Superclaw
PassAudited by ClawScan on May 1, 2026.
Overview
Superclaw is a coherent instruction-only development workflow skill, but users should expect it to enforce its process, use workspace memory, and spawn subagents for planned implementation work.
Use Superclaw if you want a stricter software-development process with design approval, written plans, batched execution, and progress memory. Before installing, be comfortable with the agent reading workspace memory/context files, writing plan and progress documents, and spawning subagents to carry out planned file changes.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may refuse to skip design, planning, or checkpoint steps even when you ask it to move faster.
The skill deliberately overrides requests to skip checkpoints. This affects user control, but it is prominently disclosed and matches the stated purpose of enforcing a disciplined workflow.
DO NOT execute tasks all at once, even if user says "no interruptions"
Install only if you want this stricter workflow; otherwise use the skill selectively or disable it for tasks where you need direct execution.
Approved implementation plans may modify or delete project files.
The plan template explicitly supports creating, editing, and deleting files. That is expected for a software development skill, but it means the generated plan can materially change the workspace.
**Files:** - Create: path/to/new/file.ext - Edit: path/to/existing/file.ext - Delete: path/to/old/file.ext
Review generated plans for exact file paths and keep normal source-control or backup practices in place before execution.
Project progress and some contextual details may be stored for future sessions and later reused by the agent.
The skill reads persistent memory/context files and writes progress back to daily memory logs. This is disclosed and supports resumability, but persistent context can contain sensitive preferences or project details.
Before starting, read: - `MEMORY.md` (if main session) - `memory/YYYY-MM-DD.md` (today's log) ... Don't wait until the end - update after each batch
Avoid using it with sensitive projects unless you are comfortable with progress being written to workspace memory; periodically review memory logs for unwanted details.
Subagents may receive enough project context to create or modify files for each task.
The skill routes individual implementation tasks through spawned subagents. This is disclosed as an isolation mechanism, but task prompts may include file paths, commands, and project context.
Spawn fresh subagent for task execution: ``` sessions_spawn --label "task-N-<short-desc>" --prompt "Execute task N from plan..."
Ensure generated subagent prompts contain only the context needed for the task and review batch outputs before continuing.