Skip to content
Provider · AI

Anthropic — Ashlr Stack Provider

Anthropic: Claude models + MCP. Key verified against the Messages API on paste. Review adapter-specific setup, credential and MCP requirements in Ashlr Stack.

AI · API key KEY-ONLY · supplied credential attachment
Anthropic — Claude models + MCP. Key verified against the Messages API on paste.
Dashboard ↗

One command

Run one provider setup from an initialized project. Its adapter may handle auth inputs, run its available credential checks, then attempt persistence through Phantom. Supabase, Neon, Turso and Vercel have source resource-creation branches; other adapters attach accounts or supplied credentials. Inspect actual results.

stack add anthropic

Or retrieve catalog matches for what you're building via stack recommend:

stack recommend "LLM provider for a coding assistant"

Auth flow

Supply the adapter's API-key inputs. Some checks only validate structure or a limited endpoint; additional credentials and manual setup may be needed.

How-to: Paste a secret key from https://console.anthropic.com/settings/keys.

Secret slots

stack add anthropic has these 1 catalog secret slot; verify which values its adapter actually materializes through Phantom:

  • ANTHROPIC_API_KEY

Use stack env export --example for an empty-key skeleton. Stack does not automatically fill env files or resolve credential placeholders. stack exec -- <cmd> forwards to installed Phantom exec; configure and verify the app separately.

Starter templates that include Anthropic

Selected template init attempts provider setup by default; add --noProvision to inspect config first:

  • stack init --template claude-agent-openai-anthropic
  • OpenAI — GPT, Realtime, embeddings. Key verified against /v1/models on paste.
  • xAI — Grok + tool use. Key verified on paste.
  • DeepSeek — Open-weight models. Key verified on paste.
  • Replicate — Open-source model inference with per-second billing. API token stored in Phantom.
  • Braintrust — LLM evals + observability + prompt playground. Verified against /v1/organization on paste.

FAQ

Do I need a Anthropic account to use it with Stack?

Use an existing Anthropic account and permissions appropriate to the adapter. Creation or attachment behavior is provider-specific. Supply the adapter's API-key inputs. Some checks only validate structure or a limited endpoint; additional credentials and manual setup may be needed. If you don't have a credential yet, create one at https://console.anthropic.com and supply the required inputs; inspect verification and persistence results.

Where does Stack store my Anthropic credentials?

Credential persistence uses the installed Phantom Secrets interface. The catalog declares slot names (ANTHROPIC_API_KEY); actual materialized outputs can differ. Stack handles supplied values in memory and records dedicated names in config; manual files and freeform metadata are not generally redacted.

Can I rotate or revoke this integration later?

Use `stack remove anthropic` for local Anthropic configuration/MCP cleanup and attempted Phantom cleanup; upstream resources remain. Cleanup can refuse or leave keys stored. Rotate or revoke the underlying Anthropic credentials through https://console.anthropic.com, then update and verify each consumer separately. Doctor does not rotate tokens and can skip already-stored names.

Outbound