skills/cy-execute-task/SKILL.md
Executes one PRD task end-to-end uninterrupted — resolves spec conflicts autonomously, implements, validates, and updates tracking without pausing for questions. Use when a prompt includes a task specification that must be implemented, validated, and reflected in task tracking files. Do not use for PR review batches, generic coding tasks without a PRD task file, or standalone verification-only work.
npx skillsauth add compozy/compozy cy-execute-taskInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Execute one PRD task from exploration through tracking updates — uninterrupted.
Compozy runs tasks in a loop. Pausing for clarification breaks the loop.
_tasks.md task graph manifest path.Ground in repository and PRD context.
_techspec.md, _tasks.md, and the contract catalogs _user_stories.md and _tests.md when present.adrs/ subdirectory of the PRD directory to understand the architectural decision context for this task.analysis/ summaries for decision context.cy-workflow-memory skill before editing code.Build the execution checklist.
Validation, Test Plan, or Testing item into a numbered working checklist.## Tests section, and implement each case as _tests.md specifies it — the assigned IDs are part of the deliverable, not a suggestion.Implement the task.
Validate and self-review.
cy-final-verify skill. This step is mandatory regardless of auto-commit mode — always verify before claiming completion.Update task tracking.
references/tracking-checklist.md when applying status, checklist, or commit updates._tasks.md during normal completion tracking; it owns graph topology, not per-task status.Handle commit behavior.
Task files paraphrase the spec, and paraphrases drift. Real incident: a task was implemented straight from its task file while the spec directory contained the canonical, complete definition of the deliverable (_examples.md) and the QA input contract (_qa.md); neither was read, seven review rounds validated engineering quality only, and the shipped result contradicted the product contract wholesale. This survey exists so that can never repeat.
Explore in Claude Code, or the closest scoped read-only subagent the current harness offers) to inventory the ENTIRE spec directory — every file, not a subset: all _*.md siblings (_prd.md, _user_stories.md, _techspec.md, _tests.md, _tasks.md, _examples.md, _qa.md, and any others present), loose siblings such as requirements.md and product-ux.md, plus the adrs/, analysis/ (including summary*.md), handoffs/, and memory/ subdirectories.analysis/ summaries (at minimum the newest summary*.md) for the decision context behind the spec.Resolve contradictions autonomously with this ladder (highest wins). Record the pick; continue.
CHECKs, typed tables, enumerated states) beat conflicting prose in the same TechSpec._tests.md assigned cases, _examples.md, _qa.md input tables, _user_stories.md acceptance criteria, parity maps) beat a task-file paraphrase of the same fact.analysis/ notes when they address the same decision._tests.md cases and remains implementable against the winning schema/state constraints.Authority for WHAT to build remains the task file + PRD + TechSpec + ADRs, read through this ladder.
The existing runtime shape is never the contract. If the current code cannot express what the resolved contract requires, extend or adapt the runtime within task scope; if that is truly out of scope, implement the closest faithful solution, record the gap as a follow-up note, and continue — never pause to ask, and never mold the deliverable to whatever the runtime happens to support today without recording the gap.
development
Expands a raw idea into a structured, research-backed spec in .compozy/tasks/<slug>/_idea.md through interactive brainstorming, web research, business analysis, and multi-advisor debate. Use when the user has a feature idea and wants to explore and structure it before creating a PRD. Do not use for PRD creation, technical specifications, task breakdown, or code implementation.
testing
Reconciles a finished workflow's planned decisions (its Accepted ADRs) against the settled reality (git diff, review issues, and task status), then promotes the proven, cross-feature-durable ones into a durable project decision log at .compozy/DECISIONS.md plus .compozy/decisions/AD-NNN.md. Use when a workflow has finished its full pipeline (review round, reviews fix, and final verify) and you want to capture its durable decisions as the final step, or when re-running capture to refresh the log after further changes. Do not use for capturing decisions mid-implementation before review remediation, for PRD or TechSpec authoring, for PR review remediation, or for generic note-taking.
development
Performs a comprehensive code review of a PRD implementation and generates a review round directory with issue files compatible with cy-fix-reviews. Use when reviewing implemented PRD tasks, creating a manual review round without an external provider, or performing a quality audit of code changes. Do not use for fetching reviews from external providers, fixing existing review issues, executing PRD tasks, or editing source code.
development
Enforces fresh verification evidence before any completion, fix, or passing claim, and before commits or PR creation. Use when an agent is about to report success, hand off work, or commit code. Do not use for early planning, brainstorming, or tasks that have not yet reached a concrete verification step.