Repository navigation
Library Integration
SYSTEMX can be copied as a self-contained folder or installed as the dotsystemx
Python distribution, which exposes the systemx command and importable module.
The package includes the same template files and uses no third-party runtime
dependencies. A package installation does not initialize or alter any project.
From the public Git repository:
python3 -m venv .venv
# Activate this environment using your platform's normal command, then:
python -m pip install "git+https://github.com/WayneTechLab/dotSYSTEMX.git@v1.8.2-alpha.1"
systemx setup --profile project
systemx install --target "/path/to/project" --profile projectOn Windows use py -3 -m venv .venv and .\.venv\Scripts\Activate.ps1 if
your execution policy permits it; alternatively call that environment's Python
and SYSTEMX executables directly. No policy change or administrator access is
needed. Git-based pip installation requires Git. An extracted release can also
be installed with python -m pip install /path/to/dotSYSTEMX, or use its published
wheel without Git. The package is distributed through GitHub; no PyPI publication
is implied. Select exact versions from the release history.
Upgrading the Python package only updates that tool environment. Each project's
default version and pin remain independent. systemx update --target ... selects
project defaults under the preservation contract.
The public release ID 1.8.2-alpha.1 is spelled 1.8.2a1 in Python package
metadata and wheel filenames (PEP 440). They identify the same release.
systemx --version shows both; status --target ... shows project defaults.
The library and CLI are alpha APIs; review release policy
before upgrading integrations.
from systemx import install, update, status, set_policy, export_chat, alias
# Plan without touching the target, using a local reviewed distribution:
plan = install("/path/to/project", source="/path/to/template/.SYSTEMX",
profile="project", dry_run=True)
# Apply only after choosing the intended project:
result = install("/path/to/project", source="/path/to/template/.SYSTEMX")
print(status("/path/to/project"))
# Updates require explicit unlocking of the default pin:
set_policy("/path/to/project", pin="none")
update("/path/to/project", source="/path/to/new-template/.SYSTEMX", dry_run=True)
# Export only to a new file; nothing is uploaded:
export_chat("/path/to/project", "/path/to/new-session.md", agent="agent.0")Management functions return structured dictionaries. run and projects return
subprocess.CompletedProcess; inspect returncode, stdout, and stderr. Importing the library performs no
filesystem or network actions. The CLI renders management results as JSON and
uses exit status 2 for a rejected operation. Project command execution preserves
the underlying command's exit status. Use systemx run --target ... -- <command>
to invoke the selected defaults against that project's canonical records.
install defaults to the package's bundled version. Supply version to fetch a
specific GitHub release, or source for offline installation. update uses the
installation's selected repository unless given a local source. Downloaded files
are bounded, path-checked, and fingerprint-verified; archives are never blindly
extracted into a project. Existing root files are opened only for inspection or
left alone, and missing files are created exclusively.
These functions are a possible future adapter boundary for MCP or other tools. Such an adapter would still need explicit target selection, access controls, permission handling, and evidence of writes. No server, remote endpoint, or additional authority is included by installing this library.
Packaging references: PyPA project metadata and entry points.
The API always manages the real .SYSTEMX directory beneath its target.
install(target, lowercase_alias=True) optionally creates a local relative
.systemx -> .SYSTEMX link. alias(target) reports the layout without writing;
alias(target, create=True, dry_run=True) previews and alias(target, create=True)
requests creation. Case-insensitive filesystems report filesystem-equivalent.
The status response includes pathLayout. Conflicting case variants raise
ValueError without merging or deleting them. See the case contract.
from systemx import first_run, audit, uninstall, restore
first_run("/path/to/project") # preview
first_run("/path/to/project", apply=True)
audit("/path/to/project") # read-only, scoped footprint and operation history
uninstall("/path/to/project", backup="/path/to/new-backup") # preview
uninstall("/path/to/project", backup="/path/to/new-backup", apply=True)
restore("/path/to/project", backup="/path/to/new-backup") # preview
restore("/path/to/project", backup="/path/to/new-backup", apply=True)Uninstall moves the whole .SYSTEMX folder, including user records, to an external
backup on the same filesystem. It does not uninstall this Python package. Run
removal/restore from an external environment or checkout. Preserve the receipt
and use the cleanup guide for package removal and intentional leftovers.
from systemx import projects, export_chat
preview = projects("/path/to/workspace", ["add", "Project-A", "--kind", "software"])
preview.check_returncode()
created = projects("/path/to/workspace", ["add", "Project-A", "--apply"])
created.check_returncode()
packet = projects("/path/to/workspace", ["context", "--project", "Project-A"])
packet.check_returncode()
print(packet.stdout)
export_chat("/path/to/workspace", "/path/to/project-a-chat.md", project="Project-A")projects(target, arguments, *, offline=True, capture=True) runs the selected
outer release's projects command. The target is the directory containing the
outer .SYSTEMX, not a child's folder. Arguments are a nonempty list/tuple of
strings. Scope commands require --project NAME or --root.
export_chat(..., project="Project-A") includes only that child's active records
with shared default guidance. Omit project for the root packet. The equivalent
CLI is systemx export-chat --target PATH --project Project-A --output NEW_FILE.
It never overwrites an existing output or uploads the packet. The returned
management dictionary includes project (null for root), version, output,
and uploaded: false. Review private context before sharing. See
SYSTEMX PROJECTS for scoping, commands, and preservation rules.
from systemx import run, projects
result = run("/path/to/workspace", ["roles-init"], offline=True, capture=True)
result.check_returncode() # preview; add --apply after reviewing the intended scope
projects("/path/to/workspace", ["agent-z", "policy", "--project", "Project-A"]).check_returncode()run is exported by the package and returns subprocess.CompletedProcess.
Arguments must be an explicit string array; they cannot override the chosen
workspace's global --root. Scoped projects ... --root remains valid.
Role commands perform local record operations; they do not schedule external
jobs, call a model, run checks, or upload evidence. Use offline=True to suppress
an independently enabled startup update check. See Agent X and
Agent Z.
.SYSTEMX · DOT SYSTEMX · Maintained by Wayne Tech Lab LLC · Repository · User manual · Prompt cookbook · MIT license
Alpha 1.11.0-alpha.7 — use at your own risk; may change daily. Release policy. Project records, configured commands, and verified evidence determine the state of each adopted project.
DOT SYSTEMX · User manual
1. Start here
- The Guided Folder: 4K walkthrough and visual chapter index
- Start with a Folder: 4K animated guide
- Introduction video and 4K poster
- Advanced walkthrough: inside the coordination layer
- Choose a workspace
- First-time setup
- Daily workflow
- One-shot project guide
- Shared workspaces, clouds, and bots
- Use-case infographics: four environments
- Four-part system infographic
- Visual Guide
- New chat setup prompt
- Prompt cookbook
2. Work and continuity
- SYSTEMX PROJECTS
- AUTOBOT and Agent Alpha: Tier 2
- Local DATA storage
- Planning and memory
- ROOT planning, rules, and task delta
- System MAP
- Access Matrix
- Task workflow
- Agent X: events and time
- Agent Y: CLI and PR channels
- Agent Z: fixed review
- 100 review questions
- Agent 0 and subagents
- Subagent processing
- Advanced use cases
- Evidence and acceptance
- Anti-drift and long-running work
- Tokens, cost, and time
3. Setup and maintenance
- Project / VS Code
- OS directory
- Google Drive
- LLM chat
- Exact casing and alias
- Installation and updates
- Upgrading
- Uninstall and cleanup
4. Advanced reference
- Standard format
- Technical Guide
- Stack Guide
- Library integration
- CLI and Tools
- Command reference
- Configuration and quality
- Operations and troubleshooting
- Security and support
- Code review and readiness
5. Project information