메뉴
Use when generating a human-readable onboarding document from HARNESS.md, AGENTS.md, and REFLECTION_LOG.md — produces a friendly guide for new team members joining a harnessed project
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
This skill should be used when the user asks to "assess AI literacy", "run an assessment", "check literacy level", "evaluate our AI collaboration", "where are we on the framework", or wants to determine their team's AI literacy level using the ALCI instrument.
This skill should be used when an agent is deciding whether to author a dynamic workflow — a self-authored, ephemeral multi-agent harness — for a task. Use when the user asks about "dynamic workflows", "multi-agent harness", "fan-out", "subagents in parallel", "adversarial verification", "the ultracode trigger", "should I use a workflow", or when a task looks long-running, massively parallel, highly structured, or adversarial. Names the six composable patterns, the election rubric for when NOT to spend the extra compute, and the two governance invariants (INV-1 ephemeral-proposes, INV-2 quarantine).
This skill should be used when the user asks about "verification slots", "integrating a linter", "adding a deterministic tool", "harness-enforcer", "constraint enforcement interface", "wrapping a tool", or needs the technical reference for how deterministic and agent-based checks work in the harness framework.
Use when setting up automatic PR constraint enforcement via GitHub Actions — covers the advisory-vs-blocking split, workflow installation, configuration options, and reading the output
Use when the user wants to estimate or predict the cost, token usage, or time of a task BEFORE it runs — "how much will this feature cost to build", "estimate the tokens for this spec", "what will this slice cost", "predict the agent-compute time before I commit" — produces a token + time estimate as a range with disclosed confidence, adding a dollar figure only when an observability/costs snapshot grounds it. The prospective sibling of cost-tracking, which records actual spend after the fact.
Use when the user wants to capture AI tool costs, review spending trends, set cost budgets, or integrate cost data into health snapshots — guides quarterly cost capture, records data in a structured format, and updates MODEL_ROUTING.md with observed cost patterns
| name | harness-onboarding |
| description | Use when generating a human-readable onboarding document from HARNESS.md, AGENTS.md, and REFLECTION_LOG.md — produces a friendly guide for new team members joining a harnessed project |
HARNESS.md is written for agents and experienced team members. New contributors need something different — a friendly, practical guide that explains "here's how we work" without requiring them to parse constraint definitions and GC rule syntax.
This skill generates ONBOARDING.md from three sources the project
already maintains. No new data entry is needed — the onboarding
document is a human-readable projection of existing harness state.
| Source | What it contributes |
|---|---|
| HARNESS.md | Stack, conventions, constraints, GC rules, observability cadence |
| AGENTS.md | Gotchas, architecture decisions, test strategy, design decisions |
| REFLECTION_LOG.md | Recent surprises with context or workflow signal |
HARNESS.md, AGENTS.md, and REFLECTION_LOG.md are the single source of truth. Sync is one-way only — these files drive ONBOARDING.md. The generated document is never edited directly.
Write for a human who just joined the team and needs to ship their first PR. The reader is a skilled developer who knows nothing about this specific project.
One paragraph. What this project is, why it matters, what a new contributor should expect. Pull from the README if it has a good description; otherwise synthesise from the stack and conventions.
From HARNESS.md Context > Stack. List each technology with a brief note on how it's used. Don't just repeat the HARNESS.md bullets — add context about what each tool does in this project.
From HARNESS.md Context > Conventions. Transform the terse bullet format into readable prose. Explain not just the rule but why the team follows it. Use examples where helpful.
From HARNESS.md Constraints. Group by when they fire:
For each constraint, explain in plain English what it checks and what happens if it fails. Skip the Tool and Enforcement fields — those are implementation details for the harness, not for the contributor.
From AGENTS.md GOTCHAS + recent REFLECTION_LOG entries where
Signal is context or workflow. Present as practical warnings:
"what happened" and "how to avoid it". Limit to the 5-7 most
relevant entries — the onboarding doc should not be exhaustive.
From AGENTS.md ARCH_DECISIONS. Each entry: what was decided, why, and what the alternatives were. Frame as "the team already decided this — here's why."
From AGENTS.md TEST_STRATEGY. Explain how quality is assured. What tools run, what they check, how to run them locally.
Brief explanation of the three enforcement loops. One sentence each for advisory (hooks), strict (CI), and investigative (GC). Mention the observability cadence from HARNESS.md Observability. Don't explain the framework — just explain what the contributor will encounter.
Synthesise from Constraints and Conventions. Include only items the contributor needs to check before pushing. Format as a numbered list they can work through. Example items:
Links to HARNESS.md, AGENTS.md, REFLECTION_LOG.md, and the docs site if one exists.
The generated document must match the section structure of
templates/ONBOARDING.md. The /harness-onboarding command
includes a validation checkpoint that verifies all 10 sections
are present and fixes missing sections in place.
A GC rule checks monthly whether ONBOARDING.md is stale — whether
any of its three sources (HARNESS.md, AGENTS.md, REFLECTION_LOG.md)
have been modified more recently than ONBOARDING.md. When stale,
the GC rule flags it and suggests re-running /harness-onboarding.
