memory-manage_skill

This skill orchestrates CLAUDE.md updates and API/documentation generation by routing inputs and prompting interactive needs assessment.
  • TypeScript

1.3k

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill catlog22/claude-code-workflow --skill memory-manage

  • SKILL.md7.8 KB

Overview

This skill provides a unified entry point for managing project memory (CLAUDE.md) updates and generating project documentation (API.md / README.md). It parses user input or runs an interactive needs assessment, then routes to focused phases for full, incremental, or single-module operations. The skill supports tool fallbacks and batching to scale across module counts.

How this skill works

The router inspects input tokens and keywords to auto-dispatch to sub-commands: update-full, update-related, update-single, docs-full, or docs-related. If input is ambiguous or missing, it runs a short interactive questionnaire to determine content type and scope, then reads and executes the referenced phase document on-demand. Remaining CLI arguments are forwarded unchanged and the chosen sub-command handles plan presentation and confirmation.

When to use it

  • You need a full CLAUDE.md refresh after a major refactor or architecture change.
  • You want incremental updates for only git-changed modules during everyday development.
  • You need a deep, single-module handbook-style CLAUDE.md entry for a specific path.
  • You want to generate or update API.md and README.md for the whole project (initialization).
  • You want incremental docs generation limited to changed modules to keep docs in sync without full rebuild.

Best practices

  • Prefer incremental (related) mode for daily workflows to reduce noise and runtime.
  • Provide explicit sub-command tokens when scripting to avoid interactive prompts (e.g., update-full, update-single).
  • Use the --tool flag to choose primary CLI tool; rely on the built-in fallback order for resilience.
  • Keep repository paths and git state clean so the router can accurately detect changed modules.
  • Rely on phase documents for execution detail; the router only dispatches and reads the selected phase on demand.

Example use cases

  • CI job: run docs-related after each merge to update API.md for changed modules only.
  • Developer CLI: memory-manage -> update-single src/auth to generate a detailed CLAUDE.md for the auth module.
  • Post-refactor maintenance: memory-manage update-full --tool qwen to rebuild CLAUDE.md across the codebase.
  • Local workflow: run memory-manage with no args to trigger the interactive questionnaire and pick between CLAUDE.md or project docs.
  • Automation pipeline: schedule docs-full daily during major releases to regenerate README and ARCHITECTURE files.

FAQ

The router prompts for a module path in the interactive flow (Q3). You can choose examples or enter a custom path.

How does tool fallback work when a primary tool fails?

The skill follows the configured hierarchy (e.g., --tool gemini → [gemini, qwen, codex]) and automatically falls back to the next tool in order.

Will the router read all phase documents before executing?

No. The router reads only the phase document for the selected route on demand; it does not preload all phase files.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational