Aa new project
Agile planning + execution for AI coding agents: grill the idea, plan with OKRs, adversarial critic panel with kill authority, cheap-model story cards, parallel wave execution. MIT, no token, ever.
npx -y skills add Peeyushmeher/agent-agile --skill aa-new-projectAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 24 days oldThe repository was created 24 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Use when the user wants to plan a new project from a clear idea, has already locked an idea with aa-grill, or invokes /aa-new-project.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
3.6 KB, as published. Nobody here has run it
aa-new-project
Resolve the Agent-Agile playbook root: use the first of these that exists — (1) ${CLAUDE_PLUGIN_ROOT}/playbooks, (2) ./.claude/agent-agile/playbooks, (3) ./.agents/agent-agile/playbooks, (4) ~/.claude/agent-agile/playbooks, (5) ~/.agents/agent-agile/playbooks, (6) ./playbooks.
Read playbooks/planner.md sections The 8 intake questions, Pushback patterns, Drafting the plan, Research fan-out and the risk register, Slicing epics into stories, The worker-readiness test, and Extracting PREREQS, in full, and follow them exactly — do NOT re-derive or improvise the system.
Intake
If .planning/IDEA.md exists, read it and, per planner.md's Grill → intake mapping, confirm the pre-filled answers back to the user instead of re-asking them. Ask only the questions the grill didn't answer, plus the three it never answers (hard constraints, riskiest unknown, review-batch appetite) — always fresh. If IDEA.md does not exist, ask all 8 questions from scratch, one at a time, applying the pushback patterns to every thin answer before moving on.
After the eighth answer and before writing any files, deliver the split statement exactly as defined at the end of planner.md's The 8 intake questions — the "my part / your part" narration plus the process preview. It is a required step, not optional color.
Drafting
Draft the plan inline in this conversation, following Drafting the plan — this is the default; do not spawn a separate agent for this step unless the user asks for it. Produce, in order:
.planning/PROJECT.mdfromplaybooks/templates/PROJECT.md..planning/ROADMAP.mdfromplaybooks/templates/ROADMAP.md— every epic, riskiest first, one-line demo sentences; only the current epic (Epic 1) gets full detail.- If the roadmap draft is multi-epic, run Research fan-out and the risk register exactly as written (including its off-ramp): spawn
aa-researcher-domainandaa-researcher-ecosystemas parallel fresh subagents with the model configured for the smart tier in.planning/CONFIG.md(ask for tiers now if CONFIG.md doesn't exist yet), synthesize.planning/RESEARCH.mdfromplaybooks/templates/RESEARCH.md, then revise the roadmap draft against the risk register —[fixture: …]tags on epic lines, Epic 1/2 riskiest-first re-examination. - Slice Epic 1 into story cards per Slicing epics into stories, writing each from
playbooks/templates/STORY.md— including one fixture story per[fixture: …]tag on Epic 1. Apply The worker-readiness test to every card before calling it done. .planning/PREREQS.mdfromplaybooks/templates/PREREQS.mdper Extracting PREREQS, statuspendingfor every row..planning/CONFIG.mdfromplaybooks/templates/CONFIG.md— ask the user for smart_tier, cheap_tier, gate, and commercial if not already set..planning/STATE.mdfromplaybooks/templates/STATE.md, Phaseintake, pointing at PROJECT.md, ROADMAP.md, and Epic 1's stories.
Handoff
If the roadmap has more than one epic, tell the user the plan must go through the panel before anything runs, and instruct them to run /aa-panel next — do not start execution first. If the roadmap is a single throwaway epic, the panel may be skipped — see the gate rule in playbooks/critics.md section Panel protocol; say so and name the next command instead.