h-explore
h-explore generates three to five structurally distinct solution variants for a given problem using the FPF (Framing, Parallel, Feedback) workflow. Each variant is assigned a different conceptual direction, includes a clearly identified weakest link, and indicates whether it functions as a stepping stone to future exploration. Use this skill when you need diverse candidate approaches to a problem and want alternatives differentiated by kind rather than incremental tweaks, ensuring each option has transparent trade-offs and architectural reasoning.
git clone --depth 1 https://github.com/m0n0x41d/haft /tmp/h-explore && cp -r /tmp/h-explore/packages/haft-pi/skills/h-explore ~/.claude/skills/h-exploreSKILL.md
# h-explore — Keep alternatives live Retrieve current FPF source with `mode="inspect"` for a known SourceID/UnitID, or `mode="concern"` for the exploration question. Inspect the direct pattern body; retrieval does not choose or rank a candidate. Treat a concern result's `candidate_set` as incomplete navigation. Before relying on one candidate, inspect its exact identifier and direct pattern body. Keep several candidates live or abstain when the basis is insufficient. Never query after exploration merely to manufacture support for candidates already chosen. Generate candidates that differ in kind, not only degree. For each state its novelty marker, strengths, risks, weakest link, and any defensible stepping- stone role. Keep unresolved branches and return conditions visible. Return candidates conversationally by default. Call `haft_solution(action="explore")` only on explicit save intent or when current Work supplies a concrete operator-named or agent-inferred receiving use. A typed durable portfolio needs one independently addressable ProjectRecord per option: persist candidate-description Notes under the same exact concern, keep each returned `Haft.ProjectRecordRef`, and pass it as the variant's `project_record_ref`. Do not derive or invent record IDs. Without exact option refs, preserve the legacy portfolio but report typed projection as underdetermined. Do not invent a ProblemCard to satisfy a sequence and do not prescribe comparison next.
|
|
Manual-only skill that creates bounded execution authority from an explicit operator grant. Never auto-trigger or infer approval from prompt text or tool schemas.
Compare two or more existing candidates under an explicit characteristic space, parity basis, and predeclared selection policy. Return trade-offs and a non-dominated set; persistence is conditional and a binding choice requires a direct unambiguous operator request.
Route one direct, unambiguous operator request for a bounded binding choice. A manual h-decide invocation is a compatible shortcut, not an authorization receipt.
Diagnose a concrete failure with rival-hypothesis testing. Stabilize the symptom, generate distinct explanations, test discriminating observations read-only in parallel, and rank by evidence while retaining losing rivals.
Shape an under-articulated engineering problem without assuming a solution or forcing a project phase. Default to a conversational frame; create a ProblemCard only on explicit save intent or when current Work supplies a concrete operator-named or agent-inferred receiving use that needs a durable accepted problem statement.
Persist an explicitly requested non-binding fact, observation, caveat, or small rationale in Haft project memory. Do not auto-persist ordinary reasoning.