Adopt Keelson

Establish a disciplined, issue-driven agentic operating model with automated tracking and strategic human-in-the-loop.

Install
cmdop skills install agensi-adopt-keelson

Adopt Keelson is a skill that installs a disciplined, issue-driven operating model onto an AI agent working inside a software repository. Rather than letting an agent make ad-hoc changes, it enforces a structured frame → plan → implement → open-change → gate → close lifecycle, addressing the problem of uncoordinated agent contributions that bypass repository conventions or fail to communicate intent.

Installing the skill triggers a multi-round configuration session, typically 8 to 20 rounds, during which the skill maps the project’s specific SDLC requirements. It discovers branching models, tracker terminology for tools such as GitHub, Jira, and Linear, and delivery stages. From this session it generates a repository-specific AGENTS.md file and a set of localized, plain-language lifecycle skills tailored to that project.

A key output is the establishment of a Tactical vs. Strategic boundary: the skill defines exactly which decisions the agent handles autonomously and which require human approval before proceeding. Quality gates and engineering standards are embedded directly into the agent’s operating instructions. The agent ends up managing its own backlog, producing self-reviewing pull requests, and halting to ask for clarification rather than proceeding under uncertainty.

This skill is appropriate when a team needs an AI contributor that respects project-specific documentation standards such as ADRs, follows defined status-flow rules, and operates within a human-in-the-loop governance framework. It is not a runtime execution tool — it is a configuration and onboarding skill that shapes how an agent behaves going forward.

Use cases

  • Configure an agent to follow your team's specific branching model and PR conventions
  • Define the boundary between autonomous agent actions and decisions that require human sign-off
  • Generate a repository-specific AGENTS.md that encodes your SDLC rules into the agent
  • Integrate issue tracker terminology (GitHub Issues, Jira, Linear) into the agent's workflow
  • Embed quality gates and ADR documentation standards into the agent's lifecycle instructions
  • Onboard an agent onto a new repository without it making uncoordinated or undocumented changes

When to use it

  • A team wants an AI agent to contribute code changes under structured, auditable governance
  • The project uses an issue tracker and requires every change to be linked to a tracked issue
  • Human approval is required at specific lifecycle stages before the agent proceeds
  • Repository conventions, branching models, or documentation standards must be enforced consistently
  • The team has experienced uncoordinated or undocumented agent changes and wants to establish guardrails

When not to use it

  • No tools are provided by this skill, so it cannot perform runtime database, API, or file operations on its own
  • Projects with no defined SDLC or issue tracking workflow will not benefit from the configuration session
  • Teams that want fully autonomous agents with no human-in-the-loop checkpoints
  • Environments that do not use any supported issue tracker (GitHub, Jira, Linear, or similar)
  • One-off scripting tasks where a persistent operating model adds unnecessary overhead