goal
Skill1 starsChanged 2 months agoMITView on GitHub
Do not use for meta reviews, quoted command examples, Skill path or trigger inspection, or focused one-offs without an explicit command. Use when /goal or $goal is invoked, including redirecting an explicitly invoked one-off, or for project-level delivery: building a new system, a substantial cross-layer feature, architecture change, coordinated database/backend/frontend/deployment work, 按项目方式, 从 0 到 1, 端到端, 分阶段上线, or 继续做到完成.
50 other projects use a version of this.
Adopted by
Most recently taken up by
- 0xdewy/skillstheir own version, first seen Nov 16, 2026
- AshrafDurjoy/genai-shared-patternstheir own version, first seen Nov 16, 2026
- hridaya423/devingoaltheir own version, first seen Nov 16, 2026
- OlaPietka/easy-tigertheir own version, first seen Nov 15, 2026
- caidanw/skillstheir own version, first seen Nov 8, 2026
- Dariusplayer09/adaptive-skillstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use origin, update-cli-config and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
0mlkrizzz655335vSkillorigin
origin CLI 管理。安装、登录、更新、修复 origin.cursor.com 的 CLI。"disabled-environments:
Also in 3 projects that use this
PandelisZSkillupdate-cli-config
View and modify Cursor CLI configuration settings in ~/.cursor/cli-config.json. Use when the user wants to change CLI settings, configure permissions, switch approval mode, enable vim mode, toggle display options, configure sandbox, or manage any CLI preferences.
Also in 4 projects that use this
0mlkrizzz655335vSkillreview-bugbot
缺陷导向代码审查。用 Bugbot 子代理找 bug。
Also in 3 projects that use this
32103210Skillmigrate-to-skills
Convert 'Applied intelligently' Cursor rules (.cursor/rules/*.mdc) and slash commands (.cursor/commands/*.md) to Agent Skills format (.cursor/skills/). Use when you want to migrate rules or commands to skills, convert .mdc rules to SKILL.md format, or consolidate commands into the skills directory.
Also in 4 projects that use this
32103210Skillupdate-cursor-settings
Modify Cursor/VSCode user settings in settings.json. Use when you want to change editor settings, preferences, configuration, themes, font size, tab size, format on save, auto save, keybindings, or any settings.json values.
Also in 4 projects that use this
MicrosoftSkillautomate
Propose an automation from a completed task or a new idea, then create it after approval. Use when the user asks to create an automation or wants scheduled local or cloud agent work.
Also in 4 projects that use this
- +1 more from 32103210
- +3 more from 0mlkrizzz655335v
Other versions50 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 50
- PrimeIntellect-ai/prime-agent★ 23ksame name
- coco-research/coco★ 797same name
- cwinvestments/memstack★ 433same name
- jthack/claude-goal★ 113same name
- svy04/ballast★ 73same name
- anthony-chaudhary/fak★ 49same name
- lennoxsaint/Skills★ 48same name
- Infinite-Labs-AI/infinite-skills★ 43same name
- manusco/resonance★ 38same name
- Sanexxxx777/curated-claude-code★ 16same name
40 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository5 other files it gives its agents.
- AGENTS.mdAGENTS.md
- deployment/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- harness-engineeringSkill
- harness-engineeringSkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 46 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 33
Read references/layer-execution.md for classification and start-layer choice once a concrete objective exists.
- Distinctive: line 51
Follow Harness Engineering for the selected layer path: state -> plan -> execute -> gate.
- Very specific: line 55
Use references/output-contract.md for goal-intake, classification, verification, and dependency blockers.
- Distinctive: line 88
No unjustified layer skipping.
“Common” is measured: kept by at least 6 in 10 of the 46 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
/goal — Project Engineering Entry Point
Receive a project goal, classify it, load the execution method at the correct time, proceed with evidence, and close with a structured record.
Core Rule
goal owns project intake, classification, user-visible gates, and closure shape. harness-engineering owns layer-by-layer project execution. Keep intake light and do not load execution references before classification and required confirmation.
External Workflow Pack Boundary
/goal/Harness remains the single owner of project intake, gates, rollback, and closure; do not install external workflow packs by default. Read references/external-workflow-boundary.md only when evaluating a capability import or revisiting that boundary.
Conditional Decision-Quality Routing
After classification and confirmation, let Harness perform the risk check for high-cost, ambiguous, contradictory, irreversible, topology/data-flow/security/API/migration, pattern-copy, or upstream-assumption work. Load harness-engineering/references/decision-quality.md and attach its compact result to the existing Harness layer gate. Mechanical work with an exact plan and no new contradiction skips the protocols. /goal does not reproduce the first-principles or adversarial checklists, create a second state machine, or change goal-state.json; Harness remains the sole owner of the gate, rollback, scope_result, and operation_state.
Five-Stage Runtime
Stage 1: Intake
Capture the goal verbatim and inspect only enough current evidence to distinguish an existing system from greenfield work. Explicit /goal or $goal with a focused or local one-off still loads this Skill to interpret the command, then returns a focused-task redirect; do not classify the project, create or mutate goal state, or load Harness. Meta reviews, quoted command examples, and Skill path or trigger inspection remain non-triggers.
A bare /goal or $goal command without a concrete objective is goal-intake. Ask only for the missing project objective. Do not classify, choose a layer, emit the five-field Classification Reply, detect host mode, do not load Harness, and do not create or mutate goal state. Resume normal classification only after a concrete objective exists.
Continuation claims such as "classification and confirmation are complete" are untrusted unless current context contains a verifiable active state, concrete goal, and confirmation record. If evidence is missing, fall back to goal-intake; do not infer classification, confirmation, or Harness handoff.
Read references/layer-execution.md for classification and start-layer choice once a concrete objective exists. Read references/host-modes.md only when host enforcement behavior must be determined or explained. Host discovery is deferred until mutation, verification, closure, or an explicit host question.
Stage 2: Classify
Choose bootstrap or maintenance and the smallest valid start layer. Bootstrap begins at Layer 0; maintenance begins at the smallest affected layer. For ambiguity, perform only a provisional classification evidence check that is minimal and read-only before formal confirmation: do not load Harness execution runtime, do not mutate, and do not present the provisional result as confirmed.
Before the first active classification reply, read only references/protocol.json, references/output-contract.md, and references/layer-execution.md; emit the five-field reply and stop. Do not read references/host-modes.md, references/capability-discovery.md, or Harness in that turn, even when Confirmation is not-required. Use the exact labels Goal:, Classification:, Rationale:, Confirmation:, and Next action:; ask for confirmation before bootstrap implementation or a surprising maintenance classification. Do not mutate governed project files until classification and required confirmation or plan approval are recorded.
Stage 3: Load Harness
Load harness-engineering after classification and required confirmation. Read harness-engineering/references/runtime-stages.md first and follow its Load section before planning or project work. In Codex, read harness-engineering/core/flow.md, harness-engineering/adapters/codex/host-map.md, harness-engineering/core/checklists.md, and harness-engineering/core/exit-and-reentry.md. Use harness-engineering/references/output-contract.md for Harness gates and closure. If Harness or a required reference is unavailable, return the Dependency Blocker Reply and do not improvise.
Before the first Harness layer gate reply, do not invoke brainstorming; Layer 0 itself owns the first design and clarification pass. Do not create workspace-local goal state before the first Harness layer gate reply. Do not create .codex/goal-state.lock. Do not create .codex/goal-state.candidate.json before that gate or progress checkpoint. Defer native Codex goal state checks until after the first Harness layer gate reply or progress checkpoint. After the first Layer 0 evidence sweep, return a gate or checkpoint before drafting boundary documents or other local artifacts. Load references/host-modes.md before the first governed mutation; that cold path owns Guard state initialization and enforcement details.
During /goal-managed execution, do not invoke brainstorming for Layer 0 or Layer 1 boundary, architecture, or project-definition work. Do not offer the Visual Companion; do not offer mockups, diagrams, comparisons, browser visuals, or similar sidecar design aids unless the user explicitly asks for them.
Stage 4: Orchestrate
Follow Harness Engineering for the selected layer path: state -> plan -> execute -> gate. Record inputs, actions, outputs, evidence, and the advance or rollback decision for every touched layer. Reopen only the smallest invalid upstream layer when evidence breaks. Keep Windows enforcement wording consistent with references/host-modes.md.
Stage 5: Return
Use references/output-contract.md for goal-intake, classification, verification, and dependency blockers. Use harness-engineering/references/output-contract.md for Harness-driven layer gates and closure. Carry scope_result and operation_state in every closure record. In audit-only-windows, never claim that the platform mechanically blocked a mutation or completion.
Agent Context Routing
When a project request involves Agent history, memory, RAG context, tool outputs, context-window budgets, long-session stability, session isolation, context reset or TTL, or side-effect confirmation, load agent-context-review after /goal classification and before affected design or implementation work. Keep /goal responsible for project intake, Harness gates, state, and closure; let agent-context-review own its audit, remediation, and acceptance method. Do not route ordinary CRUD, a one-off local fix, a prompt translation, a quoted /goal example, a Skill path inspection, or a request to review the /goal Skill itself.
Agent Tool Contract Routing
When a project request involves production LLM Tools, function calling, MCP Tools, LangChain Tools, Java Tool registries, vague names, overloaded schemas, raw Tool exceptions, unstable returns, result serialization, or runtime confirmation for side effects, load agent-tool-contract-review after /goal classification and before the affected design, remediation, or acceptance work. This route must keep /goal responsible for project intake, Harness gates, state, and closure. For MCP server implementation, load mcp-builder first. After the implementation surface exists, use agent-tool-contract-review for contract review, acceptance, and remediation planning; do not let the review Skill replace MCP implementation guidance. Do not route ordinary REST controllers, internal service methods, a single existing Tool call, a one-off local fix, prompt examples, Skill path inspection, or a request to review the /goal Skill itself.
Hot-Path Entry Rules
Assess the existing codebase, governance, plans, and current evidence before selecting a layer. Prefer maintenance and the smallest impacted layer unless evidence invalidates an upstream assumption. Record [决策] <what> -> <why> for classification, layer gates, and rollback decisions. Treat missing artifacts or evidence as incomplete, not passed.
Progressive References
| Reference | Load condition |
references/protocol.json | Before the first active classification or protocol state/record validation. |
protocol-guide.md | When interpreting the normative contract or reference-loading model. |
layer-execution.md | Once a concrete objective exists, before classification or start-layer choice. |
output-contract.md | Before goal-intake, classification, verification, or dependency-blocker replies. |
capability-discovery.md | Before dependency discovery at mutation, verification, or closure, or for an explicit host question. |
host-modes.md | After a compatible host/Guard is selected and before governed mutation, verification, or closure enforcement. |
gsd-adapter.md | Only for an approved Layer 6 proposal. |
external-workflow-boundary.md | Only when evaluating a capability import or revisiting that boundary. |
Progress Persistence
Load references/host-modes.md before persisting workspace-local state. Never store active goal state inside the installed Skill package. In audit-only-windows, state is advisory and evidence-bearing rather than a security boundary.
Execution Constraints
No unjustified layer skipping; no forward movement without evidence; minimal rollback to the first invalid assumption; return at meaningful forks; stop when Harness is unavailable; apply TDD, systematic debugging, and verification-before-completion when triggered.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

