agent awareness (`idle` / `working` / `blocked`), and cross-agent control
+from one screen — pi, or any mix of the 25 supported agent kinds. Six agents,
+six sessions, zero chaos: that's the exact workstation setup documented in
+**[pi × Herdr × pi-open-agents](./PI-WITH-HERDR.md)**.
+
+## Roadmap
+
+- **Parent permission inheritance** for subagents — #8
+- **Agent hot-reload** (or `/reload-agents`) — #9
+- **Per-call model override** in `subagent()` — #10
+- **Parallel subagent limit** — #11
+
+---
+
 ## Development
 
 **Contributing?** Please read the [Contributing Guide](./CONTRIBUTING.md) before

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-open-agents",
-  "version": "0.1.16",
+  "version": "0.1.18",
   "description": "Unified agent and subagent management for pi coding agent, with OpenCode-compatible agent definitions.",
   "keywords": [
     "pi-package",

package/src/primary/executor.ts

@@ -64,7 +64,8 @@ async function applyModel(agent: AgentDefinition, pi: ExtensionAPI, ctx: Extensi
  * Falls back to the session's current thinking level if agent doesn't specify one.
  */
 function applyThinking(agent: AgentDefinition, pi: ExtensionAPI): void {
-  pi.setThinkingLevel(agent.thinking);
+  // Cast: pi-open-agents accepts "max" (aligned with CLI); pi-agent-core type lags behind
+  pi.setThinkingLevel(agent.thinking as any);
 }
 
 // ─── Tool Filtering ──────────────────────────────────────────────────────────

package/src/primary/manager.ts

@@ -82,8 +82,8 @@ export class AgentManager {
       if (this.originalState.model) {
         await pi.setModel(this.originalState.model);
       }
-      // Restore thinking level
-      pi.setThinkingLevel(this.originalState.thinkingLevel);
+      // Restore thinking level (cast: pi-agent-core type lags behind on "max")
+      pi.setThinkingLevel(this.originalState.thinkingLevel as any);
       // Restore tools
       pi.setActiveTools(this.originalState.toolNames);
     } else {

package/src/types.ts

@@ -10,7 +10,7 @@
 
 export type AgentMode = "primary" | "subagent" | "all";
 
-export type ThinkingLevel = "off" | "minimal" | "low" | "medium" | "high" | "xhigh";
+export type ThinkingLevel = "off" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max";
 
 export type SystemPromptMode = "append" | "replace" | "replace-all";
 
@@ -19,7 +19,7 @@ export type PermissionAction = "allow" | "deny" | "ask";
 export type AgentSource = "global" | "project";
 
 export const THINKING_LEVELS: readonly ThinkingLevel[] = [
-  "off", "minimal", "low", "medium", "high", "xhigh",
+  "off", "minimal", "low", "medium", "high", "xhigh", "max",
 ] as const;
 
 export const SYSTEM_PROMPT_MODES: readonly SystemPromptMode[] = [