| name | bspdn-goal-driver |
| description | Drive BSPDN objective optimization toward target outcomes (about 8% dynamic-power reduction without area/timing regression, or 5%+ frequency uplift with non-worse power/area) using gated experiments and evidence tracking. |
BSPDN Goal Driver
When to use
Use this skill when the user asks to push objective/performance toward explicit PPA goals, not just run exploratory experiments.
Knowledge And Tool Interaction
- Read the latest route/STA/PPA evidence and milestone history before proposing the next campaign.
- If the task needs shared KB/tool lookup before campaign planning, delegate that retrieval step to
eda-context-accessor.
- Write every milestone decision back into durable evidence artifacts so later optimization rounds inherit the exact gate history.
Outputs
bspdn_goal_progress.summary.md
bspdn_goal_progress.detail.tsv
- campaign plan (
goal_campaign_plan.md)
- evidence-backed milestone decision note (
M0/M1/.../M4).
Hard rules
- No promotion claim without execution-contract PASS.
- Goal-B frequency claim must come from period sweep, not WNS proxy alone.
- Record exact selector/route-hint config for each compared run.
Operational References
- Load
references/goal-contract.md when deciding whether the task is targeting power goal A, frequency goal B, or both.
- Load
references/milestone-policy.md when deciding current milestone, promotion readiness, or whether a campaign should advance.
- Load
references/campaign-workflow.md when building the goal tracker, selecting testcase tier, or running the gated A/B campaign.
- Load
references/reporting-policy.md when deciding whether a user-facing milestone report or PDF summary is justified.