com um clique
research
Investigate a question from real sources and land cited findings.
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Menu
Investigate a question from real sources and land cited findings.
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Turn a repetitive manual procedure into a proven, repeatable mechanism. Use when the user explicitly invokes the automation pipeline.
Plan, approve, implement, verify, and record a software change. Use when the user explicitly invokes the build pipeline.
Record a durable architecture decision with alternatives and consequences.
Discover, approve, implement, prove, review, and record a software change. Use when the user explicitly invokes the delivery pipeline.
Leave unfinished repository work safe for another human or agent to resume.
Evaluate whether retained evidence supports the delivery claims being made.
| name | research |
| description | Investigate a question from real sources and land cited findings. |
| disable-model-invocation | true |
| argument-hint | <task> |
Use this pipeline when the deliverable is an answer, not a change: a question whose answer must come from sources rather than memory. Treat the invocation text as the research question. The run is read-only outside its own artifacts — findings that demand changes become work items or decision entries, never edits made mid-run.
Task: $ARGUMENTS
Record harness as claude in the run ledger.
Clarify the requested outcome from available context. Inspect the repository before asking
questions that the files can answer. Create the run folder and write task.md containing the
outcome, constraints, assumptions, and acceptance checks. Reserve an empty evidence/ folder.
Identify where the answer actually lives — repository files, primary documentation, live systems,
the web — and consult the strongest sources available, not the most convenient. Capture every
consulted source in sources.md in the run folder: what it is, where it lives, what it says that
matters, and how much weight it deserves. Separate observed fact from inference; when sources
conflict, record the conflict instead of silently picking a winner. Stop gathering when new
sources stop changing the picture.
Write findings.md in the run folder: the answer to the task's question, with each material claim
tied to a source recorded in sources.md. State confidence plainly and name what evidence would
change the conclusion. List what remains unknown instead of papering over it. If the findings
warrant action, name the follow-up explicitly — a work item, a decision entry — rather than
leaving it implied.
Always run this stage, including after rejection, failure, or abort. Append exactly one compact JSON
object line to .base/history.jsonl with slug, date, pipeline, harness, outcome, and
paths. Use completed, aborted, or failed for outcome; never rewrite previous lines.