PyAutoMind stores intent. PyAutoBrain reasons over that intent and routes it to the right specialist agent. This file defines the contract between the two: the first folder of every prompt path declares the kind of work, and PyAutoBrain maps that to a reasoning agent.
PyAutoMind organises intent by the kind of thinking required; PyAutoBrain uses that structure to choose the right reasoning agent.
Prompts live at <work-type>/<target>/<name>.md. The work-type (first folder)
determines the agent; the target (second folder) tells the agent which repo or
domain is affected.
| Work-type folder | Intent | PyAutoBrain agent |
|---|---|---|
feature/ |
new user-facing or scientific capabilities | feature planner |
bug/ |
incorrect behaviour, crashes, regressions | debugger |
refactor/ |
internal restructuring, no intended behaviour change | Refactor Agent (conductor; default-safe under --auto) |
docs/ |
documentation, tutorials, notebooks, examples | documentation agent |
test/ |
test coverage, smoke tests, validation scripts | test engineer |
release/ |
packaging, versions, deployment, release readiness | release engineer |
maintenance/ |
dependency updates, hygiene, cleanup, small technical debt | hygiene agent |
research/ |
exploratory scientific / algorithmic investigation before implementation | research analyst |
experiment/ |
prototypes, spikes, proof-of-concept work | prototype agent |
triage/ |
classification still unclear | (human triages, then re-homes) |
The canonical list of repos (with GitHub home, category and role) is
repos.yaml in this repo — the organism's body map.
The second folder names the affected repo or domain, e.g. autoarray, autofit,
autogalaxy, autolens, autolens_assistant, autolens_profiling,
autolens_workspace_developer, autobuild, pyautobrain; the workspace bucket
workspaces; or a topic series kept together as a unit (jax_substructure,
weak, cluster, priors).
Within the libraries, work classifies as library vs workspace for the
/start_library ↔ /start_workspace split — but that is decided from the
@RepoName references in the prompt body, not from the folder. The folder is
for human + agent legibility and PyAutoBrain routing.
This repository only defines the taxonomy and the metadata/documentation that
PyAutoBrain consumes. The agents themselves are not implemented here — they
live in PyAutoBrain. Prompts that implement
those agents are ordinary feature/pyautobrain/*.md prompts.
Humans reach this routing through the short PyAutoBrain commands (/feature,
/build, /health, /bug, /refactor, /docs, /research, plus the /route
NL router). The work-type verbs map onto the folders above; /route infers the
work-type from free text. The Brain stays implicit — users speak in short
commands; PyAutoBrain performs the routing. Bodies + the boundary live in
PyAutoBrain/skills/COMMANDS.md.
issued/, z_features/, z_vault/ are workflow-lifecycle folders;
autoprompt/ holds meta prompts about this repo's own infrastructure. None of
these are work-type folders and PyAutoBrain does not route them.