XFPA
XF Process Audit

Learns your judgment so the HIL prompts fade where they've earned trust.

Every approval, every override, every skip — in XFCD, in XFBA, in XSIA, in every HIL prompt XF surfaces — becomes training signal. XFPA aggregates decisions per rule with recency decay, computes rolling confidence, and graduates auto-apply when a rule earns it. First-time users get the full guardrails. Power users get the intelligence of having trained the platform to their judgment.

Part of Memory Layer

Connect with GitHub — it's free

What it does

1

Per-rule rolling confidence — not global trust.

XFPA tracks confidence per matched rule (e.g., ‘suppress-stamp-removal’, ‘approve-resolution-xfcd’). A rule you've approved 12 times is graduated; a brand-new rule still prompts. Trust is scoped.

2

Recency-weighted with 30-day half-life.

An approval from a year ago counts half as much as one from last month. Stale patterns fade. Current patterns dominate. The platform adapts as your judgment evolves.

3

Graduated auto-apply with configurable threshold.

Default: 5+ signal decisions at >= 0.8 confidence graduates a rule. Contradictions and novel rules always escalate to HIL. The gate only fires when the rule has earned it.

How it works

1

Decisions from XFCD, XFBA, XSIA, and other HIL modules log to xfcd_decisions.jsonl.

2

/xfpa ingests the log into the XFMT tree as process-audit leaves.

3

Rolling confidence computed per rule with recency decay.

4

Threshold table written to ~/.claude/xfmt/xfpa_thresholds.json.

5

Eligible rules auto-apply on next firing; ineligible rules prompt HIL as before.

6

/xfpa status shows current confidence table for review.

Other modules