Beta support and limits

Bowerloom 0.7.0-beta.1 prepares readable local definitions for your existing personal agent. Use this page to choose a supported task.

Open beta · 0.7.0-beta.1

Setup does not start workers, grant runtime access, or authorize connected actions.

Capability	Current boundary
Setup	Prepare a personal-agent profile, team definition, and working agreement in a new or existing project. Exact approval writes the planned files.
Portable skills	Pin skills from npm or GitHub with skills add, then install them for Claude Code and Codex with skills sync. A copy in the project does not prove that Claude Code or Codex finds or runs it.
Revision	Plan a change to an installed setup, then approve its exact revision before replacement.
Execution	Setup does not start workers, grant runtime access, or authorize connected actions.
Harnesses	Synthetic configuration commands support Codex and Claude Code fixtures. They do not run models or change live agent configuration.
Connections	MCP planning reads selected synthetic files. Local backend installation requires a separate plan and exact approval.
Company	Shared company access, retrieval, offboarding, and deletion are outside this beta's documented setup path.

These guides use macOS arm64 and Node 24.11.0. Other host systems are outside this documented installation path.

Running a team needs separate runtime permissions and controls. The setup commands do not grant them. Shared company access and unattended services are outside this setup walkthrough.

Beta limits

Keep your project out of iCloud Drive folders. Project commands such as bowerloom up, ls, status, apply, and skills sync refuse a project under ~/Documents or ~/Desktop when Desktop and Documents sync is on, and under the ~/Library cloud folders, with PROJECT_IN_CLOUD_FOLDER. The older forms that take an explicit path do not run this check. This beta reads skills only from public sources. See the beta limits.

Bowerloom 0.7.0 adds these commands, and each one is documented:

• bowerloom up --team <name> prepares a project one approved step at a time and ends at prepared, workers held, or at prepared, N items held with a next command for each item.
• bowerloom ls and bowerloom status read the project and write nothing.
• bowerloom team create, skill create, and prompt create make files inside .bowerloom.
• .bowerloom/skills.json pins skills at exact versions and commits. skills add, skills check, skills sync, skills recover, and skills migrate manage it.
• bowerloom skills sync puts the pinned skills in place for Claude Code and Codex. bowerloom apply puts the prompts in place, and any skill copy that is still missing.

Every change shows a plan first and needs --approve <revision>. Exit codes are 0 done, 1 refused, 2 usage, 3 approval required, and 4 held by a gate, and 130 stopped at the yes/no question. See Approvals and exit codes.

These limits are current facts of this beta:

• No worker starts. up ends at prepared, workers held.
• Native discovery of the skill and prompt copies by Claude Code and Codex is not observed.
• Skills come from public sources only, and only MIT and Apache-2.0 skills install.
• bowerloom revise refuses once a project holds content added after setup. The fix comes in a later 0.7 release.
• Bowerloom keeps history and deletes none. A full item needs a manual step. See the guide.
• The beta needs Node >=24.11.0 <25. If a global npm prefix needs sudo, set a user prefix instead. See Install Bowerloom.

Read status fields

Specification readiness means that the inspected files meet their required rules. Runtime readiness and permission to execute remain separate. ready-for-review means that the setup needs your review. It does not mean that a team runs.

Setup and revision

Setup prepares fixed Engineer, Founder, or Research definitions. Exact approval writes 20 setup files and one private receipt.

Revision plans a replacement before applying it. These paths do not import live agent configuration, create hosted agents, or start workers.

Harnesses

Synthetic configuration commands process selected Codex and Claude Code fixtures. They do not change live configuration or run models.

Portable skill installation supports a new Codex workspace and a new Claude Code workspace. It does not support an existing target. For an existing project, skills sync puts skills in place and apply puts prompts in place for both harnesses, Claude Code and Codex. A harness is the agent application that reads the copies. Native discovery by Claude Code and Codex is not observed.

A projected file does not prove that an agent discovers or runs it. Read Codex and Claude Code.

Connections

The MCP command plans from three selected synthetic inputs. It does not connect a server, resolve secrets, or invoke tools.

The local backend uses a separate macOS Docker profile. Read MCP and Local backend for their prerequisites.

Company

Shared company access, retrieval, offboarding, and deletion are outside this beta's documented setup path.

A local setup does not enroll employees or establish a shared service. Real business data needs an approved retention and deletion policy.

Security

Setup approval grants no unattended execution or connected-tool authority. Review each later action and its required controls separately.

Keep credentials, customer data, raw receipts, and private paths out of shared files. Read Trust and security boundaries.

Help and feedback

Use GitHub Issues for reproducible bugs.

Ask questions in Q&A. Share feedback in General.

Discuss feature proposals in Ideas. Bug reports and feedback explains what to include.

Release identity

Use the documentation for your installed version. Install Bowerloom gives the entry path, and Release notes describes its changes.