{"slug":"dietrichgebert-ponytail","kind":"smoke","status":"testing","tier":"smoke","title":"ponytail: the 'laziest senior dev' prompt for coding agents","category":"coding-agents/skills-plugins","topic":"coding-agents","repos":[{"repo":"DietrichGebert/ponytail","stars":166274,"license":"MIT"}],"claims":[{"claim":"Makes your AI agent think like the laziest senior dev in the room: the best code is the code you never wrote","source":"https://x.com/trending_repos/status/2106718156463182216"}],"plan":"Smoke first: confirm the single prompt file (SKILL.md) and AGENTS.md exist and parse, and that the Claude Code plugin marketplace manifest is valid JSON. A real before/after agent run, measuring whether ponytail makes an agent reuse code instead of writing new code, is queued as a claim test.","who_its_for":"Developers who want their coding agent to stop over-engineering.","explainer":"A single widely-shared prompt (166k stars) that tells a coding agent to avoid writing unnecessary code and reuse what exists. Rocky first checks it installs cleanly, then will measure whether it actually changes agent behavior.","measure":["Does the prompt and plugin install cleanly","Does the agent write less new code with it than without (claim run, queued)"],"notes_so_far":["Method: clone at 09c434a (v5.3.2); smoke only for now. The behavioral claim run needs a controlled agent task and is queued."],"sneak_peek":["Smoke test: the repo cloned, and its skill file and AGENTS.md are both present.","The first run failed because of a quoting bug in Rocky's own test script, not in ponytail. It's fixed, and the real claim test (18 agent runs, with and without the skill) is queued."],"charts":[],"gallery":[],"results":{"jobs":[{"id":19,"title":"ponytail: SKILL.md/AGENTS.md parse, plugin marketplace manifest valid JSON (smoke)","kind":"smoke","status":"failed","started":"2026-10-18T05:10:02Z","finished":"2026-10-18T05:10:04Z","wall_seconds":2,"install_friction":null,"artifacts":[]},{"id":40,"title":"ponytail claim: with vs without skill, coding packet, 3 repeats (18 agent runs, local model) [rerun: /out permissions]","kind":"claim","status":"done","started":"2026-10-18T18:39:18Z","finished":"2026-10-18T19:54:29Z","wall_seconds":4511,"install_friction":null,"artifacts":[]}],"smoke_only":false,"ingested":"2026-10-18T20:31:30Z"},"last_test_finished":"2026-10-18T19:54:29Z","results_in":true}