# pi ↔ MemPalace extension

The canonical source of `~/.pi/agent/extensions/mempalace.ts` — the bridge
that wires the [MemPalace](https://github.com/MemPalace/mempalace) MCP
server into the [pi coding-agent](https://github.com/mariozechner/pi-coding-agent)
harness.

`install.sh` at the repo root symlinks `mempalace.ts` from this directory
into `~/.pi/agent/extensions/` so the live file on every machine tracks
version control. Works on macOS and Linux (the extension itself is plain
Node / TypeScript; the symlink is a POSIX `ln -s`).

---

## What it does

1. **Spawns `mempalace-mcp`** as a subprocess and does the MCP stdio
   JSON-RPC handshake (`initialize` + `notifications/initialized` +
   `tools/list`).
2. **Registers each MCP tool** as a pi tool with its real `inputSchema`
   passed through via `Type.Unsafe(...)` (see gotcha below).
3. **Wake-up auto-injection** (`before_agent_start`, one-shot per fresh
   session): calls `mempalace_status` + `mempalace_diary_read` and
   injects the result as a `mempalace-wakeup` system message so the
   agent orients itself the way `~/.agents/skills/mempalace/SKILL.md`
   describes. Skipped on resume/fork (context is already in the thread).
4. **Manual wind-down** via a `/mempalace-diary [topic]` slash command:
   sends a prompt asking the LLM to call `mempalace_diary_write` with
   an AAAK-formatted entry summarizing the session. Not fully auto
   because pi sessions are typically short/tactical and
   `session_shutdown` fires too late to drive another LLM turn.

## Fail-soft

If `mempalace-mcp` can't be spawned (PATH missing, binary crashes at
startup, …) the extension logs to stderr and returns early. pi keeps
working without palace tools rather than refusing to start.

## Identity

`agent_name` for diary calls comes from `$MEMPALACE_AGENT_NAME`, defaulting
to `"pi"`. First diary write against that identity creates `wing_<name>`
in the palace. Set the env var if you want to run pi under a distinct
identity on a given machine (e.g. `pi-laptop` vs `pi-server`).

## Debugging

- `MEMPALACE_EXT_DEBUG=1` — surface `mempalace-mcp` stderr into pi's
  stderr. Without this, stderr is drained silently so a misbehaving
  server doesn't flood the TUI.
- If a tool call fails with a generic "Internal tool error", spawn
  `mempalace-mcp` manually with raw JSON-RPC on stdin to read the
  server-side error — much faster than guessing.

## The `Type.Unsafe` gotcha

Earlier versions of this extension registered every MCP tool with
`parameters: Type.Object({}, { additionalProperties: true })`, which
discarded each tool's real `inputSchema`. The LLM then saw no parameter
names and had to guess, leading to bugs like `mempalace_diary_read`
being called with `agent=` instead of the required `agent_name=` and
crashing the Python server with `TypeError: missing 1 required
positional argument`.

The fix (≈ lines 160-170) is to wrap the incoming JSON Schema with
`Type.Unsafe<...>(tool.inputSchema)`. TypeBox schemas are plain JSON
Schema at runtime plus a `Symbol` marker, so wrapping an
externally-sourced schema with `Unsafe` is sufficient — no conversion
to a full TypeBox tree is needed, and the LLM now sees every tool's
real parameter names.

If you ever need to re-loosen the schema for debugging, fall back to
the `Type.Object({}, { additionalProperties: true })` default only for
that specific tool, not globally.

## Keybindings (mosh/tmux newline fix)

`keybindings.json` is symlinked so edits flow through git. Default:

```json
{
  "tui.input.newLine": ["shift+enter", "ctrl+j", "alt+j"]
}
```

Rationale: when pi runs over `kitty → mosh → tmux`, shift+enter doesn't
forward cleanly (mosh uses vt220-ish emulation, no kitty-keyboard-protocol
or csi-u extended keys). `ctrl+j` and `alt+j` pass through as plain
control/meta bytes and give you reliable newline insertion.

## Settings template (start pi without `--model`)

`settings.example.json` is a template — **not symlinked**. pi rewrites
its `settings.json` at runtime (`lastChangelogVersion` bumps on upgrade),
which would dirty a symlinked repo file. Instead, bootstrap with:

```bash
cp /path/to/mempalace-toolkit/extensions/pi/settings.example.json \
   ~/.pi/agent/settings.json
$EDITOR ~/.pi/agent/settings.json
```

The Bedrock inference-profile prefix on model IDs (`eu.`, `us.`) is
**region-specific** and must match `AWS_REGION` in `~/.config/pi/.env`.
For a bare Anthropic provider (non-Bedrock) drop the prefix entirely
and use `anthropic:claude-...`. Run `pi --list-models` to confirm what
your credentials can actually invoke.

`install.sh` warns (non-fatal) if `settings.json` is missing.

## Environment setup

pi with `defaultProvider=amazon-bedrock` needs `AWS_PROFILE` and
`AWS_REGION` exported into the shell that launches it. Recommended
layout (matches the tor-ms22 dotfiles pattern):

```
~/.config/pi/.env               ← AWS_PROFILE=..., AWS_REGION=...
                                  (git-crypt encrypted in dotfiles repo)
~/.oh-my-zsh/custom/pi-env.zsh  ← set -a; source ~/.config/pi/.env; set +a
```

Historical note: these vars used to live under a `# Environment variables
for pi` block inside `~/.config/opencode/.env`. Split out 2026-05-05 so
each tool owns its own env file. `install.sh` runs a `check_aws_env`
probe that warns if the vars are missing and points back here.

## File layout

```
mempalace-toolkit/
└── extensions/
    └── pi/
        ├── README.md               ← this file
        ├── mempalace.ts            ← symlinked into ~/.pi/agent/extensions/
        ├── keybindings.json        ← symlinked into ~/.pi/agent/
        └── settings.example.json   ← template; copy + edit into ~/.pi/agent/
```

`install.sh` detects pi by probing for `~/.pi/agent/extensions/` and
only creates symlinks when that directory exists. On machines without
pi the files stay dormant in the repo. Re-runs are idempotent (same
pattern as `bin/` and `SKILL.md`).