Skip to content
Apothem
Command pipeline

/research-ideate

/research-ideate — Formulates the problem space from a domain seed: opportunity-and-gap scanning, invalidated-prior-hypothesis scanning, and ranked candidate research-question generation — the head stage of the /research pipeline, feeding /research-spec.

Role: Principal Investigator (problem formulation) Pipeline position: head

Formulates the problem space from zero — opportunity-and-gap scanning, invalidated-prior-hypothesis scanning, and candidate research-question generation with ranking — the entry stage of the /research pipeline. Every candidate is framed against an a-priori theoretical anchor, grounded in prior art, and stated in a falsifiable form, ready for /research-spec.

Canonical invocation

/research-ideate

With arguments:

/research-ideate [path/to/domain-notes] [--suite-name NAME] [--override] [--domain NAME]

Inputs

A domain seed — free-form domain notes (operator-supplied inline or as a path), or a bare --domain name when starting from scratch.

Outputs

_inputs/ideation.md (the framed problem space, the opportunity-and-gap scan, the invalidated-prior-hypothesis scan, and the ranked candidate-question slate) plus the Handoff Manifest at the research-suite folder.

Downstream

/research-spec (downstream consumer — frames the chosen candidate into a falsifiable spec).

Workflow phases

The command follows the standard /research pipeline workflow:

  1. Load context — Read the domain seed, the rules, and any existing suite state.
  2. Scan the opportunity and gap space — Map what is studied, what is contested, and what is unaddressed in the domain.
  3. Scan invalidated priors — Surface hypotheses already refuted by prior art so the slate does not re-tread settled ground.
  4. Generate and rank candidates — Produce candidate research questions, each stated in a falsifiable form, ranked by tractability and impact.
  5. Verify outputs — Confirm every candidate is grounded in primary prior art and refutable; resolve every ambiguity through the structured-inquiry channel.
  6. Emit the ideation slate_inputs/ideation.md plus the handoff manifest.

Failure modes

SymptomCauseRecovery
No falsifiable candidateDomain seed too broad or purely descriptiveNarrow the domain via structured inquiry; re-frame candidates as refutable questions
Candidate restates an invalidated priorPrior-art scan skippedRun the invalidated-prior scan; drop or re-frame the overlapping candidate
Missing domain seedNo notes or --domain suppliedSupply domain notes inline / as a path, or pass --domain NAME

Examples

# Scan a domain and generate ranked candidate questions
/research-ideate ~/notes/retrieval-systems.md --suite-name retrieval-latency

# Start from a bare domain name
/research-ideate --domain "vector-index recall under drift"

Cross-references

On this page