Open source Grok Bot alternative, Bring your own agent and subscription: message Claude Code, Codex, Gemini CLI, Cursor, Copilot and 40+ AI agents as bots on your own computer, from iPhone, Mac, Windows, Linux or terminal. Agent-agnostic and free. Also a Muse and Dots alternative. Built in Rust.
Memory evaluation
Scripted conversations that check what a bot remembers, run against a real host and a real harness (design).
cases.json: 52 zh-Hant and English cases. Extraction, multi-session reasoning, temporal reasoning, knowledge update and abstention (LongMemEval's five abilities); selective forgetting (MemoryAgentBench); detail that only a long-form note can hold; short CJK queries; and instructions planted in pasted text.run.py: one bot per case. It sends each session's messages, starts a new session (which runs dreaming), then asks the probe in a fresh session and grades the reply by keywords, plus what memory may not hold.
Run it on a disposable host, since it creates and deletes bots:
CODYNC_HOME=/tmp/eval-home host/target/debug/codync-host serve --port 19555
tools/memory-eval/run.py --home /tmp/eval-home --backend claude --jobs 3
tools/memory-eval/run.py --home /tmp/eval-home --only tm,lf-01 # some kinds or cases
It prints a pass count per kind and each failure with the reply; results.json keeps every reply, entry and note
name. Temporal answers are written as {d+3:%Y-%m-%d}, {next_friday:…} or {next_month:%B} and expanded on
the day of the run.
Prompt cache
cache.py reads Claude Code's and Codex's session logs (~/.claude/projects, ~/.codex/sessions) and prints,
per harness, how much of each conversation's input came from the prompt cache and why the cold calls were cold;
for Claude also the input cost against no cache. It needs no host:
tools/memory-eval/cache.py # the bots of ~/.codync and ~/.codync-dev
tools/memory-eval/cache.py eval-home-bots # workspaces whose path contains this ("/" and "." read as "-")
The latest numbers are under prompt cache.