What does this skill do, and when should you use it?
using-agent-skills is the meta-skill in the Agent Skills collection that helps AI coding agents identify the appropriate skill for a task. It maps work to development phases including define, plan, build, verify, review, and ship. It also establishes shared operating rules such as surfacing assumptions, managing confusion, enforcing simplicity, maintaining scope discipline, and requiring verification evidence. It does not implement code itself; it directs agents to select and follow other skills.
Reads the incoming task description, identifies its development phase, and selects the corresponding skill from the collection; provides a typical multi-skill lifecycle for larger features; requires assumptions to be stated before non-trivial implementation, pauses on conflicts, discourages unrelated changes, and requires evidence such as passing tests, build output, or runtime data before completion.
- A developer starting a coding session and unsure which workflow skill applies.
- A developer handling a new feature, bug, test failure, or release task who needs a fast workflow mapping.
- A team that wants agents to state assumptions, surface conflicts, and keep changes within scope before coding.
- A project that needs to connect specification, planning, implementation, testing, review, and release into a coherent process.
How do you install this skill?
- The skill requires verification for every skill but does not define recovery when a skill is missing, triggers conflict, or a referenced file is unavailable.
- Routing conditions are broad and overlapping, creating a risk of incorrect semantic invocation; users should confirm the applicable phase and objective.
- The skill guides agents through code changes and release-related work without explicitly defining permissions, sensitive-data handling, external-system effects, or rollback requirements.
- No Chinese localization is provided, and compatibility with mainland-China network conditions is not demonstrated.
Install the full repository collection with: npx skills add addyosmani/agent-skills. The README also documents the individual-skill pattern: npx skills add addyosmani/agent-skills --skill using-agent-skills. For Codex CLI, the repository documents: codex plugin marketplace add addyosmani/agent-skills.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/addyosmani/agent-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/using-agent-skills" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Based on this task, identify which Agent Skills apply and list your assumptions first.
Trigger using-agent-skills at the start of a session or whenever you need to determine the applicable skill. Example prompt: "Based on this task, identify which Agent Skills apply and list your assumptions first." Then invoke the mapped skills. The required companion skills must already be installed or available to the agent.
What are this skill's strengths and limitations?
- Provides a clear mapping across define, plan, build, verify, review, and ship phases.
- Makes assumption handling, conflict management, simplicity, scope control, and verification explicit operating rules.
- Useful for establishing a consistent engineering workflow at session start.
- It is a navigation and governance meta-skill, not a direct coding, testing, or deployment skill.
- Its practical value depends on the other mapped skills being installed and callable.
- The supplied material does not document an independent test suite or platform-specific validation for this individual skill.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Engineering Workflow Navigator this page | 46 · Use with care | ★ 107k | 13d ago | MIT |
| Ask Matt Skill Router | 50 · Use with care | ★ 291k | 18d ago | MIT |
| FableCodex Evidence-Gated Workflow Skill | 63 · Recommended | ★ 436 | 4mo ago | AGPL-3.0 |
| Make Plan | 51 · Use with care | ★ 101k | 13d ago | Apache-2.0 |
| Autonomous Agent Harness | 49 · Use with care | ★ 18k | 15d ago | MIT |
How did FollowSkills review this skill?
The skill requires explicit assumptions, stopping on ambiguity, surfacing tradeoffs, scope discipline, and verification, providing some user-confirmation and safety structure. It does not define permission boundaries, data flows, sensitive-data handling, external side effects, dependency security, or rollback, so points are deducted.
The phase-to-skill routing table, shared operating rules, and failure-mode list make the happy path reasonably clear. However, the workflow depends on other skills and references not supplied here and lacks diagnosable handling for missing skills, discovery failures, or abnormal inputs. The static cap also prevents a score above 10, so 7 is warranted.
The intended audience, lifecycle scenarios, and primary triggers are fairly clear and broad. Trigger overlap remains significant, non-fit boundaries and input/output contracts are absent, and Chinese-language or mainland-China environment support is not addressed, so points are deducted.
The file has a stable name and description and is organized around overview, rules, lifecycle, and quick reference sections. Repository context supplies installation guidance, an MIT license, maintainer clues, and contribution guidance, but the selected skill lacks versioning, changelog, troubleshooting, and explicit update ownership, so it does not receive full marks.
The routing table and shared behavior rules could reduce omissions of specifications, tests, and verification in agent workflows. Actual value depends on the other skills and correct semantic triggering, while no representative directly usable outputs or independent outcome evidence are provided. The static cap supports only 6 points.
Repository context includes CI workflows and commands for validating and evaluating skills, showing an intent to automate checks. The supplied test fixture does not directly cover this meta-skill's routing paths, so key-path reproduction and independent corroboration remain unproven; 3 points is appropriate.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →