see-it-live
Skill1 starsChanged 3 months agoMITView on GitHub
Launch the app and confirm the specific change is visible and working: screenshot, curl, or run it. Doubles as the fast smoke pass. Use for run the app, start the server, see it live, smoke test, or post-deploy check.
In this repository12 other files it gives its agents.
- templates/AGENTS.mdAGENTS.md
- templates/CLAUDE.mdCLAUDE.md
- adversarial-reviewSkill
- architect-reviewSkill
- delivery-loopSkill
- fast-pr-workflowSkill
- orchestratorSkill
- post-merge-cleanupSkill
- qa-automationSkill
- security-reviewSkill
- task-researchSkill
- web-devSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
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
- Very specific: line 29
Before launching, write down the one signal that proves the change ("the Save button shows a spinner", "/api/jobs returns status: queued").
- Distinctive: line 39
No error spew in logs or console.
- Very specific: line 43
Keep it deterministic: no flaky waits, no giant fixtures, no unstubbed third parties, a clear pass or fail.
- Distinctive: line 48
Stop the server and spawned processes.
- Very specific: line 50
Not automated testing: repeatable checks belong to qa-automation.
≈ “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
See it live
Green type-checks and tests do not prove the change works. Behavior lives in the running app: start it and observe the specific change before saying done.
Launch path
The project knows its run command better than this table. Inventory the targets it declares first: build-file targets (Makefile, justfile), package.json scripts, CI job names, the README run and validate section. A named journey, playtest, E2E or smoke target outranks a generic launch, and its graphical variant outranks the headless one when the display allows. Headless-only proof shows absence of errors, not that the change is visible; say so when you fall back to it.
| Project type | Launch | Observe |
| Game engine | declared playtest or journey target; a headless quit run catches script errors fast | the scripted outcome passes and the changed behavior happens, not just "scene loads" |
| Web frontend | dev server, open the changed route | golden path and error state rendered; screenshot |
| Backend or API | start the server, curl the changed endpoint with a real payload | health returns 200; status and response shape match |
| CLI or library | run the binary with real args | stdout, stderr and exit code are correct |
Confirm the specific change
- Before launching, write down the one signal that proves the change ("the Save button shows a spinner", "
/api/jobsreturnsstatus: queued"). - Launch, reproduce the exact path that exercises it, and check for that signal. "It still boots" is necessary, not sufficient.
Smoke mode
The same paths are the fast critical-path gate between built and shipped. Four checks, answered in seconds:
- The app boots with no fatal error.
- The core path works (the one thing the product must do).
- A key endpoint responds (health or the primary surface).
- No error spew in logs or console.
Run it after integrating a change and again after deploy or merge. A failed smoke pass is a hard stop for shipping. The post-deploy run hits the deployed URL, not localhost, and confirms the new build is live (version endpoint, build hash, or a marker from the change). A post-deploy failure is a rollback trigger: surface it loudly.
Keep it deterministic: no flaky waits, no giant fixtures, no unstubbed third parties, a clear pass or fail. The tenth assertion means you have drifted into qa-automation.
Evidence and teardown
- UI: a screenshot of the changed state. API or CLI: the actual response body or stdout, pasted, with the command that produced it.
- Stop the server and spawned processes; revert seeded data or toggled flags.
Not automated testing: repeatable checks belong to qa-automation. This is the "I watched it happen" step.
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.