Skip to content
Design
Skill

/balise-handoff

Audits and prepares designs, specifications, prototypes, design-system artifacts, and implemented interfaces for developer handoff using explicit scope, source-of-truth mapping, evidence levels, risk-based gates, open-decision ownership, and acceptance criteria. Use to assess

BOOST
From plugin
balise-skills
1215 skills
Install
$ npx -y skills add mrstev3n/balise-skills --skill balise-handoff --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/balise-handoff

Context preview

The summary Claude sees to decide when to auto-load this skill.

Audits and prepares designs, specifications, prototypes, design-system artifacts, and implemented interfaces for developer handoff using explicit scope, source-of-truth mapping, evidence levels, risk-based gates, open-decision ownership, and acceptance criteria. Use to assess

SKILL.md

balise-handoff.SKILL.md
name: balise-handoff
description: Audits and prepares designs, specifications, prototypes, design-system artifacts, and implemented interfaces for developer handoff using explicit scope, source-of-truth mapping, evidence levels, risk-based gates, open-decision ownership, and acceptance criteria. Use to assess whether a Figma selection or product surface is ready for implementation, improve developer handoff, review Ready for dev claims, assemble a handoff contract, trace design-to-code provenance, or revalidate a design after changes.

Handoff Readiness

Determine whether a specific version can be implemented without unresolved critical ambiguity. Treat readiness as an evidence-based agreement between people, artifacts, and systems—not as file cleanliness, a universal checklist, or a promise that future code will be correct.

Operating contract

  • Audit the requested selection, flow, component, feature, or release—not the entire workspace by default.
  • Preserve source designs, components, specifications, code, tickets, and status markers unless the user explicitly authorizes edits.
  • Identify the target platform, implementation team, release context, design-system relationship, and risk before applying gates.
  • Reuse the team's existing vocabulary, components, tokens, documentation, tickets, tests, and delivery workflow.
  • Distinguish **observed**, **structurally inferred**, **documented**, **simulated**, **linked to code**, **runtime verified**, and **unknown**.
  • Never treat a Figma status, annotation, prototype, generated snippet, component name, or visual match as runtime proof.
  • Never invent API behavior, data shape, permissions, breakpoints, accessibility semantics, analytics, asset rights, or implementation constraints as facts.
  • Do not require every dimension mechanically. Apply gates proportionally, but never average away an unresolved critical blocker.
  • Make every open critical decision actionable with impact, owner, next action, and fallback when one exists.
  • Do not turn handoff into a final ceremony. Prefer evidence produced and maintained throughout product work.

Choose the mode

Infer the mode from the request. Default to **Audit** when edit intent is absent.

| Mode | Action | | --- | --- | | **Audit** | Inspect evidence, apply gates, and issue a readiness verdict without editing. | | **Prepare** | Add or repair handoff structure, annotations, mappings, links, and decision records after explicit edit intent. | | **Verify** | Reinspect affected gates after corrections, implementation, or design-system changes. | | **Contract** | Produce a durable handoff summary from available evidence without changing the source. | | **Complete** | Audit, prepare authorized changes, reverify, and deliver the contract. |

Audit and Verify are read-only. In a design tool, Prepare should use a labeled handoff area or copy by default. Edit source frames, shared components, status markers, or production code only when the user explicitly requests that mutation.

Load only the needed references

  • Read [readiness-model.md](references/readiness-model.md) when selecting gates, classifying severity, assigning evidence, or deciding the verdict.
  • Read [figma-execution.md](references/figma-execution.md) when the source or output includes Figma frames, components, variables, prototypes, annotations, Dev Mode, Code Connect, or MCP context.
  • Read [source-of-truth-and-change.md](references/source-of-truth-and-change.md) when design and code disagree, sources are distributed, dependencies changed after handoff, or provenance is unclear.
  • Read [implementation-evidence.md](references/implementation-evidence.md) when inspecting code, Storybook, tests, runtime behavior, accessibility, assets, data contracts, or acceptance proof.
  • Use [handoff-readiness-contract-template.md](assets/handoff-readiness-contract-template.md) when the user requests a durable readiness report or implementation package.

Workflow

1. Frame the handoff candidate

Establish:

  • selected surface and stable artifact link or path;
  • exact version, branch, revision, status, or review date;
  • included and excluded flows, states, platforms, and viewports;
  • intended release, implementation audience, and decision owner;
  • user and product goal;
  • known design-system, codebase, data, asset, and policy dependencies;
  • risk profile and requested outcome.

If scope or version cannot be identified, do not inspect arbitrary nearby work and call it the target. Mark the candidate **Not assessable** until the missing boundary can be resolved.

2. Choose a proportional profile

Use the lightest profile that still controls the real risk:

  • **Quick:** small, reversible change; colocated team; known implementation; low consequence.
  • **Standard:** ordinary feature or component handoff; inspect every dimension and activate applicable gates.
  • **Deep:** external or distant team, new platform, wide reuse, migration, money, identity, permissions, sensitive data, safety, security, localization, or consequential accessibility risk.

State the profile and why. A Deep profile changes evidence requirements; it does not justify auditing unrelated surfaces.

3. Inspect the real delivery ecosystem

Inspect the selected artifact and relevant adjacent evidence:

  • parent flow, alternate branches, states, prototypes, and annotations;
  • components, instances, properties, overrides, libraries, variables, modes, styles, tokens, and layout rules;
  • product requirements, decisions, tickets, API or data contracts, content sources, asset sources, and analytics requirements;
  • canonical code components, routes, state owners, tests, Storybook stories, Code Connect mappings, and release history;
  • status, version history, change notes, owners, and communication channel.

Presence is not freshness. A linked ticket may be stale; a matching name may represent a different concept; a clean frame may omit behavior. Record access limitations in

Read more
Ships withbalise-skills

Des skills portables pour Claude Code, Codex, Cursor, le Figma Agent, Figma Make et les autres outils compatibles avec le standard Agent Skills. Installez un skill précis ou une collection prête à l’emploi. Chaque skill utilise SKILL.md comme format canonique.

Get the whole plugin

Other skills on balise-skills.