Turn Colleagues into AI Skills with Teammate Skill

Where this fits

Your team relies on a senior engineer who answers the same architecture questions daily. Instead of interrupting them, you want an AI agent that thinks like them. Teammate Skill solves this by mining Slack, Teams, and GitHub history to create a structured "Work Skill" and a 5-layer persona—so your agent can respond with the same reasoning, preferences, and domain shortcuts. It’s for teams that want to preserve institutional knowledge without manual prompting or RAG pipelines.

Why agents benefit

• Context injection without manual setup – The skill automatically scopes data collection to channels, repos, and conversations, giving the agent a ready-to-use persona instead of you writing system prompts.
• Multi-source correlation – It merges Slack/Teams chat patterns with GitHub commit history and PR discussions, so the agent understands both what the teammate does and how they communicate.
• Continuous evolution – The persona updates as new interactions occur, meaning your agent stays current with shifting team practices—no stale snapshots.
• Drop-in compatibility – Works with Claude Code, OpenClaw, and any AgentSkills-compatible agent. You install once and reuse across your toolchain.
• Role-specific behavior – The 5-layer persona (identity, knowledge, style, workflow, and boundaries) gives the agent guardrails that match the original teammate’s decision-making patterns.

Practical scenarios

Onboarding a new teammate

A junior developer inherits a codebase maintained by someone who left. Instead of reading months of Slack history, they invoke the Teammate Skill to generate a persona from the departed engineer's GitHub PRs and Slack Q&A. The agent now answers "why is this function structured this way?" with the same rationale the original engineer would give.

Scaling an architect's review bandwidth

Your principal architect is overloaded with design doc reviews. You distill them into a skill that evaluates new proposals against the same criteria they apply in Slack threads. The agent gives initial feedback, flagging likely objections before the human architect reviews—cutting review cycles from days to hours.

Preserving tribal knowledge during reorgs

When teams shift domains, key context evaporates. Use the skill on a departing product manager’s Slack history and GitHub issues to create an agent that answers stakeholder questions about roadmap decisions, priority trade-offs, and historical constraints—without needing the person around.

Add it to your agent workflow

Install with a single command:

npx skills add LeoYeAI/teammate-skill

Then configure your agent (e.g., in Claude Code's agents directory) to load the skill and point it at your data sources. A minimal usage example:

name: "teammate-alex"
source: "slack://channels/engineering#arch; github://repos/core-api"
persona_depth: 5
update_frequency: "weekly"

After the first generation, the skill emits a persona file you can import into any AgentSkills-compatible agent. Run it in a cron job to keep the persona fresh.

Compare before adopting

Before wiring Teammate Skill into your stack, evaluate: quality signals (does the persona output actually pass a blind test with a human? Check the repo’s issues and PRs for real user feedback), maintenance freshness (last commit date, responsiveness to issues), alternatives (manual prompt engineering, full RAG setups, or simpler context bots—do you need a full 5-layer persona or just a summary?), and workflow fit (does your team already use Slack/Teams and GitHub heavily? The skill’s value scales with data volume).

Why it is worth tracking

Teammate Skill has 255 stars and rapid adoption because it tackles a pain every distributed team feels: making individual expertise reusable. The continuous evolution feature and agent-agnostic design make it a strong candidate for production use, especially if you already rely on Claude Code or OpenClaw. Given its active development and clear use case, it’s worth evaluating this month—especially if you’re building internal knowledge tools. For a quick reference, see the skill’s official listing on OpenAgentSkill to track updates and community notes.