Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →nwave-ai/nw-speculative-dispatch
Speculative parallel implementation methodology — dispatch N candidate implementations, audit all, score, pick best. Auditability mandate: ALL candidates logged (not just winner).
Passed 18 of 21 test cases with the skill, and 6 of 21 without it.
and 15 more in the full report
13 fixed6 passed either way3 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 12 test cases the skill fixed. Same model, same prompt, with and without the skill.
Candidate A and Candidate B both pass all unit tests and both have a complexity delta of 2. Candidate A adds 4 lines of code and runs in 1.2s. Candidate B adds 8 lines of code and runs in 0.3s. Which candidate wins?
| Checked | Without skill | With skill |
|---|---|---|
| Selects Candidate A as the winner | Missed | Passed |
| States that lines_added takes precedence over runtime as a tiebreaker when complexity is equal | Missed | Passed |
The instructions your model reads when the skill loads.
Speculative dispatch is a technique where the orchestrator generates N candidate implementations of the same TDD step in parallel, scores each by measured properties, and picks the best — while logging every candidate, including discarded ones, for human audit.
Use speculative dispatch when the decision is ambiguous and at least 3 feasible candidate strategies exist.
| Signal | Use speculative dispatch? |
|---|---|
| Step has ≥3 plausible implementation strategies | YES |
| Decision is clear from prior context | NO — implement directly |
| Only 1-2 strategies, clear winner | NO — implement directly |
| Performance matters and strategies have measurable trade-offs | YES |
| Domain is novel and "correct" structure is unknown | YES |
Speculative dispatch is orthogonal to TDD stages — it applies at any phase (RED unit-test authoring, GREEN, COMMIT under the 3-phase canon per ADR-025; or RED_UNIT, GREEN, COMMIT under the legacy 5-phase contract) where ambiguity is genuine and 3 candidates are feasible within the time budget.
Generate exactly 3 candidates per speculative step by default:
| Role | Strategy |
|---|---|
| minimal-change | Inline, no new abstractions. Lowest complexity, fewest lines. |
| refactor-heavy | Extracts helpers, adds guards, defensive patterns. Higher complexity. |
| pattern-extraction | Generalises to a factory/template. Highest lines. Justified if reuse follows. |
Three candidates is the minimum viable set for meaningful scoring. More than 5
test cases passed, vs 6 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.