define-goals
Skill1 starsChanged 3 months agoMITView on GitHub
Pins down what a new project or initiative is for: the objective, the stakeholders, how success is measured, and the guardrails. Use when a new project or initiative starts before scoping or building, or when the user wants to build X with no stated objective, or when the objective is vague or missing. Skip a single feature, and skip while a decision or approval reply is pending.
In this repository49 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- working-from-a-ticketSkill
- writing-specsSkill
- clarifying-intentSkill
- dispatching-parallel-agentsSkill
- handoffSkill
- mapping-the-codebaseSkill
- prototypingSkill
- using-git-worktreesSkill
- using-sdlc-skillsSkill
- viewing-artifactsSkill
- writing-skillsSkill
- yagniSkill
- finishing-a-branchSkill
- release-readinessSkill
- architecture-decisionsSkill
- coding-standardsSkill
- data-modelSkill
- migration-strategySkill
- reviewing-plansSkill
- system-architectureSkill
- ui-ux-designSkill
- writing-plansSkill
- adding-observabilitySkill
- changing-a-live-schemaSkill
- executing-plansSkill
- rollout-controlsSkill
- subagent-driven-developmentSkill
- test-driven-developmentSkill
- complexity-auditSkill
- containing-an-incidentSkill
- debuggingSkill
- deprecating-an-apiSkill
- diagnosing-a-sessionSkill
- eliminating-correctionsSkill
- large-scale-changesSkill
- post-mortemSkill
- refactor-architectureSkill
- brainstormingSkill
- feasibility-checkSkill
- scopingSkill
- receiving-code-reviewSkill
- requesting-code-reviewSkill
- reviewing-a-prSkill
- security-auditsSkill
- verification-before-completionSkill
- verification-strategySkill
- visual-ui-verificationSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Before scope or design, name the outcome, and how you will know you reached it.
- Very specific: line 18
Skip: when a drafted goal or brief is awaiting a direct answer.
- Distinctive: line 26
Ask "why this, why now?" until the answer is what the world does differently afterward, not code that exists.
- Distinctive: line 38
Write the guardrails and the failure criterion.
- Distinctive: line 39
Write the value in one sentence.
- Very specific: line 43
Write the ## Goals section, which is never edited once its identity is issued, to .sdlc-skills/briefs/{{YYYY-MM-DD}}-{{topic}}.md or the …
- Very specific: line 64
Ask this and end the turn.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Define Goals
Before scope or design, name the outcome, and how you will know you reached it.
When to use
- Starting a new project, product, or substantial initiative with a fuzzy "why".
- Skip for a single feature or task. Use
clarifying-intentonly for genuine ambiguity about which initiative or procedure is intended;writing-specsowns settled, detailed feature requirements. - Skip when a drafted goal or brief is awaiting a direct answer. If the
reply informs but does not answer, handle it with
clarifying-intent, not with a new goal pass.
Step 1: Find the objective
Open assets/goals-section.md before starting; each step fills its section.
- Ask "why this, why now?" until the answer is what the world does differently afterward, not code that exists.
- Name who benefits, who operates or bears risk, what changes for each, and who owns the outcome. If goals compete, give each its own metric, never one averaged metric.
- Record one accountable decision owner, or else the approvers, the conflict
resolver, and the decision rule. If the user is the only stakeholder (a
solo owner), write the user as the decision owner in
Decision rule:and name no approvers or conflict resolver. - For each outcome, record its current baseline, target, time horizon, measurement source, and accountable owner. If an outcome has no source or date, it cannot be checked, so it is not a goal.
- Write the guardrails and the failure criterion.
- Write the value in one sentence.
Step 2: Write and present
- Write the
## Goalssection, which is never edited once its identity is issued, to.sdlc-skills/briefs/{{YYYY-MM-DD}}-{{topic}}.mdor the user-set path, leaving the approved sections around it untouched. When the work came from a captured ticket, fillSourcewith the capture's path and identity. Present:
Goals {{path}} — version {{identity (per template)}} Objective: {{one line}} Measure: {{baseline}} → {{target}} by {{horizon}}, source {{source}} Guardrails: {{list}} 1. Approve the goals 2. Request changes 3. Reject the objective 4. Cancel Recommendation: {{option the open assumptions support}} — {{one sentence}}.Ask this and end the turn;
clarifying-intentowns what closes it.- REQUIRED SUB-SKILL: on option 1, invoke
feasibility-checkunless an existing feasibility section already fills every dimension row from evidence and no row is unknown; then, if the boundary is next, invokescoping. Never impose a phase that is already complete.
Gotchas
- A guardrail can look complete just because it is written down while no
observation could ever show it broken.
scopingtreats every goal guardrail as a non-negotiable constraint on each cut, so a guardrail with no observable threshold lets a cut violate it with nothing in place to catch that. - Naming several approvers for the objective without also naming a conflict resolver and a decision rule leaves no way to settle it when they disagree about what the goal actually is; the brief then stalls on a tie nobody is placed to break.
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.

