Team Agents Orchestrator

Run complex tasks with explicit role separation (operator, researcher, builder, editor), structured handoff contracts, and memory hygiene to prevent context...

Security Scan

The skill is an instruction-only orchestrator for multi-role workflows; its declared purpose matches the instructions and it requests no installs, credentials, or system access.

Purpose & Capability

Name/description describe multi-role orchestration and the SKILL.md contains step-by-step orchestration, handoff contract, and memory rules that align with that purpose. There are no unrelated env vars, binaries, or install steps requested.

Instruction Scope

Runtime instructions are confined to defining roles, creating handoff contracts, running role steps, and memory hygiene. They reference only the included handoff template and do not instruct the agent to read system files, export data, call external endpoints, or access credentials.

Install Mechanism

No install spec and no code files — instruction-only skill. Nothing will be downloaded or written to disk by the skill itself.

Credentials

No required environment variables, credentials, or config paths. The skill does not declare any primary credential or request access beyond the agent's normal execution context.

Persistence & Privilege

always is false and the skill does not request persistent system presence or modify other skills' configs. It allows normal autonomous invocation (platform default) but does not itself expand privileges.

Assessment

This skill appears coherent and low-risk because it's purely procedural text with no installs or credential requests. Before enabling it for sensitive or autonomous runs: (1) review and test the handoff template with non-sensitive tasks so you understand how outputs are produced; (2) do not grant unrelated credentials or external tool access unless you explicitly need them for a task — the orchestrator itself won't request them, but an agent using other skills might; (3) verify that the agent follows the memory-hygiene guidance in practice (to avoid accidental promotion of transient/private data); and (4) monitor first runs to ensure no unexpected external calls or data exfiltration occur. If you plan to give the agent additional capabilities (APIs, file access), reassess proportionality at that time.

Use this skill when one task spans multiple modes (research + implementation + communication) and quality can degrade without structure.

Roles

• Operator: owns user interaction and final decisions
• Researcher: gathers facts/options/constraints
• Builder: executes implementation steps and reports exact diffs/errors
• Editor: refines final messaging for clarity and actionability

Activation criteria

Activate when any of these are true:

• task is multi-step and high impact
• task crosses domains (e.g., web research + config changes + user-facing output)
• task has external/public consequences
• task likely needs iterative review

Orchestration loop

1. Define objective and constraints.
2. Create a handoff contract (see references/handoff-contract.md).
3. Run role steps in order (Researcher → Builder → Editor).
4. Operator reviews and decides final output/action.
5. Log durable outcomes and follow-ups.

Handoff discipline

Every handoff must include:

• objective
• constraints
• output format
• done criteria
• open questions

Receiver must restate scope before executing.

Memory hygiene rules

• Durable outcomes/decisions → daily memory
• Stable preferences/facts → long-term memory
• Temporary chatter/debug text should not be promoted unless reused

Common failure modes

• role blur (single messy pass)
• overreach without approval
• verbose output without added value
• memory pollution from transient logs

Output contract

Return:

• concise final result
• what changed
• unresolved risks/questions
• next best action