Coordinate agent teams for adversarial review or parallel implementation, including team spawning, task sizing, review protocol, and teammate health monitoring. Only loaded when agentTeams is enabled. Use when dispatching multi-agent work or adversarial review.
From flownpx claudepluginhub synaptiai/synapti-marketplace --plugin flowThis skill is limited to using the following tools:
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Searches prompts.chat for AI prompt templates by keyword or category, retrieves by ID with variable handling, and improves prompts via AI. Use for discovering or enhancing prompts.
Enables AI agents to execute x402 payments with per-task budgets, spending controls, and non-custodial wallets via MCP tools. Use when agents pay for APIs, services, or other agents.
Domain skill for orchestrating agent teams. Only relevant when agentTeams: true in settings.
Requires: CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS environment variable.
INDEPENDENT ANALYSIS BEFORE SHARED CONCLUSIONS. Teammates who see each other's findings are contaminated reviewers.
The value of a team is independent perspectives. Sharing findings before independent analysis destroys that value.
| Scenario | Recommendation | Rationale |
|---|---|---|
| Review with 3+ facets | Spawn team | Independent analysis enables adversarial protocol |
| Large feature (>5 acceptance criteria, independent modules) | Suggest team | Parallel implementation across modules |
| Debugging with competing hypotheses | Spawn team | Each teammate tests a different hypothesis |
| Small feature (<3 criteria) | Single session | Team overhead exceeds benefit |
| Sequential dependencies | Single session | Can't parallelize sequential work |
When spawning a review team:
Each reviewer works in isolation:
Teammate(security-reviewer):
"Review PR diff for security issues independently. Do not share findings until asked.
Return P1/P2/P3 findings with file:line citations."
Teammate(code-reviewer):
"Review PR diff for logic, quality, edge cases independently. Do not share findings.
Return P1/P2/P3 findings with file:line citations."
Teammate(convention-checker):
"Validate conventions independently. Return findings."
Collect all findings from teammates.
Each reviewer challenges others' findings:
Lead synthesizes results:
From settings: timeouts.teammateTimeout (default: 300 seconds)
If a teammate exceeds timeout:
If a teammate crashes or returns an error:
The nudge-idle-teammate.sh hook handles idle teammates:
When spawning an implementation team:
When teams are disabled or spawn fails, all team patterns fall back to:
The calling command handles the fallback — this skill only provides team-specific knowledge.