doc: fix stale session management references

Update s/new syntax (no 'new' prefix), rm -r for kill, and correct
Python script to use s/new and shutil.rmtree instead of top-level ctl.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

doc/IDEAS.md

@@ -38,9 +38,9 @@ cd ~/mnt/ollie
 # Count existing sessions so we can identify the newly created ones by offset.
 n=$(ls s/ | wc -l)
 
-echo "new agent=developer cwd=$cwd" > ctl
-echo "new agent=reviewer  cwd=$cwd" > ctl
-echo "new agent=tester    cwd=$cwd" > ctl
+echo "agent=developer cwd=$cwd" > s/new
+echo "agent=reviewer  cwd=$cwd" > s/new
+echo "agent=tester    cwd=$cwd" > s/new
 
 dev=$(ls s/ | sort | sed -n "$((n+1))p")
 rev=$(ls s/ | sort | sed -n "$((n+2))p")
@@ -89,7 +89,7 @@ Each agent only sees what the script explicitly sends it. Each can use a differe
 
 ### Sub-Agents
 
-A session can spawn ephemeral child sessions to handle a focused task, then tear them down when done. The parent uses `execute_code` to write `new` to `ctl`, send an initial context and task to the child's `prompt`, poll until `reply` is ready, read the result, and write `kill <id>` to `ctl`. From the child's perspective it is just a normal session; it has no knowledge of its own ephemerality.
+A session can spawn ephemeral child sessions to handle a focused task, then tear them down when done. The parent uses `execute_code` to write a key=value spec to `s/new`, send an initial context and task to the child's `prompt`, poll until `reply` is ready, read the result, and `rm -r s/{id}` to clean up. From the child's perspective it is just a normal session; it has no knowledge of its own ephemerality.
 
 **Sub-agents as function calls** — the parent primes the child with exactly the context it needs (excerpts from its own `chat`, relevant beads, local file contents) and receives a single focused reply. Failures are isolated; a child that errors or stalls can be killed and retried without affecting the parent. The pattern composes naturally: a sub-agent can itself spawn sub-agents, building a call tree bounded only by the number of sessions open at once.
 
@@ -105,7 +105,7 @@ Output (stdout): JSON array of {"agent": str, "reply": str}
 Install in ~/.config/ollie/tools/ to make available as a {tool} step.
 """
 
-import json, os, sys, threading, time
+import json, os, shutil, sys, threading, time
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from pathlib import Path
 
@@ -120,7 +120,7 @@ def session_ids():
 def spawn_session(agent="default", cwd=""):
     with _spawn_lock:
         before = session_ids()
-        (BASE / "ctl").write_text(f"new agent={agent} cwd={cwd}\n")
+        (BASE / "s" / "new").write_text(f"agent={agent} cwd={cwd}\n")
         while True:
             new = session_ids() - before
             if new:
@@ -128,7 +128,7 @@ def spawn_session(agent="default", cwd=""):
             time.sleep(0.1)
 
 def kill_session(sid):
-    (BASE / "ctl").write_text(f"kill {sid}\n")
+    shutil.rmtree(BASE / "s" / sid)
 
 def wait_reply(sid):
     path = BASE / "s" / sid / "reply"

prompts/01_ollie.md

@@ -52,7 +52,7 @@ Use this to discover other sessions for cooperation. Sessions sharing your `cwd`
 
 **Session lifecycle:**
 - Create: write a key=value string to `{{.OllieMountPath}}/s/new`
-- Kill: write `kill` to `{{.OllieMountPath}}/s/{id}/ctl` (do not use `rm -r`)
+- Kill: `rm -r {{.OllieMountPath}}/s/{id}`
 - Rename: `mv {{.OllieMountPath}}/s/{id} {{.OllieMountPath}}/s/{newname}` — this also updates `$OLLIE_SESSION_ID` in the session environment
 
 **`ctl` commands** (write to `{{.OllieMountPath}}/s/{id}/ctl`):