Subagent Development

ReviewAudited by ClawScan on May 1, 2026.

Overview

The provided artifacts describe a coherent instruction-only development workflow, with the main caveat that it delegates code changes and commits to subagents.

This skill appears coherent and instruction-only. Before installing or using it, make sure you are comfortable letting subagents edit, test, and commit code for the selected plan; use a feature branch, review diffs, and verify any referenced external skills or unpinned install sources.

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.

What this means

If invoked on the wrong plan, branch, or repository, the agent could make multiple code changes before you review them.

Why it was flagged

This shows the workflow intentionally delegates code-changing and commit actions to subagents and proceeds between tasks without a human checkpoint.

Skill content
"Faster iteration (no human-in-loop between tasks)" ... "Implementer subagent implements, tests, commits, self-reviews"
Recommendation

Use it on a feature branch, keep the plan tightly scoped, and require diff or commit review before accepting the results.

What this means

Following that install command later could fetch content that differs from the reviewed artifacts.

Why it was flagged

The README includes a user-directed install path from a mutable GitHub branch rather than a pinned version or commit.

Skill content
npx add https://github.com/wpank/ai/tree/main/skills/meta/subagent-development
Recommendation

Prefer the registry install or pin the source to a reviewed version or commit before installing from GitHub.

What this means

The behavior of the full workflow can depend on separately installed reviewer skills or templates.

Why it was flagged

The code quality review step depends on an external skill/template that is not part of the provided artifact set.

Skill content
Task tool (superpowers:code-reviewer):
  Use template at requesting-code-review/code-reviewer.md
Recommendation

Review referenced skills/templates separately before relying on them in the same workflow.

What this means

Sensitive information included in task descriptions or context may be shared with subagents during the workflow.

Why it was flagged

The workflow passes task text and context to a general-purpose subagent as part of its normal operation.

Skill content
Task tool (general-purpose): ... [FULL TEXT of task from plan - paste it here, don't make subagent read file]
Recommendation

Keep task context limited to what the subagent needs and avoid including unrelated secrets or private data in plans.