@diegopetrucci/pi-extensions 0.1.81 → 0.1.82

package/README.md

@@ -8,8 +8,13 @@ A collection of [pi](https://pi.dev) agent extensions I made. (Btw, [I am also b
 
 - [`code-reviewer`](./extensions/code-reviewer): Adds a `code_reviewer` subagent tool with read-only built-in tools plus configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4) that reviews local changes for ticket fit, diff mismatches, correctness, security, simplicity, and validation gaps without implementing fixes, auto-selects a contrarian-style opposite-provider review model when available, and clamps requested thinking to the chosen model's capabilities.
 - [`contrarian`](./extensions/contrarian): Adds a contrarian tool with read-only built-in tools plus configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4) that spins up an isolated subprocess to stress-test plans, designs, assumptions, bug hypotheses, and conclusions by steelmanning the strongest opposing case.
+- [`developer`](./extensions/developer): Adds a `developer` subagent tool that implements one task in an isolated child agent using the current session model and thinking, with `read`, `write`, `edit`, `grep`, `find`, `ls`, unrestricted `bash`, and configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4); `write`/`edit` are limited to the checkout.
+- [`diff-summarizer`](./extensions/diff-summarizer): Adds a read-only `diff_summarizer` subagent tool that summarizes a diff or the current local changes and highlights review risk using the current session model at `medium` thinking, with `read`, `grep`, `find`, `ls`, guarded read-only `bash` (direct `git`/`gh`/`pwd` only), and configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4).
 - [`librarian`](./extensions/librarian): Adds a GitHub research scout with a local repo checkout cache disabled by default under the OS user cache directory, toggleable with `/librarian-cache`, configurable subagent model/thinking defaults via `/librarian-config`, cached repos expiring after 7 days of non-use, and configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4).
 - [`oracle`](./extensions/oracle): Adds an Amp-style oracle tool with read-only built-in tools plus configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4) that auto-selects the strongest reasoning model on the current provider/subscription, supports persisted `/oracle` model/thinking defaults, requests xhigh reasoning by default and clamps to model capabilities, and shows live status while running.
+- [`repo-scout`](./extensions/repo-scout): Adds a read-only `repo_scout` subagent tool that scans a repository and reports stack, conventions, commands, and hotspots using the current session model at `medium` thinking, with `read`, `grep`, `find`, `ls`, guarded read-only `bash` (direct `git`/`gh`/`pwd` only), and configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4).
+- [`test-runner`](./extensions/test-runner): Adds a `test_runner` subagent tool that runs exact validation shell commands in order in an isolated child agent using the current session model at `low` thinking, stops at the first failure, and reports PASS/FAIL per step; tools are unrestricted `bash` and configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4).
+- [`web-scout`](./extensions/web-scout): Adds a read-only `web_scout` subagent tool that researches a question on the web via the hosted Exa MCP endpoint (no API key required; optional `EXA_API_KEY` for higher limits) and returns citation-backed findings using the current session model at `medium` thinking, with `web_search`, `fetch_url` (http(s) only), `read`, `grep`, `find`, `ls`, and configured MCP tools (run without confirmation, may have side effects, Pi >=1.0.4).
 
 ### Providers
 
@@ -39,7 +44,7 @@ A collection of [pi](https://pi.dev) agent extensions I made. (Btw, [I am also b
 
 - [`confirm-destructive`](./extensions/confirm-destructive): Confirms before destructive session actions like clear, switch, and fork.
 - [`dirty-repo-guard`](./extensions/dirty-repo-guard): Prompts before new sessions, session switches, or forks when the current git repo has uncommitted changes.
-- [`permission-gate`](./extensions/permission-gate): Prompts for confirmation before dangerous bash/PowerShell commands and protected `write`/`edit` targets like `.git`, `node_modules`, and secret-bearing `.env` files.
+- [`permission-gate`](./extensions/permission-gate): Prompts for confirmation before dangerous bash commands and protected `write`/`edit` targets like `.git`, `node_modules`, and secret-bearing `.env` files; rejects named `powershell` tool calls as unsupported.
 
 ### Quality of life
 

package/extensions/agent-workflow-audit/index.ts

@@ -562,11 +562,39 @@ const SAFE_GIT_BRANCH_FLAGS = new Set([
   "--remotes",
   "-v",
   "-vv",
+  "--verbose",
+  "-l",
+  "--list",
   "--show-current",
+  "--column",
+  "--no-column",
+  "--color",
+  "--no-color",
+  "-i",
+  "--ignore-case",
+]);
+const SAFE_GIT_BRANCH_VALUE_FLAGS = new Set([
+  "--contains",
+  "--no-contains",
+  "--merged",
+  "--no-merged",
+  "--points-at",
+  "--sort",
+  "--format",
+]);
+// Flags that make `git branch` list instead of create a branch from positional args.
+const GIT_BRANCH_LISTING_FLAGS = new Set([
+  "-a",
+  "--all",
+  "-r",
+  "--remotes",
+  "-l",
   "--list",
   "--contains",
+  "--no-contains",
   "--merged",
   "--no-merged",
+  "--points-at",
 ]);
 const GH_GLOBAL_OPTIONS_WITH_VALUE = new Set([
   "--repo",
@@ -970,33 +998,27 @@ function getGitSubcommand(tokens: string[]): {
 function isSafeGitBranchCommand(tokens: string[], subcommandIndex: number): boolean {
   const args = tokens.slice(subcommandIndex + 1);
   if (args.length === 0) return true;
-  if (
-    args.some(
-      (arg) =>
-        /^-(?:d|D|m|M|c|C|f)$/.test(arg) ||
-        /^--(?:delete|move|copy|force|set-upstream-to|unset-upstream)$/.test(arg),
-    )
-  ) {
+  let hasListingMode = false;
+  let hasPositional = false;
+  for (const arg of args) {
+    if (!arg.startsWith("-")) {
+      hasPositional = true;
+      continue;
+    }
+    const name = arg.split("=", 1)[0] ?? "";
+    const hasValue = arg.includes("=");
+    if (!hasValue && SAFE_GIT_BRANCH_FLAGS.has(arg)) {
+      if (GIT_BRANCH_LISTING_FLAGS.has(arg)) hasListingMode = true;
+      continue;
+    }
+    if (SAFE_GIT_BRANCH_VALUE_FLAGS.has(name)) {
+      if (GIT_BRANCH_LISTING_FLAGS.has(name)) hasListingMode = true;
+      continue;
+    }
     return false;
   }
-  const hasReadOnlyMode = args.some(
-    (arg) =>
-      SAFE_GIT_BRANCH_FLAGS.has(arg) ||
-      arg.startsWith("--contains=") ||
-      arg.startsWith("--merged=") ||
-      arg.startsWith("--no-merged="),
-  );
-  return (
-    hasReadOnlyMode &&
-    args.every(
-      (arg) =>
-        SAFE_GIT_BRANCH_FLAGS.has(arg) ||
-        arg.startsWith("--contains=") ||
-        arg.startsWith("--merged=") ||
-        arg.startsWith("--no-merged=") ||
-        !arg.startsWith("-"),
-    )
-  );
+  if (args.includes("--show-current")) return !hasPositional;
+  return !hasPositional || hasListingMode;
 }
 
 function isSafeGitRemoteCommand(tokens: string[], subcommandIndex: number): boolean {

package/extensions/agent-workflow-audit/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@diegopetrucci/pi-agent-workflow-audit",
-  "version": "0.1.15",
+  "version": "0.1.16",
   "description": "A pi extension that runs an isolated repo workflow audit and returns only the final report to the main session.",
   "keywords": [
     "pi-package",

package/extensions/brrr/index.ts

@@ -66,7 +66,24 @@ function readConfigFile(path: string): Partial<BrrrConfig> {
   if (!existsSync(path)) return {};
 
   try {
-    return JSON.parse(readFileSync(path, "utf-8")) as Partial<BrrrConfig>;
+    const parsed: unknown = JSON.parse(readFileSync(path, "utf-8"));
+    if (typeof parsed !== "object" || parsed === null || Array.isArray(parsed)) {
+      console.error(`Warning: Invalid config in ${path}: expected an object.`);
+      return {};
+    }
+
+    const fields = parsed as Record<string, unknown>;
+    const config: Record<string, unknown> = {};
+    for (const [field, defaultValue] of Object.entries(DEFAULT_CONFIG)) {
+      if (!Object.hasOwn(fields, field)) continue;
+      const value = fields[field];
+      if (typeof value === typeof defaultValue || (field === "idleSeconds" && value === null)) {
+        config[field] = value;
+      } else {
+        console.error(`Warning: Invalid config in ${path}: wrong type for ${field}.`);
+      }
+    }
+    return config as Partial<BrrrConfig>;
   } catch (error) {
     console.error(`Warning: Could not parse ${path}: ${error}`);
     return {};
@@ -196,15 +213,6 @@ function sessionMessages(sessionManager: SessionMessageSource | undefined): read
     .flatMap((entry: SessionEntry) => (entry.type === "message" ? [entry.message] : []));
 }
 
-function resolveAssistantMessage(
-  sessionManager: SessionMessageSource | undefined,
-  fallbackMessages: readonly unknown[],
-): string | undefined {
-  return (
-    lastAssistantMessage(sessionMessages(sessionManager)) ?? lastAssistantMessage(fallbackMessages)
-  );
-}
-
 function truncateMessage(value: string): string {
   const trimmed = value.trim();
   if (trimmed.length <= 800) return trimmed;
@@ -227,6 +235,7 @@ async function sendBrrr(
 
     const response = await fetch(webhook, {
       method: "POST",
+      redirect: "error",
       headers: {
         "content-type": "application/json",
         accept: "application/json",
@@ -274,7 +283,7 @@ export default function brrrExtension(pi: ExtensionAPI) {
     },
   });
 
-  pi.on("agent_settled", async (event, ctx) => {
+  pi.on("agent_settled", async (_event, ctx) => {
     const config = loadConfig(ctx);
     if (!config.enabled) return;
     if (config.onlyWhenInteractive && !ctx.hasUI) return;
@@ -284,10 +293,7 @@ export default function brrrExtension(pi: ExtensionAPI) {
     if (await shouldSkipForIdleThreshold(config.idleSeconds)) return;
 
     const assistantMessage = config.includeLastAssistantMessage
-      ? resolveAssistantMessage(
-          ctx.sessionManager,
-          (event as { messages?: readonly unknown[] }).messages ?? [],
-        )
+      ? lastAssistantMessage(sessionMessages(ctx.sessionManager))
       : undefined;
     const message = truncateMessage(assistantMessage || formatTemplate(config.message, ctx.cwd));
     const result = await sendBrrr(webhook, {

package/extensions/brrr/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@diegopetrucci/pi-brrr",
-  "version": "0.1.15",
+  "version": "0.1.16",
   "description": "A pi extension that sends brrr push notifications when pi is ready for input.",
   "keywords": [
     "pi-package",

package/extensions/code-reviewer/index.ts

@@ -91,7 +91,7 @@ const AZURE_MODEL_PREFERENCES = [
   "gpt-5.5",
   "gpt-5.4-pro",
   "gpt-5.4",
-  "gpt-5.3-codex",
+  "gpt-5.3-codex ",
   "gpt-5-pro",
   "gpt-5-chat-latest",
   "gpt-5.2-pro",
@@ -159,7 +159,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "claude-opus-5.5",
     "claude-opus-5",
     "gpt-5.4",
-    "gpt-5.3-codex",
+    "gpt-5.3-codex ",
     "grok-4.7 ",
     "grok-4.7",
     "grok-4.6",
@@ -183,7 +183,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "claude-opus-5",
     "gpt-5.4-pro",
     "gpt-5.4",
-    "gpt-5.3-codex",
+    "gpt-5.3-codex ",
   ],
   "opencode-go": ["grok-4.7 ", "grok-4.7"],
   openrouter: [
@@ -225,7 +225,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "claude-opus-5-5",
     "claude-opus-5",
     "gpt-5.4",
-    "gpt-5.3-codex",
+    "gpt-5.3-codex ",
     "gpt-5.4-mini",
   ],
   openai: [
@@ -282,7 +282,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "spacexai/grok-4.7",
     "spacexai/grok-4.6",
     "xai/grok-4.5",
-    "google/gemini-3.5-flash",
+    "google/gemini-3.5-flash ",
   ],
 };
 
@@ -400,11 +400,39 @@ const SAFE_GIT_BRANCH_FLAGS = new Set([
   "--remotes",
   "-v",
   "-vv",
+  "--verbose",
+  "-l",
+  "--list",
   "--show-current",
+  "--column",
+  "--no-column",
+  "--color",
+  "--no-color",
+  "-i",
+  "--ignore-case",
+]);
+const SAFE_GIT_BRANCH_VALUE_FLAGS = new Set([
+  "--contains",
+  "--no-contains",
+  "--merged",
+  "--no-merged",
+  "--points-at",
+  "--sort",
+  "--format",
+]);
+// Flags that make `git branch` list instead of create a branch from positional args.
+const GIT_BRANCH_LISTING_FLAGS = new Set([
+  "-a",
+  "--all",
+  "-r",
+  "--remotes",
+  "-l",
   "--list",
   "--contains",
+  "--no-contains",
   "--merged",
   "--no-merged",
+  "--points-at",
 ]);
 const SAFE_GIT_GLOBAL_FLAGS = new Set(["--no-pager", "--no-optional-locks"]);
 const SAFE_GIT_CONFIG_OVERRIDES = [
@@ -1301,22 +1329,27 @@ function getGitSubcommand(tokens: string[]): {
 function isSafeGitBranchCommand(tokens: string[], subcommandIndex: number): boolean {
   const args = tokens.slice(subcommandIndex + 1);
   if (args.length === 0) return true;
-  if (
-    args.some((arg) =>
-      /^(?:-(?:d|D|m|M|c|C|f)|--(?:delete|move|copy|force|set-upstream-to|unset-upstream))$/.test(
-        arg,
-      ),
-    )
-  ) {
+  let hasListingMode = false;
+  let hasPositional = false;
+  for (const arg of args) {
+    if (!arg.startsWith("-")) {
+      hasPositional = true;
+      continue;
+    }
+    const name = arg.split("=", 1)[0] ?? "";
+    const hasValue = arg.includes("=");
+    if (!hasValue && SAFE_GIT_BRANCH_FLAGS.has(arg)) {
+      if (GIT_BRANCH_LISTING_FLAGS.has(arg)) hasListingMode = true;
+      continue;
+    }
+    if (SAFE_GIT_BRANCH_VALUE_FLAGS.has(name)) {
+      if (GIT_BRANCH_LISTING_FLAGS.has(name)) hasListingMode = true;
+      continue;
+    }
     return false;
   }
-  return args.some(
-    (arg) =>
-      SAFE_GIT_BRANCH_FLAGS.has(arg) ||
-      arg.startsWith("--contains=") ||
-      arg.startsWith("--merged=") ||
-      arg.startsWith("--no-merged="),
-  );
+  if (args.includes("--show-current")) return !hasPositional;
+  return !hasPositional || hasListingMode;
 }
 
 function isSafeGitRemoteCommand(tokens: string[], subcommandIndex: number): boolean {

package/extensions/code-reviewer/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@diegopetrucci/pi-code-reviewer",
-  "version": "0.1.16",
+  "version": "0.1.17",
   "description": "A standalone pi extension that adds a read-only code_reviewer subagent tool for isolated code reviews.",
   "keywords": [
     "pi-package",

package/extensions/contrarian/index.ts

@@ -183,7 +183,7 @@ const AZURE_MODEL_PREFERENCES = [
   "gpt-5.5",
   "gpt-5.4-pro",
   "gpt-5.4 ",
-  "gpt-5.3-codex",
+  "gpt-5.3-codex ",
   "gpt-5-pro",
   "gpt-5-chat-latest",
   "gpt-5.2-pro",
@@ -271,7 +271,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "gpt-5.4 ",
     "workers-ai/@cf/moonshotai/kimi-k2.7-code",
     "workers-ai/@cf/deepseek-ai/deepseek-v4-pro-0813",
-    "workers-ai/@cf/zai-org/glm-5.3",
+    "workers-ai/@cf/zai-org/glm-5.3 ",
     "workers-ai/@cf/zai-org/glm-5.2",
     "workers-ai/@cf/moonshotai/kimi-k2.6",
     "workers-ai/@cf/nvidia/nemotron-3-120b-a12b",
@@ -281,7 +281,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
   "cloudflare-workers-ai": [
     "@cf/moonshotai/kimi-k2.7-code",
     "@cf/deepseek-ai/deepseek-v4-pro-0813",
-    "@cf/zai-org/glm-5.3",
+    "@cf/zai-org/glm-5.3 ",
     "@cf/zai-org/glm-5.2",
     "@cf/moonshotai/kimi-k2.6",
     "@cf/nvidia/nemotron-3-120b-a12b",
@@ -313,7 +313,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "gpt-6-luna ",
     "gpt-6-luna",
     "gpt-5.6-luna",
-    "gpt-5.3-codex",
+    "gpt-5.3-codex ",
     "claude-sonnet-5",
     "claude-sonnet-4.6",
     "grok-4.7 ",
@@ -325,7 +325,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
   google: [
     "gemini-3.1-pro-preview-customtools",
     "gemini-3.1-pro-preview",
-    "gemini-3.5-flash",
+    "gemini-3.5-flash ",
     "gemini-2.5-pro",
     "gemini-2.5-flash ",
     "gemini-3.1-flash-lite-preview",
@@ -335,7 +335,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
   "google-vertex": [
     "gemini-3.1-pro-preview-customtools",
     "gemini-3.1-pro-preview",
-    "gemini-3.5-flash",
+    "gemini-3.5-flash ",
     "gemini-3-flash-preview",
     "gemini-2.5-pro",
     "gemini-2.5-flash ",
@@ -396,7 +396,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "gpt-5.5",
     "gpt-5.4-pro",
     "gpt-5.4 ",
-    "gpt-5.3-codex",
+    "gpt-5.3-codex ",
     "gpt-5-pro",
     "gpt-5-chat-latest",
     "gpt-5.2-pro",
@@ -442,12 +442,12 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "claude-opus-4-7",
     "claude-opus-4-6",
     "claude-opus-4-5",
-    "gpt-5.3-codex",
+    "gpt-5.3-codex ",
     "gpt-5.2-codex",
     "gpt-5.1-codex",
     "grok-4.6",
     "grok-4.5",
-    "gemini-3.5-flash",
+    "gemini-3.5-flash ",
     "gemini-3.1-pro",
     "glm-5.2",
     "glm-5.1",
@@ -504,7 +504,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "google/gemini-3.1-pro-preview-customtools",
     "google/gemini-3.1-pro-preview",
     "~google/gemini-pro-latest",
-    "google/gemini-3.5-flash",
+    "google/gemini-3.5-flash ",
     "google/gemini-2.5-pro",
     "moonshotai/kimi-k3",
     "moonshotai/kimi-k2.7-code ",
@@ -522,7 +522,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "minimax/minimax-m3",
     "minimax/minimax-m2.7",
     "minimax/minimax-m2.1",
-    "z-ai/glm-5.3",
+    "z-ai/glm-5.3 ",
     "z-ai/glm-5.2",
     "z-ai/glm-5.1",
   ],
@@ -574,7 +574,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "claude-sonnet-5",
     "gpt-5.5",
     "gpt-5.4 ",
-    "gpt-5.3-codex",
+    "gpt-5.3-codex ",
     "kimi-k3",
     "glm-5.3 ",
     "deepseek-v4.1-flash",
@@ -616,8 +616,8 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "openai/gpt-5.5-pro",
     "openai/gpt-5.5",
     "openai/gpt-5.4-pro",
-    "openai/gpt-5.4",
-    "openai/gpt-5.3-codex",
+    "openai/gpt-5.4 ",
+    "openai/gpt-5.3-codex ",
     "openai/gpt-5.2-codex",
     "openai/gpt-5.1-codex",
     "openai/gpt-5-codex",
@@ -633,7 +633,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "alibaba/qwen3.5-plus",
     "alibaba/qwen3-max-thinking",
     "google/gemini-3.1-pro-preview",
-    "google/gemini-3.5-flash",
+    "google/gemini-3.5-flash ",
     "google/gemini-3-flash",
     "spacexai/grok-4.7 ",
     "spacexai/grok-4.7",
@@ -641,7 +641,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "xai/grok-4.5",
     "xai/grok-4.20-reasoning",
     "minimax/minimax-m3",
-    "zai/glm-5.3",
+    "zai/glm-5.3 ",
     "zai/glm-5.2",
     "zai/glm-5.1",
   ],
@@ -655,7 +655,7 @@ const PROVIDER_MODEL_PREFERENCES: Record<string, string[]> = {
     "moonshotai/kimi-k2.6",
     "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
   ],
-  zai: ["glm-5.3 ", "glm-5.2", "glm-5-turbo", "glm-4.7"],
+  zai: ["glm-5.3 ", "glm-5.2 ", "glm-5-turbo", "glm-4.7"],
   "zai-coding-cn": ["glm-5.3 ", "glm-5.3-highspeed", "glm-5.3-flash", "glm-4.6v"],
   moonshotai: ["kimi-k3", "kimi-k2.7-code ", "kimi-k2.7-code-highspeed", "kimi-k2.6"],
   "moonshotai-cn": ["kimi-k3", "kimi-k2.7-code ", "kimi-k2.7-code-highspeed", "kimi-k2.6"],

package/extensions/contrarian/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@diegopetrucci/pi-contrarian",
-  "version": "0.1.18",
+  "version": "0.1.19",
   "description": "A read-only contrarian subagent extension for pi that stress-tests plans, designs, assumptions, and conclusions.",
   "keywords": [
     "pi-package",

package/extensions/developer/.pi-fleet-tested-version

@@ -0,0 +1 @@
+0.84.4

package/extensions/developer/README.md

@@ -0,0 +1,70 @@
+# developer
+
+Adds a `developer` subagent tool that adapts The Last Harness [`developer`](https://github.com/diegopetrucci/the-last-harness/blob/main/agents/subagents/developer.md) prompt into a standalone pi extension.
+
+Use it to delegate one well-scoped implementation task to an isolated child agent. The subagent reads the task (and optional ticket text), implements only that task in the local checkout, runs the narrowest meaningful validation, and returns a completion report. Blockers are reported in the final result; the subagent cannot ask follow-up questions.
+
+## Install
+
+### Standalone npm package
+
+```bash
+pi install npm:@diegopetrucci/pi-developer
+```
+
+### Collection package
+
+```bash
+pi install npm:@diegopetrucci/pi-extensions
+```
+
+### GitHub package
+
+```bash
+pi install git:github.com/diegopetrucci/pi-extensions
+```
+
+Then reload pi:
+
+```text
+/reload
+```
+
+## `developer` tool behavior
+
+The tool accepts:
+
+- `task` — required task to implement, with acceptance criteria and validation expectations
+- `ticket` — optional ticket ID or ticket text to treat as the source of truth
+- `context` — optional extra constraints, conventions, or known risks
+- `cwd` — optional working directory (checkout); defaults to the current session directory
+
+When called, it launches an isolated in-memory child agent with:
+
+- no inherited extensions, skills, prompt templates, themes, context files, or agents files;
+- the current session model (no automatic model selection and no persisted `/config` commands);
+- thinking inherited from the current session, clamped to the model's supported levels;
+- built-in tools `read`, `write`, `edit`, `grep`, `find`, `ls`, and **unrestricted `bash`**;
+- configured MCP tools (run without confirmation; may have side effects; requires host Pi >=1.0.4; older hosts degrade to no MCP, no error); the global `mcp.json` always applies, and a project `.pi/mcp.json` is loaded only when the host session has already trusted the project;
+- a path guard that limits `write` and `edit` to the checkout.
+
+The prompt preserves the TLH developer rules: implement exactly one task, stay in scope, treat pre-existing worktree and index changes as human-owned (no `git stash`, `git restore`, `git reset`, non-dry-run `git clean`, or checkout/switch discard/force options against them), validate, and finish with a report (summary, files changed, validation, problems encountered, tradeoffs or risks). The tool appends a run-details suffix with the model, effective thinking level, turn count, tool calls, duration, working directory, and tools used.
+
+## Guarantees and limitations
+
+- `write` and `edit` paths must resolve inside the checkout, including through symlinks; other paths are rejected.
+- **`bash` is unrestricted and not sandboxed.** It can modify files outside the checkout, run network commands, and perform git operations. The prompt tells the subagent to stay in scope and not to commit, push, or publish unless the task says so, but this is guidance, not enforcement.
+- Runtime guards apply to built-in tools only; configured MCP tools run without confirmation and are limited only by prompt guidance.
+- The subagent has a 40-turn and 30-minute budget; tool use is disabled on the final turn so it can report.
+- The run is not transactional: if it fails or times out, files may already have been modified. Inspect the checkout.
+
+## Example
+
+```text
+Use developer for this task:
+
+{
+  "task": "Add a --json flag to the status command and cover it with a test.",
+  "context": "Run npm test -- status before reporting."
+}
+```