feat: inject ponytail ruleset into subagents via SubagentStart hook (#254)

* feat: inject ponytail ruleset into subagents via SubagentStart hook

SessionStart additionalContext is parent-thread only, so every Task-spawned
agent ran ponytail-unaware. Add a SubagentStart hook that injects the active
ruleset into each subagent, reusing getPonytailInstructions. Native Claude
needs the hookSpecificOutput JSON form (not raw stdout), so writeHookOutput
grows a SubagentStart branch; readMode exposes the live flag.

Workflow- and team-spawned coverage is undocumented upstream; verify in a
fresh session once installed.

Closes #252

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01H1dMag33yz1Kf1jS24Aira

* fix: address QA round 1 — make hook tests hermetic + cover Codex SubagentStart

QA round 1 (panel + deepseek-v4-pro) found two real test issues:
- The new subagent block (and the pre-existing claudeEnv block) used a no-op
  `delete env.PLUGIN_DATA`; run() spreads process.env, so a PLUGIN_DATA /
  COPILOT_PLUGIN_DATA leaked from the shell would steer writeHookOutput into the
  codex/copilot branch and silently mis-fire the native-Claude assertions.
  Fixed at the source: neutralize both vars once at the top, like CLAUDE_CONFIG_DIR.
- The Codex SubagentStart branch (claude-codex-hooks.json is shared by both plugin
  manifests) had zero coverage. Added a codex-path assertion.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01H1dMag33yz1Kf1jS24Aira

* fix: use PowerShell-safe command form for SubagentStart hook

Match the post-#265 'node ...; exit 0' form used by the sibling hooks. The old 'command -v node ... || exit 0' form fails tests/hooks-windows.test.js (POSIX-guard and non-blocking asserts) once this branch merges onto current main.

---------

Co-authored-by: Shane McCarron <shane.mccarron@corvexconnect.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Emeriko <dietrich.gebert@gmail.com>

hooks/claude-codex-hooks.json

@@ -14,6 +14,19 @@
         ]
       }
     ],
+    "SubagentStart": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "node \"${CLAUDE_PLUGIN_ROOT}/hooks/ponytail-subagent.js\"; exit 0",
+            "commandWindows": "if (Get-Command node -ErrorAction SilentlyContinue) { node \"$env:CLAUDE_PLUGIN_ROOT\\hooks\\ponytail-subagent.js\" }",
+            "timeout": 5,
+            "statusMessage": "Loading ponytail mode..."
+          }
+        ]
+      }
+    ],
     "UserPromptSubmit": [
       {
         "hooks": [

hooks/ponytail-runtime.js

@@ -21,6 +21,15 @@ function clearMode() {
   try { fs.unlinkSync(statePath); } catch (e) {}
 }
 
+// Live mode written by activate/mode-tracker. Absent flag = ponytail off.
+function readMode() {
+  try {
+    return fs.readFileSync(statePath, 'utf8').trim() || null;
+  } catch (e) {
+    return null;
+  }
+}
+
 function writeHookOutput(event, mode, context = '') {
   if (isCopilot) {
     // Copilot reads additionalContext on SessionStart; ignores output elsewhere.
@@ -39,6 +48,13 @@ function writeHookOutput(event, mode, context = '') {
     process.stdout.write(JSON.stringify(output));
     return;
   }
+  // Native Claude: SessionStart accepts raw stdout, but SubagentStart needs the
+  // hookSpecificOutput JSON form or the context is dropped.
+  if (event === 'SubagentStart') {
+    process.stdout.write(JSON.stringify(
+      { hookSpecificOutput: { hookEventName: event, additionalContext: context } }));
+    return;
+  }
   process.stdout.write(context);
 }
 
@@ -46,6 +62,7 @@ module.exports = {
   clearMode,
   isCodex,
   isCopilot,
+  readMode,
   setMode,
   writeHookOutput,
 };

hooks/ponytail-subagent.js

@@ -0,0 +1,22 @@
+#!/usr/bin/env node
+// ponytail — Claude Code SubagentStart hook
+//
+// SessionStart context is parent-thread only and never reaches subagents, so
+// without this every Task-spawned agent runs ponytail-unaware (issue #252).
+// When ponytail mode is active, inject the same ruleset into each subagent.
+
+const { getPonytailInstructions } = require('./ponytail-instructions');
+const { readMode, writeHookOutput } = require('./ponytail-runtime');
+
+const mode = readMode();
+
+// Absent flag or off → ponytail isn't active; inject nothing.
+if (!mode || mode === 'off') {
+  process.exit(0);
+}
+
+try {
+  writeHookOutput('SubagentStart', mode, getPonytailInstructions(mode));
+} catch (e) {
+  // Silent fail — a stdout error at hook exit must not surface as a hook failure.
+}

tests/hooks.test.js

@@ -26,9 +26,14 @@ function run(script, env, input = '') {
   });
 }
 
-// Keep the base env clean so the default-dir checks are deterministic; the
+// Keep the base env clean so the default-dir / native-Claude checks are
-// CLAUDE_CONFIG_DIR case sets it explicitly.
+// deterministic; the CLAUDE_CONFIG_DIR and codex/copilot cases set these
+// explicitly where needed. run() spreads process.env, so a PLUGIN_DATA /
+// COPILOT_PLUGIN_DATA leaked from the dev or CI shell would otherwise steer
+// writeHookOutput into the wrong branch and mis-fire the native assertions.
 delete process.env.CLAUDE_CONFIG_DIR;
+delete process.env.PLUGIN_DATA;
+delete process.env.COPILOT_PLUGIN_DATA;
 
 const temp = fs.mkdtempSync(path.join(os.tmpdir(), 'ponytail-hooks-'));
 // Runs on normal exit and on assertion-throw exit; force makes it idempotent.
@@ -168,4 +173,40 @@ assert.equal(
 output = JSON.parse(result.stdout);
 assert.deepEqual(output, {});
 
+// SubagentStart hook: when ponytail mode is active it injects the ruleset into
+// each subagent (issue #252). Native Claude must get the hookSpecificOutput JSON
+// form, not raw stdout, or the context is dropped.
+const subHome = path.join(temp, 'sub-home');
+const subFlag = path.join(subHome, '.claude', '.ponytail-active');
+fs.mkdirSync(path.dirname(subFlag), { recursive: true });
+const subEnv = { HOME: subHome, USERPROFILE: subHome };
+
+fs.writeFileSync(subFlag, 'full');
+result = run('ponytail-subagent.js', subEnv);
+assert.equal(result.status, 0, result.stderr);
+output = JSON.parse(result.stdout);
+assert.equal(output.hookSpecificOutput.hookEventName, 'SubagentStart');
+assert.match(
+  output.hookSpecificOutput.additionalContext,
+  /PONYTAIL MODE ACTIVE — level: full/,
+);
+
+// No flag → ponytail off → inject nothing (empty stdout, no failure).
+fs.unlinkSync(subFlag);
+result = run('ponytail-subagent.js', subEnv);
+assert.equal(result.status, 0, result.stderr);
+assert.equal(result.stdout, '', 'SubagentStart must stay silent when ponytail is off');
+
+// Codex shares claude-codex-hooks.json, so SubagentStart is reachable under Codex
+// too — assert the codex branch emits the badge plus hookSpecificOutput.
+const subCodex = path.join(temp, 'sub-codex');
+fs.mkdirSync(subCodex, { recursive: true });
+fs.writeFileSync(path.join(subCodex, '.ponytail-active'), 'full');
+result = run('ponytail-subagent.js', { HOME: subHome, USERPROFILE: subHome, PLUGIN_DATA: subCodex });
+assert.equal(result.status, 0, result.stderr);
+output = JSON.parse(result.stdout);
+assert.equal(output.systemMessage, 'PONYTAIL:FULL');
+assert.equal(output.hookSpecificOutput.hookEventName, 'SubagentStart');
+assert.match(output.hookSpecificOutput.additionalContext, /PONYTAIL MODE ACTIVE — level: full/);
+
 console.log('hook compatibility checks passed');