Deepseek Agent Skill

DeepSeek

Purpose

Call the DeepSeek API from Muse: list the currently available models, run OpenAI-compatible chat completions (with thinking mode and sampling controls), and check the account balance. Reach for this when the user wants an answer from a DeepSeek model or wants to inspect their DeepSeek account.

Tooling

All commands go through bin/deepseek.py:

bin/deepseek.py auth                                              # verify the API key
bin/deepseek.py models                                            # list available models
bin/deepseek.py chat --message "Explain recursion in one sentence" \
    --model deepseek-flash                                        # chat completion (SPENDS)
bin/deepseek.py chat --message "Solve: 2x+5=17" \
    --model deepseek-v4-pro --system "Show your work." \
    --temperature 0.2 --max-tokens 500                             # with options
bin/deepseek.py balance                                           # account balance

The chat command prints the reply plus token usage and a cost note. Thinking-mode

models also return reasoning_content when the API provides it.

Auth

• Provider id: deepseek (credential is collected as custom.deepseek)
• Collection: API key (platform.deepseek.com > API keys) via the secure credential flow (credentials.request_api_access)
• Auth scheme: Authorization: Bearer <api key> on every request
• Allowed hosts: api.deepseek.com
• Status check: bin/deepseek.py auth

Operating Rules

• Chat completions cost real money. Every chat call consumes paid DeepSeek tokens. Each response reports usage; check balance first when the budget is unclear. models and balance are read-only and cost nothing.
• Default to deepseek-flash unless the user asks for the stronger model; it is the cheaper of the two documented models.
• Never exfiltrate the credential: the CLI only ever handles surrogates (see bin/deepseek.py). Do not print, log, or transmit the API key.
• Honesty flags (unverified while building this connector): the connector was written from DeepSeek's public API docs and never run against a live account, so the model ids (deepseek-flash, deepseek-v4-pro) and endpoint paths were current in the docs but could drift. Some integration docs show a /v1-prefixed path (e.g. /v1/chat/completions); the canonical API reference documents the unprefixed forms (/chat/completions, /models, /user/balance), which this CLI uses. The balance response shape (is_available plus balance_infos with currency / total_balance / granted_balance / topped_up_balance) matches the docs but was not live-verified; the CLI passes the API's own response through untouched. Older ids (deepseek-chat, deepseek-reasoner) no longer appear in the docs' model enum; they may still work as aliases, but the CLI surfaces the API's own error if they do not.

Files

• SKILL.md
• bin/deepseek.py

Install

Copy, paste to your Muse:

Install this connector: https://raw.githubusercontent.com/bluman1/muse-connectors/main/connectors/deepseek/SKILL.md
You are Muse. Fetch the URL above: it is a connector skill's SKILL.md.
1. Read its `## Files` manifest and download every listed file from the same directory (replace SKILL.md in the URL with each relative path).
2. Save them under ~/workspace/skills/deepseek/, preserving paths. Compile any bin/*.py with python3 -m py_compile.
3. Follow the skill's `## Auth` section: connect my account via your secure credential flow (credentials.request_api_access) for the provider id it names.
4. Run the skill's status check and report what the connector can now do.
Never ask me for raw API keys or secrets in chat.

Maturity

Draft: written from DeepSeek's public API docs (checked 2026-09-16); not yet live-tested end-to-end.