Skill category
Use this category when the goal is to improve how Claude Code integrates into your development workflow — commits, reviews, parallel agents, and external tool connections.
Profiles for Git, MCP, subagents, and day-to-day Claude Code operating patterns.
Use the hub when you need a wider choice set, review methodology, or a different workflow path. Return to the skills overview
Use this profile when one Claude Code session is becoming a bottleneck and the work can be split into parallel, scoped subtasks with cleaner context boundaries.
3 source signals • Last reviewed 2026-04-01
Each subagent has its own tool permissions, so define scope and allowed actions explicitly before delegation.
Use this profile when Claude Code needs live access to databases, APIs, file systems, or operator tools through MCP and you want the connection model to stay legible.
3 source signals • Last reviewed 2026-04-01
MCP servers can have full system access, so confirm the source and permissions model before connecting one to Claude Code.
Use this profile when Claude Code is part of a real Git workflow and you need the collaboration pattern around branching, reviewing, and landing work to stay disciplined.
3 source signals • Last reviewed 2026-04-01
Avoid letting Claude Code automatically push to main or perform force pushes without an explicit human checkpoint.
Use this profile when the main goal is diff analysis: spotting regressions, missing tests, risky assumptions, or release-boundary mistakes before a change merges.
2 source signals • Last reviewed 2026-04-01
AI review is an assistive layer and cannot replace human review for security-sensitive, compliance-critical, or high-blast-radius code paths.
Use this profile when you want to author, package, and iterate on Claude Code skills instead of keeping workflow logic trapped in ad-hoc prompts. It fits teams standardizing repeatable development flows across repositories.
1 source signals • Last reviewed 2026-04-02
Review any generated skill for shell commands, file writes, and tool permissions before treating it as reusable automation.
Use this profile when you need a general Claude Code operating pattern rather than one narrow sub-workflow. It is useful for teams wiring Claude Code into daily development and repo maintenance.
1 source signals • Last reviewed 2026-04-02
Treat Claude Code hooks and connected tools as execution surfaces that can affect the local machine, not just the chat transcript.
Use this profile when the problem is less about installation and more about getting consistently strong Claude Code output. It fits operators who want better prompt framing, review habits, and session control.
1 source signals • Last reviewed 2026-04-02
Do not mistake confident model output for verified results; expert workflows still need repo-level checks and manual review.
Use this profile when you want to shape Claude Code into a repeatable build system for prompts, hooks, and project bootstrap steps. It fits teams formalizing how new repos get AI-ready defaults.
1 source signals • Last reviewed 2026-04-02
Audit any bootstrap step that writes files, installs dependencies, or edits repo config before shipping it to teammates.
Use this profile when a task is too large for one flat execution pass and needs staged planning. It is useful for operators who want Claude Code to break delivery into deliberate waves before implementation.
1 source signals • Last reviewed 2026-04-02
Do not let a wave plan justify broad permissions up front; keep each stage scoped to the files and commands it actually needs.
Use this profile when you need Claude Code to trigger structured actions around development events such as setup, validation, or workflow handoff. It is most useful when hook behavior needs to stay predictable and reviewable.
1 source signals • Last reviewed 2026-04-02
Hooks often end up running shell commands, so review every command path and environment assumption before enabling them broadly.
Use this profile when you are initializing Claude Code for a new or messy repository and need the project-level setup to become explicit. It is a good fit for teams standardizing CLAUDE.md, conventions, and baseline workflow files.
1 source signals • Last reviewed 2026-04-02
Do not copy a generic setup bundle into every repo without checking whether the project actually uses those tools, scripts, and permissions.
Use this profile when Claude Code already runs in a repo but the project-level config is weak, inconsistent, or drifting. It is helpful for teams refining hooks, instructions, and repo defaults after the first install.
1 source signals • Last reviewed 2026-04-02
Project config can indirectly authorize file edits, command execution, and hook behavior, so treat it as an operational control plane.
Use this profile when a repository already has Claude instructions but they are outdated, incomplete, or drifting from reality. It is especially useful after toolchain changes, workflow shifts, or repeated assistant mistakes.
1 source signals • Last reviewed 2026-04-02
Do not add secrets, tokens, or customer-specific data to CLAUDE.md just because the assistant benefits from more context.
Use this profile when the challenge is not creating CLAUDE.md but deciding how to format persistent working memory inside it. It fits teams that want repo instructions, plans, and context blocks to stay readable for both humans and Claude Code.
1 source signals • Last reviewed 2026-04-02
Keep working memory concise so Claude Code is not forced to parse a bloated instruction file full of stale context.
Use this profile when you need Claude Code to operate with a more opinionated execution stack for QA, review, or shipping workflows. It is a good fit for teams that want repeatable command flows instead of improvising every task from scratch.
1 source signals • Last reviewed 2026-04-02
A stacked workflow can hide a lot of shell and browser automation, so review what commands and tools it is actually allowed to run.
Use this profile when you want a general development workflow for Claude Code rather than a narrow skill or tool integration. It is useful for operators standardizing how AI-assisted coding moves from spec to implementation to validation.
1 source signals • Last reviewed 2026-04-02
Do not let a generic AI coding workflow authorize broad repo changes without narrowing the task and file scope first.
Use this profile when an OpenClaw or Claude-style workflow benefits from clearly separating context gathering from mutation steps. It is useful for operators who want safer execution in workspaces where reading and writing should not blur together.
1 source signals • Last reviewed 2026-04-02
Keep the read phase broad enough to build context, but narrow the write phase to the exact files or actions that were actually approved.
Use this profile when you are authoring or refining subagent workflows and need a repeatable way to verify them. It is useful for teams that treat subagents as reusable operational units rather than one-off experiments.
1 source signals • Last reviewed 2026-04-02
Test subagents against realistic tasks so they expose scope leaks, missing context, or confusing handoff behavior before wider rollout.
Use this profile when a repo has a repeatable flow for introducing new weapon-related data, content, or logic and you want Claude Code to follow that path consistently. It fits domain workflows where new entries should be added with the same review pattern every time.
1 source signals • Last reviewed 2026-04-02
Keep the workflow scoped to the exact files or records needed for the new entry so a domain task does not spill into unrelated changes.
Use this profile when a repo has a standard process for introducing new character-related content, data, or logic and you want Claude Code to follow it consistently. It works best for structured content systems where additions should conform to an existing pattern.
1 source signals • Last reviewed 2026-04-02
Constrain the workflow to the files and records related to the new character so the task does not widen without review.
Use this profile when implementation work naturally splits into bounded subtasks that can be delegated to subagents. It is useful for teams trying to turn parallel Claude Code execution into a repeatable development pattern.
1 source signals • Last reviewed 2026-04-02
Only delegate work with clear ownership boundaries so subagents do not trample each other’s files or assumptions.
Use this profile when the main challenge is not spawning one subagent, but coordinating several of them cleanly. It is useful for teams designing a more formal orchestration layer on top of delegated Claude workflows.
1 source signals • Last reviewed 2026-04-02
Keep orchestration rules explicit so subagents do not receive overlapping ownership or conflicting task definitions.
Use this profile when a team has a named internal workflow and wants Claude Code to follow it consistently across tasks. It is useful when the value comes from reusing a house process rather than inventing a new prompt each time.
1 source signals • Last reviewed 2026-04-02
A named workflow can become trusted too easily, so re-check its assumptions whenever the repo or tooling changes.
Use this profile when you want to create new subagents more systematically instead of hand-authoring each one from scratch. It is useful for teams scaling a subagent library and needing a more repeatable generation flow.
1 source signals • Last reviewed 2026-04-02
Generated subagents should never be trusted as-is; review their instructions, tool scope, and failure modes before reuse.
Use this profile when delegated implementation work should stay tied to a stronger specification or planning layer. It is useful for teams combining subagent execution with a more formal delivery method.
1 source signals • Last reviewed 2026-04-02
Keep the specification boundary visible to every delegated agent so subagents do not optimize against their own local interpretation.
Use this profile when subagent quality problems come from context overload, missing inputs, or fuzzy task boundaries. It is useful for teams that already use delegated workflows and now need sharper context discipline.
1 source signals • Last reviewed 2026-04-02
Do not pass broad repo or secret context to every subagent by default; include only what the delegated task truly needs.
Need broader discovery? Skills overview
Need OpenClaw context? OpenClaw workflow