agent

Executes Agent engineering tasks for tools, skills, context, and runtime integration.

Updated Jun 27, 2026
One-click install
npx skills add https://github.com/HKUST-QUANT-SOCIETY/quantcode --skill agent-hkust-quant-society
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent
Source: https://github.com/HKUST-QUANT-SOCIETY/quantcode/tree/main/.opencode/groups/agent/skills/agent
Command: npx skills add https://github.com/HKUST-QUANT-SOCIETY/quantcode --skill agent-hkust-quant-society

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It standardizes how the Agent engineering group handles development tasks on Agent tools, Skills, context, and runtime integration within QuantCode, ensuring capability reuse, task grading, and controlled approvals instead of ad-hoc changes. ## Core Features & Use Cases - Capability Reuse First: Queries the Capability Catalog and group long-term Memory before building, recording real coverage with organization_reuse and proposing plans with organization_solution. - Graded Task Workflow: Classifies tasks by level, requiring L2/L3 work to wait for a frozen SolutionDoc before implementation begins. - Controlled Delivery: Produces change summaries, test evidence, artifact provenance, and open items, with knowledge promoted only after confirmation. - Use Case: A group member is asked to integrate a new runtime component; the Skill guides them to check existing organizational capabilities, draft a solution for user confirmation, implement in the authorized workspace, run tests, and report verified results. ## Quick Start Start a new Agent group task to integrate a runtime component, first checking the Capability Catalog and group Memory, then propose a solution plan for my confirmation before implementing.

Frequently Asked Questions about agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I start an Agent engineering task in QuantCode?

Begin by querying the Capability Catalog and the group's long-term Memory to reuse existing components, then record coverage with organization_reuse and propose a plan with organization_solution. The user confirms the specific version in the task interface before implementation starts.

What is the task grading process for L2 and L3 tasks?

Tasks are graded by level, and L2/L3 tasks must wait until the SolutionDoc is frozen before any implementation begins. This ensures design decisions are reviewed and locked before code changes.

Can the agent group deploy to production or get shell access?

No. The agent group only handles merge and permission decisions through the standard HumanGate. Production deployment belongs to a separate Admin management plane, and the group receives no production shell or deployment privileges.

Does group membership grant access to GitHub repositories?

No. GitGraph only consumes verified GitHub subjects whose repositories are visible under the mapped team. The group name itself does not grant any repository permissions.

What outputs does an Agent group task produce?

Each task outputs a summary of changes, test evidence, artifact provenance, and any unfinished items. New knowledge is first recorded as a candidate and only promoted after explicit confirmation.