Build Team

Build Team

The user-invoked counterpart to hire-agent. Where hire-agent suggests one agent reactively when it spots a repeated pattern, build-team looks at the whole picture (the user's profile, judgment rules, and real task history) and proposes a tailored team in one pass. It is a Maker-Checker flow (Proposal 11, E6): Claudia proposes, an independent Checker validates, the user approves, and only then is anything written.

What this is, and is not

• It proposes and, on approval, scaffolds agent definitions. It never auto-spawns agents or takes external actions.
• It reuses the existing roster first (the agents in .claude/agents/) and adds a new role only when the user's work clearly justifies one.
• It starts minimal. A small team that covers the real work beats a sprawling org chart the user will never use. Growth comes later, from hire-agent and the proactive team-update suggestions (Proposal 11, E7).
• It does not duplicate agent-dispatcher routing logic, and it does not change how Claudia delegates. It only proposes which agents exist.

When to fire

Trigger phrases: "build my team", "set up my agents", "what agents should I have", "design my team", "/build-team", "tailor your team to my work".

Also the landing spot when a proactive suggestion (from capability-suggester or hire-agent) escalates from "add one agent" to "let's set up your whole team".

Do NOT fire for: adding a single agent for one observed pattern (use hire-agent), folder/structure changes (use structure-generator), or command/workflow additions (use capability-suggester).

The flow

Step 1: Read the profile

Gather, silently:

• context/me.md: role, archetype, priorities, the shape of a typical week.
• context/judgment.yaml (if it exists): priorities, delegation preferences, surfacing rules. A user who said "just auto-process transcripts" wants a processing agent; a user who said "always run things by me" wants a smaller, more ask-first team.
• Recent task patterns: what the user has actually asked Claudia to do repeatedly (lean on the same signals hire-agent uses, plus memory_recall for recurring work). The team should map to real work, not a generic template.

If context/me.md does not exist, the user has not onboarded. Route to onboarding first; do not propose a team into a vacuum.

Step 2: Maker proposes the team

Claudia (the Maker) drafts a team:

1. Start from the minimal seed for the user's archetype (see the seed table below).
2. Adjust to the user's actual work: drop a seed role they will not use, add a role a recurring task clearly needs.
3. For each role, write a one-line rationale tied to the user's real work ("You process client transcripts most weeks, so a Document Processor handles the extraction while I keep the judgment"). A role with no concrete rationale does not belong in the proposal.
4. Every new (non-roster) role must pass the hire-agent candidate test: compute-heavy, judgment-light, structured in and out, repeatable. If it needs relationship context or would take external actions, it is not an agent.

Step 3: The Checker validates (bounded)

Dispatch the loop-checker agent with the team proposal as the artifact and the team rubric below. The Checker scores independently and returns the standard verdict JSON (verified, score, issues, hard_constraint_violated).

If the Checker returns verified: false (most often: too large, or a role that needs judgment), the Maker revises once based on the issues (usually by trimming) and re-checks. Cap at 2 revisions. This is the same bounded Maker-Checker discipline every loop follows; the team proposal is not exempt.

The team rubric:

Hard constraints (any one forces verified: false): more than 5 roles in a first team, or any role that would require relationship judgment or take external actions.

Step 4: Status file + approval gate

Write team_status.md in the standard schema (docs/loop-status-schema.md): last_input (one-line profile summary), maker_proposal (the team), score, checker_verdict, verified, next_action: await user approval. Write it to the workspace, atomically (temp sibling then rename).

Then present the validated proposal to the user: each role, its model tier, and its one-line rationale, plus what is reused vs new. Write nothing to .claude/agents/ or .claude/skills/ yet. This is a Human-Approved action (see claudia-principles.md). Ask plainly: "Want me to set this team up?"

Step 5: Apply and rollback (only after explicit approval)

On a clear yes:

1. For each new role, scaffold .claude/agents/<name>.md using the agent definition format from hire-agent (frontmatter: name, description, model, dispatch-category, dispatch-tier, auto-dispatch; body: role, job, output format, constraints).
2. For any existing file you modify (for example, adding the team to .claude/agents/README.md), write a .bak sibling first, the same mechanism the installer upgrade flow uses. New files need no .bak.
3. Report exactly what was created or changed, and how to undo it: "Rollback by restoring the .bak files and deleting the new agent files, or just say 'undo the team setup' and I will."
4. Do not edit agent-dispatcher.md routing here. The new agents exist; routing them is the dispatcher's job and can be tuned separately.

Rollback on request: restore each .bak, delete each newly created agent file, and report what was reverted.

Minimal seed teams (Proposal 11, D5)

Starting points, not prescriptions. Every seed gets pruned and adjusted to the user's actual work in Step 2. All roles are drawn from the existing roster.

Two or three roles is a healthy first team. Five is the ceiling, not the target.

Safety

• Approval gate is non-negotiable. Nothing is written to agents/ or skills/ without an explicit yes (Step 4).
• Reversible. .bak siblings for any modified file; new files are deletable; "undo the team setup" restores the prior state.
• Judgment stays human. Agents handle processing. Relationship judgment, strategy, and external actions never get delegated to a generated agent.
• Minimal by default. When the Checker and the Maker disagree on size, the smaller team wins. The user can always add more later.

See also

• hire-agent for proactive, single-agent suggestions from one observed pattern.
• structure-generator for the archetype folder structures this composes with.
• capability-suggester for command and workflow additions (not agents).
• .claude/agents/README.md for the current roster and the two-tier model.
• .claude/skills/_loop/checker.md for the Checker brief the validation uses.
• docs/loop-status-schema.md for the team_status.md format.