goal-execution
Skill6 starsChanged 6 months agoApache-2.0View on GitHub
Complete a bounded implementation goal in the current agent run, with explicit finish criteria and evidence-based verification. Use for requests that should continue through implementation and checks instead of stopping after a plan.
In this repository169 other files it gives its agents.
- AGENTS.mdAGENTS.md
- apps/AGENTS.mdAGENTS.md
- design-templates/AGENTS.mdAGENTS.md
- e2e/AGENTS.mdAGENTS.md
- .github/AGENTS.mdAGENTS.md
- packages/AGENTS.mdAGENTS.md
- plugins/AGENTS.mdAGENTS.md
- skills/AGENTS.mdAGENTS.md
- tools/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- od-contributeSkill
- 8-bit-orbit-video-templateSkill
- ad-creativeSkill
- after-hours-editorial-templateSkill
- agent-browserSkill
- ai-music-albumSkill
- algorithmic-artSkill
- apple-higSkill
- article-magazineSkill
- artifacts-builderSkill
- brainstormingSkill
- brand-extractSkill
- brand-guidelinesSkill
- brandkitSkill
- industrial-brutalist-uiSkill
- canvas-designSkill
- card-twitterSkill
- card-xiaohongshuSkill
- color-expertSkill
- competitive-ads-extractorSkill
- copywritingSkill
- creative-directorSkill
- d3-visualizationSkill
- data-reportSkill
- deck-guizang-editorialSkill
- deck-open-slide-canvasSkill
- deck-swiss-internationalSkill
- design-briefSkill
- design-consultationSkill
- design-mdSkill
- design-reviewSkill
- digits-fintech-swiss-templateSkill
- doc-kami-parchmentSkill
- docSkill
- docxSkill
- domain-name-brainstormerSkill
- ecommerce-image-workflowSkill
- editorial-burgundy-principles-templateSkill
- emilkowalski-motionSkill
- enhance-promptSkill
- export-download-debuggingSkill
- fal-3dSkill
- fal-generateSkill
- fal-image-editSkill
- fal-kling-o3Skill
- fal-lip-syncSkill
- fal-realtimeSkill
- fal-restoreSkill
- fal-trainSkill
- fal-tryonSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 9 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
Use this skill for one bounded goal that should end in a verified result, not only a plan or progress update.
- Distinctive: line 37
Before changing anything, derive a compact goal contract from the user's request and the selected project context:
- Distinctive: line 47
Do this internally when the request is already clear.
- Distinctive: line 56
Run proportional verification and inspect its actual output.
- Distinctive: line 58
If a criterion is unmet and a safe in-scope fix is available, iterate.
- Distinctive: line 61
Keep the loop efficient: reuse already-read context, avoid broad rescans and duplicate commands, and do not inflate the prompt with …
- Distinctive: line 70
do not perform destructive, production, publishing, messaging, purchasing, or credential-changing actions unless the user already …
- Distinctive: line 72
stop for a required approval or material product decision instead of guessing.
- Distinctive: line 74
do not weaken tests, security controls, or acceptance criteria merely to claim completion.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Goal execution
Use this skill for one bounded goal that should end in a verified result, not only a plan or progress update. It applies to the current agent run. It is not a background scheduler and does not create a third chat mode.
Establish the finish line
Before changing anything, derive a compact goal contract from the user's request and the selected project context:
- Objective: the concrete outcome to deliver.
- Completion criteria: observable checks that prove the outcome is done.
- Constraints: scope, compatibility, quality, security, and performance requirements already stated by the user or repository instructions.
- Verification: the smallest reliable test, build, inspection, or runtime check that covers the changed behavior.
Do this internally when the request is already clear. Ask the user only when a missing choice would materially change the result or when new authority is required.
Execution loop
- Inspect only the relevant project evidence and current state.
- Make a short plan tied directly to the completion criteria.
- Implement the next useful change.
- Run proportional verification and inspect its actual output.
- Compare the result with every completion criterion.
- If a criterion is unmet and a safe in-scope fix is available, iterate.
- Finish only when all required criteria pass or a genuine blocker remains.
Keep the loop efficient: reuse already-read context, avoid broad rescans and duplicate commands, and do not inflate the prompt with unrelated files or resource catalogs.
Safety boundary
Selecting this skill never expands authority. Continue to obey repository, sandbox, approval, credential, and external-system policies. In particular:
- do not perform destructive, production, publishing, messaging, purchasing, or credential-changing actions unless the user already authorized them;
- stop for a required approval or material product decision instead of guessing;
- do not weaken tests, security controls, or acceptance criteria merely to claim completion;
- report an unavailable external dependency or exhausted safe retry path as a blocker, with the completed evidence preserved.
Completion report
At the end, state the delivered outcome first, then the verification evidence and any remaining limitation. Never report the goal as complete when a required criterion was skipped, failed, or could not be observed.
For work that must run again later, on a schedule, or after the current app session ends, create an Automation instead. Automations provide durable scheduled execution; this skill provides completion discipline inside the current run.
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.

