By gotalab
Draft and activate compact /goal instructions within Claude Code to define verifiable outcomes, evidence, constraints, stop rules, and limited clarification for persistent autonomous runs with optional review and separate threads
Turn rough long-running work into a compact Codex /goal with a clear finish line.
goal-setter is a Codex skill for work that needs more than a one-off prompt. It turns a rough request into a short goal that fixes the result, the evidence for success, the boundaries, and the stop rules, while leaving implementation judgment to the runner.
It favors verification targets over rule-heavy procedures: the goal should say what must be proven, not prescribe every step.
Built for Codex. Works on Claude Code too.
Do not turn every request into a Goal. Use goal-setter when the work can keep going across several checks and the finish line must be judged by evidence.
| Work | Better format |
|---|---|
| One clear edit, explanation, or review | Normal prompt |
| Narrow work that may need a few tries | One sentence or short paragraph Goal |
| Migration, performance work, broad bug fixing, or evidence-backed research | Standard Goal |
| Long research or high-risk change | Goal plus the smallest needed plan, checklist, or evaluation file |
Details live in docs/RUNTIME.md. Examples live in docs/EXAMPLES.md.
Pick one install path:
| If you use | Install | Invoke with |
|---|---|---|
Codex App /plugins | Codex App Plugin | $goal-setter:goal-setter ... |
| Codex local skills | Codex Skill | $goal-setter ... |
| Claude Code | Claude Code marketplace | /goal-setter:goal-setter ... |
| Another agent with Skills CLI support | Skills CLI | the agent's skill invocation syntax |
Most Codex App users should install only the Codex App Plugin.
In Codex, open /plugins, choose Add plugin marketplace, and enter:
Source: gotalab/goal-setter-skill
Git ref: main
Sparse paths: plugins/goal-setter
Then install Goal Setter from the Plugins screen.
In Codex chat:
$skill-installer install https://github.com/gotalab/goal-setter-skill/tree/main/skills/goal-setter
Restart Codex, then invoke with $goal-setter.
/plugin marketplace add gotalab/goal-setter-skill
/plugin install goal-setter@goal-setter
Invoke with /goal-setter:goal-setter.
npx skills add gotalab/goal-setter-skill
Draft a goal without activating it:
$goal-setter draft a goal for migrating our API client to v2
Shape and activate a goal when Codex does not need extra worker tools:
$goal-setter set a goal: all checkout tests pass after the refactor
When the goal must launch Codex worker tools such as spawn_agent or
create_thread, goal-setter returns an exact /goal ... line for you to send.
That user-sent line is what authorizes those tools.
skills/goal-setter/SKILL.md # the skill
skills/goal-setter/scripts/ # goal length validator
scripts/ # packaging and release checks
plugins/goal-setter/ # Codex plugin bundle
.claude-plugin/ # Claude Code plugin metadata
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub gotalab/goal-setter-skill --plugin goal-setterA UX regression testing skill for browser-based and webview-based apps. Runs after E2E to catch usability, journey, accessibility, and interface-quality issues before they ship.
Durable goal-following for Claude Code: contracts with definition-of-done, subagent judge gates, executor-subagent chain execution, and adaptive missions. Inspired by OpenAI Codex /goal and the Ralph loop pattern.
Turn broad Codex and Claude Code work into pressured /goal runs with oracles, local boards, receipts, and verification.
A no-nonsense goal workflow engine for Claude Code and Codex: turn a plain-English goal into tracked steps with plain-language decisions and evidence-based done.
Tell Claude what you want once; it works until the job is verifiably done, then saves what it learned. Goals with checkable specs, independent verification, and a memory that grows.
Agent discipline engine — locks the goal, enforces PLAN-DO-VERIFY-FINALIZE-OUTPUT loop, detects 11 success masquerading patterns.
Use when user wants to execute long-running tasks that require multiple sessions to complete. This skill manages task decomposition, progress tracking, and autonomous execution using Claude Code headless mode with auto-continuation. Trigger phrases: autonomous, long-running task, multi-session, 自主执行, 长时任务, autonomous skill.