RetentaRetenta

Evidence record · captured 2026-10-11

Claude API

A reference skill for building on the Claude API and Anthropic SDKs: it first checks the project for other-provider markers and stops if found, detects the language, then routes to per-language docs (Python, TypeScript, Java, Go, C#, curl and more) for streaming, tool use, files, batches, caching and Managed Agents. It carries a dated table of current model IDs and migration rules and warns the agent its training prior on the API is likely stale.

Official publisherUnverifiedSource reviewedBehavior not tested
Source review

Editorial assessment

This is a source-only assessment. It does not claim runtime or security testing.

What it does

Front-loads a stale-knowledge check: the agent scans for other providers and stops if it finds one, then works from a dated model table, an API drift table and per-language SDK docs, not recall.

Best for

  • Writing or reviewing code that calls the Claude API through an official SDK in Python, TypeScript, Java, Go, Ruby, C# or PHP
  • Moving an integration to a newer Claude model with the migrate subcommand, which confirms file scope first
  • Choosing between a manual tool loop, the SDK Tool Runner and Managed Agents for an agent build

Strengths in the source

  • Before editing, scans the target for non-Anthropic markers such as import openai and stops to ask, so it does not rewrite another provider's code with Anthropic calls.
  • A drift table lists API shapes that changed, such as budget_tokens being rejected on newer models, and states that the bundled language files outrank recalled patterns.
  • Forbids guessing SDK names: bindings must come from the bundled language files or the SDK repositories listed in shared/live-sources.md.
  • Subcommands that spend money on model calls, such as cost-optimize and build-eval, require the user's approval of the budget before anything runs.

Limitations

  • The model table is marked cached as of 2026-10-06; model IDs, prices and beta header names here change with each release, so the pinned copy goes stale quickly.
  • Sets firm defaults: it mandates claude-opus-5-5 unless the user names another model, and turns on the server-side fallbacks beta by default for several models.
  • Needs Anthropic or cloud-platform credentials for any call, and routes into large per-language folders and live docs, so the roughly 100 KB SKILL.md is not the whole behavior.

When not to use it

  • The project targets OpenAI, Gemini or another provider; the skill is written to stop in that case.
  • You want Claude Agent SDK code, which the skill says it does not generate.

Reviewed by Retenta editorial on 2026-10-11; scope: source only.

Files included in this review
  • skills/claude-api/SKILL.md

Evidence and boundaries

Execution · not tested

Retenta has not executed this skill.

Security · not tested

Bundles 2 scripts (shared/evals/report/*.mjs, shared/preserved-thinking-migration/*.py) that the agent may run. Retenta has not run them or scanned them.

Pinned publisher source
skills/claude-api/SKILL.md
Revision
dbd4588f9e1033efb41dad4bef2f7947c8993d44
License evidence
Apache-2.0Retenta preserves the license text captured with this immutable upstream snapshot; no independent legal or security review is claimed.
Last source change
2026-10-09
Independent verification
Published by Anthropic, but not independently verified by a third-party reviewer.
Upstream repository
Open publisher repository

Compatibility

Claude Code (plugin marketplace)

documented

Named in the upstream skill or repository documentation; not tested by Retenta.

Claude.ai

documented

Named in the upstream skill or repository documentation; not tested by Retenta.

Claude API (Skills API)

documented

Named in the upstream skill or repository documentation; not tested by Retenta.

Requirements and access

  • An Anthropic SDK for the project language
  • ANTHROPIC_API_KEY (or cloud-platform credentials)

Referenced tools

No named tool is required by the reviewed SKILL.md.

Install from the source

Clones the current default branch, which may be newer than the reviewed revision dbd4588. The destination shown is Claude Code's personal skills folder; other agents read the same folder from their own skills directory.

View the SKILL.md at its source Verify the publisher's install documentation

More AI development skills →

Install the full skill folder

git clone --depth 1 https://github.com/anthropics/skills.git
cp -R skills/skills/claude-api ~/.claude/skills/claude-api

Thyme RAG Capsules

Make this skill’s state durable, governed, and portable.

A Retenta capsule lets agents recall prior work, append provenance-stamped progress, and carry the memory to the next run.

Create durable agent memoryUnderstand the evidence labels