Index of every page on this site: llms.txt
All skills
Run it yourself

Audit or design agent-facing product capabilities: action/context parity, discoverable contracts, safe mutations, verifiable outcomes, and recovery. Use for explicit agent-readiness reviews of apps, repos, CLIs, APIs, MCP/A2A integrations, or Agent Skills; not every general code or documentation task.

Use this Skill: https://skilld.dev/gh/tristanmanchester/agent-skills/agent-use

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencesapi-sdk-mcp-readiness.md

≈815 tokens on demand. Your agent reads this file only when SKILL.md points to it.

API, SDK, MCP, and A2A contracts

APIs and SDKs

Specify operation inputs, stable IDs, pagination boundaries, scopes, errors, side effects, concurrency preconditions, and result states. Choose only applicable features: a synchronous read does not need a background task model. Document the actual retention/scope of write deduplication and distinguish partial/unknown outcomes. SDK retries must not silently repeat ambiguous mutations. Preserve API error details and versions; verify code against installed types.

MCP

Use the chosen protocol version and transport, actual tool/resource discovery, typed input/output, bounded results, and explicit errors. Resources and tool annotations describe behaviour; they do not prove a read is harmless or grant permission. Require server-side authorization and verify real effects. Do not promote this package's local server-card example into an MCP standard. Use the current official schema/SDK for wire-format conformance.

A2A 1.0 baseline

The bundled assets/templates/a2a-agent-card.json follows the reviewed 1.0 field shape: required name/description/version, ordered supportedInterfaces, a capabilities object, default input/output media types, and described skills. Each interface declares its service URL, protocolBinding, and protocolVersion (1.0, not the agent's release version). The card URL and service URL differ.

Security declarations use securitySchemes with the selected union member and securityRequirements; the example uses httpAuthSecurityScheme with Bearer and an empty scope list. This does not implement authentication. Replace the example with the actual enforced policy; no credentials belong in the card. Optional streaming/push/extended-card capabilities default false until implemented and tested.

A2A 1.0 also changes message/event encodings from older kind-discriminator forms. Use the versioned SDK/protocol schema rather than reusing old JSON samples. Send A2A-Version: 1.0 for the chosen interface and preserve any declared tenant routing. No silent version downgrade is needed for new designs in this skill.

Validate generated cards with the actual current SDK/schema before publication, then test discovery, authentication, messages/tasks, denied operations, declared optional capabilities, progress, and cancellation. Cancellation is an operation whose outcome must be observed, not guaranteed remote termination. Retain task and context identities when resuming. The local regression tests check selected scaffold fields only, not complete protocol conformance.

Events and webhooks

Verify authentication/signatures before trusting state changes, durably deduplicate receipts, acknowledge durable acceptance, and test concurrent replay/out-of-order updates. Bound payloads and validate callback destinations to avoid forwarding credentials or making unintended internal requests. Distinguish delivered events, accepted jobs, completed tasks, and verified external effects.

Source reviewed 2026-09-13: A2A 1.0 specification, particularly sections 3.6 and 4.4–4.5. Check the exact current MCP/API contract for the selected implementation; these design checks are not substitute schemas.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning1mo3 checks · Risk SAFE
  • Gen Agent Trust Hub1mo

    The skill 'agent-use' is a comprehensive toolkit for auditing and designing agent-ready systems. It includes documentation, templates, and Python scripts for scanning repositories and websites. Security analysis found no malicious intent; identified network and command execution patterns are aligned with the skill's primary purpose as an auditing tool.

  • Socket1mo

    1 alert: gptAnomaly

  • Snyk1mo

    Risk: LOW · No issues

Signed by skilld at 3323bc9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 weeks ago.

Steadyupdated last month
metadata
{
  "version": "3.0.0",
  "reviewed": "2026-09-13"
}
Other metadata
compatibility
Local helper scripts require Python 3.10+; structural validation also requires PyYAML 6.0.3 or later in major 6. Network and execution checks need separately authorised tools and environments.

README badge

README badge for tristanmanchester/agent-skills/agent-use