Skip to content
Back to skills

Integration Change Builder

ASecurity

Use `analysis-agent` or `task-agent` for shared-contract or delivery changes between internal runtime components or external providers. AI task/worker output merges alone do not establish this boundary.

  • 7 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 8, 2026
businessgo

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned October 4, 2026

npx -y skills add machenjie/rd-skills --skill integration-change-builder --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Integration Change Builder?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Integration Change Builder
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/machenjie-integration-change-builder/badge)](https://www.skillsdirectory.com/skills/machenjie-integration-change-builder)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: integration-change-builder
description: "Use `analysis-agent` or `task-agent` for shared-contract or delivery changes between internal runtime components or external providers. AI task/worker output merges alone do not establish this boundary."
---

# integration-change-builder

## Role

- **Analysis mode (`analysis-agent`):** Decide contract, authority, and failure behavior.
- **Task mode (`task-agent`):** Apply the boundary and reconciliation.

## When To Use

- internal runtime component or external provider integration change
- producer-consumer shared contract or delivery boundary alignment

## Do Not Use

- isolated change with no integration edge
- merging AI task or worker outputs without a runtime integration boundary
- unrelated source inspection

## Required Inputs

- component handoffs
- contract summary
- **Analysis mode (`analysis-agent`):** producer, consumer, credential, failure, and reconciliation evidence.
- **Task mode (`task-agent`):** accepted boundary decision with provider, duplicate, and recovery checks.

## Professional Decision Rules

- Keep integration decisions within declared owners, inputs, stops, and outputs.
- Align producer, consumer, provider, version, and applicable credential contracts; when the provider/protocol uses signing, bind its exact signed representation before changing the adapter or resolving implementation conflicts.
- Define timeout, retry, idempotency, ordering, verification, replay, unknown-outcome, partial-failure, compensation, and reconciliation behavior at the owning boundary.
- Validate the integrated diff, credential containment, mapping compatibility, and recovery behavior across affected consumers; isolated component success is insufficient.

## High-Value Gotchas

- A successful request can still leave a duplicate or unknown external effect.
- Provider sandbox behavior does not prove production credentials, quotas, ordering, or recovery.
- Signature, serialization, or adapter drift can invalidate an otherwise correct contract.

## Execution Checklist

- **Analysis mode:** Map producer, consumer, provider, credential, contract, and reconciliation authority.
- **Task mode:** Apply the accepted adapter boundary with idempotency and failure handling.
- Verify timeout, retry, duplicate, malformed, denied, and unknown-outcome behavior.
- Record provider assumptions and untested recovery paths as residual risk.
- Minimal validation: run contract and failure tests at the real adapter or calibrated sandbox.

## Stop / Escalation Conditions

Block unresolved authority or proof for affected provider, credential, ordering, duplicate/failure, sensitive-data, or adapter boundaries. Require signed-byte proof only when the current provider/protocol uses signing; do not introduce signing or reconciliation mechanisms absent from the contract. Escalate material production/provider/combined-impact/weak-recovery risk.

## Output Contract

- **Analysis mode (`analysis-agent`):** integration design; authority and failure decisions; reconciliation model.
- **Task mode (`task-agent`):** integrated boundary changes; conflict and credential decisions; unresolved provider risk.

## Targeted References

| Path | Type | Load when | Do not load when | Required by | Required output |
|---|---|---|---|---|---|
| [checklist](references/checklist.md) | decision-checklist | Closing or reviewing an external integration change needs provider-contract, timeout, retry/backoff, circuit-breaker, idempotency, webhook/replay, credential, reconciliation, test, and monitoring checks | The inline Skill quality gate is sufficient for an read or a deeper capability reference already covers the same checklist with task-specific detail | analysis-agent, task-agent | checklist-result, validation-plan |
| [index](references/index.md) | index | competing integration change builder references require dependency, conflict, or output-fragment selection | the integration change builder root or a task-named reference already resolves selection | analysis-agent, task-agent | reference-selection |
| [solution optimality](references/solution-optimality.md) | targeted | An outbound call, webhook, provider migration, retry, concurrency, delivery, or reconciliation design has a material failure, latency, cost, or state-consistency tradeoff | The provider contract and accepted task already determine a bounded adapter change with no material delivery or failure-mode choice | analysis-agent, task-agent | failure-decision, residual-risk |

Files in this skill

  • SKILL.md5.5 KB
  • examples/example-output.md817 B
  • references/checklist.md1.2 KB
  • references/index.md1.9 KB
  • references/solution-optimality.md1.5 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…