prep docs and service updates
Some checks failed
Docs CI / lint-and-preview (push) Has been cancelled
Some checks failed
Docs CI / lint-and-preview (push) Has been cancelled
This commit is contained in:
@@ -0,0 +1,25 @@
|
||||
# Advisory AI Knobs Prep — PREP-POLICY-ENGINE-31-001-ADVISORY-AI-KNOBS-R
|
||||
|
||||
Status: Draft (2025-11-20)
|
||||
Owners: Policy Guild
|
||||
Scope: Outline the configuration knobs for Advisory AI scoring once trust weighting (30-101) is defined.
|
||||
|
||||
## Dependencies
|
||||
- Trust weighting API/contracts (30-101).
|
||||
- Advisory AI signal list and confidence model (not yet finalized).
|
||||
|
||||
## Draft configuration fields
|
||||
- `weights` (from 30-101) applied to AI-produced signals.
|
||||
- `knobs[]` array of `{name, default_value, min, max, step, description}` for:
|
||||
- `ai_signal_weight`
|
||||
- `reachability_boost`
|
||||
- `time_decay_half_life_days`
|
||||
- `evidence_freshness_threshold_hours`
|
||||
- Output preview: recomputed severity/exploitability per advisory/component.
|
||||
|
||||
## Acceptance to close PREP
|
||||
- Map knobs to actual AI signals once published; confirm allowed ranges.
|
||||
- Document schema at `docs/modules/policy/schemas/advisory-ai-knobs@draft.json` and sample request at `docs/modules/policy/samples/advisory-ai-knobs@draft.json`.
|
||||
|
||||
## Handoff
|
||||
Use this document as the prep artefact for PREP-POLICY-ENGINE-31-001-ADVISORY-AI-KNOBS-R. Update with final signal list and ranges once trust weighting is frozen.
|
||||
Reference in New Issue
Block a user