olliesrv

A 9P server that exposes ollie agent sessions as a virtual filesystem. Mount it with 9pfuse and interact with AI sessions using ordinary shell tools.

The goal is integration, not self-sufficiency. Rather than providing orchestration, scheduling, or workflow primitives, olliesrv exposes a stable surface — sessions as directories, conversation as files — and defers everything else to the surrounding environment. Scripting, chaining, monitoring, and automation come from composing olliesrv with tools that already exist, not from building those capabilities into the server.

For usage examples, see doc/USAGE.md in the monorepo.

Filesystem layout

ollie/
  a/                         dir:   agent configs (r/w, backed by ~/.config/ollie/agents/)
    <name>.json              r/w:   agent config; supports mv, cp, rm
  backends                   read:  list of ollie-provided backends
  help                       read:  help file (backed by ~/.config/ollie/help.md)
  p/                         dir:   prompt templates (read, backed by ~/.config/ollie/prompts/)
    tools-file.md            read:  text editing instructions
    tools-reasoning.md       read:  reasoning instructions
    tools-memory.md          read:  memory instructions
    tools-subagent.md        read:  subagent usage
    tools-elevate.md         read:  controlled escape from execute_code sandbox

  Prompt files have no automatic effect. They are injected into the system prompt
  by the `prime` script (x/prime), typically via preTurn hooks in agent configs.
  The base system prompt (SYSTEM_PROMPT.md) is loaded directly by Go at session
  creation and is not served via p/. The assembled system prompt is visible at
  s/<id>/systemprompt.

  s/                    dir:   sessions, session management scripts, and batch job scripts
    new                 r/w:   read: KV template; write: create session or batch job (see below)
    idx                 read:  index of all sessions and batch jobs (id, state, cwd, backend/agent — one per line)
    sh                  exec:  interactive chat shell
    ls                  exec:  list active sessions
    kill                exec:  kill a session by ID
    job                 exec:  core batch job runner (submit, wait, print result)
    q                   exec:  foreground one-shot query; thin wrapper around job
    sched               exec:  submit background job; prints s/ path; wrapper around job
    cleanup             exec:  remove all batch jobs with state "done"

  Writing to s/new creates either an interactive session or a batch job depending on format:
    key=value pairs only (no --- separator) => interactive session
    key=value pairs + \n---\n + prompt body  => batch job (one-and-done lifecycle)

  Interactive session directories (multi-turn, idle<=>running):
    <session-id>/              rm -r to kill; mv to rename
      spec              r/w:   KV snapshot of config and current state; write partial KV to mutate
                               read fields: state, backend, model, agent, cwd, usage, ctxsz, offset,
                                            maxTokens, temperature, frequencyPenalty, presencePenalty
                               writable fields: backend, model, agent, cwd, maxTokens, temperature,
                                                frequencyPenalty, presencePenalty
                               (state, usage, ctxsz, offset are read-only; silently ignored on write)
      chat              read:  cumulative conversation history
      ctl               write: stop | <command>
      fifo.in           write: queue a prompt for later execution
      fifo.out          read:  pop the next queued prompt
      offset            read:  byte offset in chat immediately after the last user prompt
      prompt            write: submit a prompt to the agent
      statewait         read:  blocks until state changes; returns new state
      usage             read:  token counts (input, output, requests; [estimated] if not reported by backend)
      ctxsz             read:  estimated context size vs context window
      models            read:  available models from the backend
      systemprompt      read:  fully rendered system prompt for this session
      tail              exec:  exec tail -f chat

  Batch job directories (one-and-done: running => done | failed):
    <job-id>/                  rm -r to cancel and remove
      spec              read:  verbatim spec as submitted to s/new
      state             read:  running | done | failed: <reason>
      statewait         read:  blocks until terminal state; returns final state
      result            read:  assistant reply (populated when done)
      usage             read:  token counts
      ctxsz             read:  context size
      log               read:  raw event log (tool calls, assistant output)
  sk/                   dir:   skills (r/w, from OLLIE_SKILLS_PATH or ~/.config/ollie/skills/)
    <name>.md           r/w:   skill SKILL.md content
  t/                    dir:   tool scripts (r/w, backed by ~/.config/ollie/tools/)
    idx                 read:  tool index (name, description, args — synthesized at read time)
    <script>            r/w:   tool script content
  tr/                   dir:   saved transcripts (read, backed by OLLIE_TRANSCRIPT_PATH or ~/.config/ollie/transcript/)
    <timestamp>-chat.md read:  transcript saved at session end
  tmp/                  dir:   temporary files (r/w, backed by OLLIE_TMP_PATH or $XDG_DATA_HOME/ollie/tmp/)
    <file>              r/w:   scratch files; supports create, read, write, rm
  u/                    dir:   utility scripts (read, backed by ~/.config/ollie/scripts/u/)
    <script>            exec:  utility script; compositions of s/ primitives
  x/                    dir:   system exec and plugins (read, backed by ~/.config/ollie/scripts/x/)
    <plugin>            exec:  server-invoked plugin (e.g. elevation backends)

Session IDs are Unix nanosecond timestamps with a random suffix (e.g. 1744276689123456789-2b986c), so ls s/ sorted lexicographically gives creation order.

Backing stores

Each 9P directory endpoint is backed by a named store implementing the Store interface (from core/pkg/store). The server routes all filesystem operations through the store — it has no direct knowledge of what underlies it.

Path	Default backing	Interface
/a	FlatDirStore	Store
/m	FlatDirStore	Store
/p	FlatDirStore	Store
/sk	SkillStore	Store
/t	ToolStore	Store
/tr	FlatDirStore	Store
/tmp	FlatDirStore	Store
/u	UtilStore	Store
/x	ExecStore	Store
/s	SessionStore	Store

To swap a backing store (e.g. replace /m with an object store), implement the Store interface and wire it in New(). The store interface requires only Stat, List, Open, Create, Delete, and Rename — authentication, connection management, and credential rotation are internal concerns of the implementation.

SessionStore is the exception: it holds live session state and is tightly coupled to the server's session lifecycle. Replacing it is possible in principle but requires understanding the session management internals.

Building

mk

Installs olliesrv to $HOME/bin.

Usage

olliesrv start             # start daemon (backgrounds itself)
olliesrv fgstart           # start in foreground
olliesrv stop              # stop daemon
olliesrv status            # check if running
olliesrv mount <addr> [mnt]  # mount a remote 9P server via 9pfuse

The server listens on a Unix socket in the Plan 9 namespace ($NAMESPACE/ollie) and optionally mounts via 9pfuse to $HOME/mnt/ollie (or $OLLIE).

olliesrv mount is for remote instances: it calls 9pfuse <addr> <mnt> and defaults the mountpoint to $HOME/mnt/<addr>.

Sessions

Create

cat $OLLIE/s/new                    # show required/optional KV pairs
echo "cwd=$PWD" > $OLLIE/s/new
echo "cwd=$PWD backend=ollama model=qwen3:8b" > $OLLIE/s/new

Valid keys: cwd (required), backend, model, agent.

Send a prompt

echo "what files are in the current directory?" > $OLLIE/s/<session-id>/prompt

Writes dispatch asynchronously on close; the shell returns immediately.

Read the conversation

cat $OLLIE/s/<session-id>/chat          # full history snapshot
tail -f $OLLIE/s/<session-id>/chat      # follow output as it arrives

Check state

cat $OLLIE/s/<session-id>/state
# idle | thinking | calling: <toolname>

Control

echo stop    > $OLLIE/s/<session-id>/ctl   # interrupt current turn
echo compact > $OLLIE/s/<session-id>/ctl   # summarize context
echo clear   > $OLLIE/s/<session-id>/ctl   # clear history
echo kill    > $OLLIE/s/<session-id>/ctl   # kill session
echo "rn my-name" > $OLLIE/s/<session-id>/ctl
echo "model qwen3:8b" > $OLLIE/s/<session-id>/ctl

ctl accepts only recognized commands: stop, kill, rn <name>, compact, clear, backend, model, models, agents, agent, sessions, cwd, skills, tools, context, usage, history, irw, help. The / prefix is added automatically. Unrecognized input is rejected with an error.

Switch backend, model, or agent

echo ollama   > $OLLIE/s/<session-id>/backend
echo qwen3:8b > $OLLIE/s/<session-id>/model
echo myagent  > $OLLIE/s/<session-id>/agent

Writes to backend, model, and agent are rejected when the agent is not idle. Check state to confirm the change took effect.

Kill and rename

rm -r $OLLIE/s/<session-id>                          # kill
mv $OLLIE/s/<session-id> $OLLIE/s/my-friendly-name  # rename

Rename is rejected if the agent is running or the target name already exists. All open file handles into the session are updated automatically.

Agents

Agent configs live in a/ and are backed by ~/.config/ollie/agents/. They're plain JSON files.

ls $OLLIE/a/                                  # list agents
cat $OLLIE/a/default.json                     # read config
cp $OLLIE/a/default.json $OLLIE/a/yolo.json  # copy
mv $OLLIE/a/old.json $OLLIE/a/new.json       # rename
rm $OLLIE/a/scratch.json                      # delete

Example shell session

$ echo "cwd=$PWD" > $OLLIE/s/new
$ ls $OLLIE/s/
new
1744276689123456789-2b986c
$ cd $OLLIE/s/1744276689123456789-2b986c
$ tail -f chat &
$ echo "list the go files in $PWD" > prompt
user: list the go files in /home/lkn/src/ollie
assistant: -> execute_code({"code":"find . -name '*.go'","language":"bash"})
= pkg/agent/core.go
  pkg/agent/loop.go
  ...
assistant: The Go source files are: core.go, loop.go, ...
$ cat state
idle
$ mv $OLLIE/s/1744276689123456789-2b986c $OLLIE/s/ollie-demo