From gsd
Captures authentication recipes for external systems (GitHub, AWS, npm, SSH) so future sessions know how to authenticate. Supports manual capture and auto-detection from Bash commands.
How this skill is triggered — by the user, by Claude, or both
Slash command
/gsd:remember-access [--review] [<system-name>][--review] [<system-name>]This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
<context>
Two modes:
/gsd:remember-access <system-name>: manual capture. Walks you through documenting how to authenticate to a system, writes to .planning/AUTH-RECIPES.md (per-project), optionally promotes to ~/.claude/auth-recipes/<system>.md (cross-project)./gsd:remember-access --review: surface the auto-detection inbox (.planning/.pending-auth-captures.jsonl, populated by the gsd-auth-detector PostToolUse hook), let you confirm each detection and save it as a recipe.Auto-detection runs continuously: every Bash command you execute is checked against auth-shaped patterns (gh auth login, aws configure, ssh-keygen, export GITHUB_TOKEN=..., etc.). Secret-looking content is redacted before the inbox entry is written. Review at your own pace.
Privacy: the detector NEVER stores the actual secret values. It stores the SHAPE of the command (which tool, which flags) with credential payloads replaced by [REDACTED] markers. The recipe is "how to authenticate," not "what the password is."
<execution_context> @${CLAUDE_PLUGIN_ROOT}/workflows/remember-access.md </execution_context>
Execute the remember-access workflow from @${CLAUDE_PLUGIN_ROOT}/workflows/remember-access.md.npx claudepluginhub jnuyens/gsd-pluginProvides Python patterns for external service authentication using API keys, OAuth, tokens. Includes verification flows, smoke tests, env checks, and error handling with leyline.
Guides claude-mem plugin usage for persisting context, work history, and key findings across Claude Code sessions. Integrates with fireauto skills for research, PRD, SEO, and security tracking.
Writes an explicit entry to auto-memory when something is important enough that you don't want to rely on Claude noticing it automatically.