From hindsight-cc
First-run setup wizard: configure the Hindsight LLM provider and write the config file
How this command is triggered — by the user, by Claude, or both
Slash command
/hindsight-cc:setupThis command is limited to the following tools:
The summary Claude sees in its command listing — used to decide when to auto-load this command
# Hindsight Setup Skill ## When to Use This Skill Run this once on first use of the plugin (or any time you want to change the LLM provider Hindsight uses for its memory operations). It walks you through choosing a provider and model, then writes a config file that the container-startup script reads when the Hindsight server is next created. ## How to Execute Perform the following steps IN ORDER. This is an interactive wizard, so ask each question and wait for the answer before moving on. ## Steps ### 1. Check Docker availability Run a quick check that Docker exists and the daemon is...
Run this once on first use of the plugin (or any time you want to change the LLM provider Hindsight uses for its memory operations). It walks you through choosing a provider and model, then writes a config file that the container-startup script reads when the Hindsight server is next created.
Perform the following steps IN ORDER. This is an interactive wizard, so ask each question and wait for the answer before moving on.
Run a quick check that Docker exists and the daemon is reachable:
command -v docker && docker info >/dev/null 2>&1 && echo "docker: ok" || echo "docker: unavailable"
If Docker is missing or the daemon is down, clearly WARN the user that the plugin needs Docker to run the Hindsight server, and that they should install or start Docker before they expect memory features to work. Do NOT hard-fail: continue the wizard so the config is still written and ready for later.
Use AskUserQuestion to ask which provider Hindsight should use. Offer these
options:
Note that Ollama and LM Studio run a model locally on the user's machine; the other four are cloud providers that need an API key.
Ask which model to use, presenting a sensible default for the chosen provider. Make clear these are suggestions the user can override with any model their provider supports:
gpt-5-nanoclaude-haiku-4-5gemini-2.0-flashllama-3.3-70b-versatilellama3.1Only for Ollama and LM Studio, ask for the LLM base URL with these suggested defaults:
http://localhost:11434/v1http://localhost:1234/v1Skip this question entirely for cloud providers (OpenAI, Anthropic, Gemini, Groq).
Only for OpenAI, Anthropic, Gemini, and Groq, ask for the provider API key. Treat this value as sensitive: do NOT echo it back in plaintext in any summary or confirmation. Skip this question entirely for local providers (Ollama, LM Studio).
Write ~/.config/hindsight-cc/config.env. Do this carefully so a secret is
never exposed in shell history or a process listing any more than necessary:
First create the directory with Bash:
mkdir -p ~/.config/hindsight-cc
Write the file contents with the Write tool (NOT with echo, cat, or a
heredoc that would put the API key inline on a command line). Resolve the
path to an absolute path first (e.g. $HOME/.config/hindsight-cc/config.env).
The format is plain KEY=value, one per line, no surrounding quotes. Write
ONLY the keys that apply to the chosen provider:
HINDSIGHT_API_LLM_PROVIDER and HINDSIGHT_API_LLM_MODELHINDSIGHT_API_LLM_API_KEY only for cloud providersHINDSIGHT_API_LLM_BASE_URL only for local providersOmit any inapplicable key entirely — the parser treats an absent key as unset and falls back to the built-in default or leaves it empty.
Example for a cloud provider (OpenAI):
HINDSIGHT_API_LLM_PROVIDER=openai
HINDSIGHT_API_LLM_MODEL=gpt-5-nano
HINDSIGHT_API_LLM_API_KEY=sk-...
Example for a local provider (Ollama):
HINDSIGHT_API_LLM_PROVIDER=ollama
HINDSIGHT_API_LLM_MODEL=llama3.1
HINDSIGHT_API_LLM_BASE_URL=http://localhost:11434/v1
Tighten permissions and confirm the file was written, with Bash:
chmod 600 ~/.config/hindsight-cc/config.env && ls -l ~/.config/hindsight-cc/config.env
Show the ls -l output to confirm the file exists with 0600 permissions.
Do NOT print the file's contents (it may contain an API key).
Tell the user what happens next:
docker rm -f hindsight — or simply start a fresh Claude Code
session. On the next create, ensure-hindsight.sh picks up the new config./setupInitializes or resumes project setup via interactive Q&A, creating conductor/ artifacts for product definition, guidelines, tech stack, workflow, and style guides.
/setupWalks an enterprise admin through configuring the Claude Office add-in to call their own cloud (Vertex, Bedrock, Foundry, or gateway), producing a customized manifest.xml for M365 deployment.
/setupChecks local Codex CLI readiness, prompts to install if unavailable via npm, and optionally toggles stop-time review gate.
/setupInteractive setup wizard that detects installed AI providers, configures authentication, and optimizes token usage. Auto-runs on first install and surfaces a status dashboard on manual invocation.
npx claudepluginhub gcswan/hindsight-cc --plugin hindsight-cc