From 4f24c7e6c520dff7aee46f80b5e2638e9d31efa5 Mon Sep 17 00:00:00 2001 From: Levi Neely Date: Sat, 18 Apr 2026 11:41:33 +0200 Subject: [PATCH] doc: add AI pipelines observation Co-Authored-By: Claude Sonnet 4.6 --- doc/OBSERVATIONS.md | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/doc/OBSERVATIONS.md b/doc/OBSERVATIONS.md index 9f1d3c1..9a8e585 100644 --- a/doc/OBSERVATIONS.md +++ b/doc/OBSERVATIONS.md @@ -19,3 +19,11 @@ **Passive memory** — Every session's chat history accumulates at `s/{id}/chat`, searchable with `grep`. An agent doesn't need to actively decide what to remember. This extends across sessions: an agent can search another session's chat for answers that have already been worked through. **Sub-agents** — `subagent_spawn` implements fork-join over the `b/` namespace: spawn N specialized agents with a shared prompt, wait for all to complete, collect results in submission order. For heterogeneous parallel delegation (N agents, N different prompts), use a `{parallel: [...]}` stage in `execute_code` — each parallel step calls `subagent_spawn` with a different prompt, all running concurrently. + +**AI pipelines with standard tools** — ollie scripts compose into multi-stage AI pipelines via Unix pipes. Proof-of-concept: + +``` +; s/optimize -n 2 -cm qwen/qwen3-8b -jm 'anthropic/claude-opus-4.6' 'generate a hello world program. use c++.' >[2]/dev/null | s/q +``` + +`s/optimize` generates and judges candidate prompts, emitting the best to stdout; `s/q` acts on it. Any tool that reads stdin and writes stdout is already a pipeline stage.