agent-afk 3.92.0 → 3.92.2

package/dist/bundled-plugins/awa-bundled/skills/shadow-verify/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: shadow-verify
-description: "Dispatch a parallel adversarial verifier wave after any high-stakes sub-agent investigation (code reviews, audits, findings reports, large refactors). Shadow verifiers independently re-derive 2–3 key claims from scratch and flag disagreements before the user acts. Use when sub-agent output will drive decisions, file changes, commits, or external side-effects."
+description: "Dispatch a parallel adversarial verifier wave after any high-stakes sub-agent investigation (code reviews, audits, findings reports, large refactors) — or whenever a sub-agent asserts a claim with high-confidence language (confident, certain, clearly, ≥80%), since confidence is a trigger, not a verdict. Shadow verifiers independently re-derive 2–3 key claims from scratch using tool calls only, returning CONFIRMED/REFUTED/UNVERIFIABLE, and flag disagreements before the user acts. Use when sub-agent output will drive decisions, file changes, commits, or external side-effects."
 context: fork
 ---
 
@@ -12,14 +12,17 @@ When a sub-agent (or wave) returns investigation findings, code-review conclusio
 **Wave 2 — Adversarial verifiers (parallel, independent):**
 1. Extract 2–3 concrete, re-checkable claims from the returned report (e.g., "X function is unused", "file Y exceeds 300 lines", "PR targets main", "no tests cover Z").
 2. Dispatch one shadow sub-agent per claim, in parallel. Each receives ONLY the claim + the user's original goal — never the original agent's reasoning or cited evidence. **Default to `subagent_type: "research-agent"` (mechanically locked to Read/Grep/Glob/WebFetch/WebSearch — cannot Edit/commit/push).** If the claim requires Bash to verify (running a failing test, `gh pr view`, `git log origin/...`), fall back to a Bash-capable subagent type with `isolation: "worktree"` and prepend this prefix to the prompt: *"Verifier sub-agent — do not Edit, Write, commit, push, `gh pr create`, or `curl`. Return findings only."*
-3. Each verifier re-derives the verdict independently. Returns `{claim, verifier_verdict, evidence_pointer}`.
+3. Each verifier re-derives the verdict independently using tool calls only — never re-reading the original report's reasoning. Returns `{claim, verifier_verdict, evidence_pointer}`, where `verifier_verdict` is one of `CONFIRMED`, `REFUTED`, or `UNVERIFIABLE`. On `REFUTED`, the verifier also emits a corrected finding.
 
 **Merge:**
-- All verifiers confirm → surface the original report as validated.
-- Any verifier disagrees → show the original claim alongside the verifier's counter-claim with evidence. Do not act until the conflict is resolved.
+- `CONFIRMED` → surface the claim as validated.
+- `REFUTED` → replace the claim with the verifier's corrected finding, annotated `[was: confident, now: refuted]`, and show it alongside the original with evidence. Do not act until the conflict is resolved.
+- `UNVERIFIABLE` → surface with a `[needs-human-review]` tag rather than passing it through silently.
+
+Bound the loop: at most 3 verification rounds per session. Claims still unresolved after 3 rounds are escalated to the user, never silently dropped.
 
 **When to invoke:**
-Any time sub-agent output will drive user decisions, file edits, commits, external side-effects, or is the basis of a user-facing summary.
+Any time sub-agent output will drive user decisions, file edits, commits, external side-effects, or is the basis of a user-facing summary. Treat **high-confidence language as a trigger in its own right**: when a review/audit sub-agent asserts a claim with markers like "confident", "certain", "clearly", "obviously", "must be", or a stated probability ≥ 80%, verify it as if it were decision-driving regardless of stakes. Confidence is a trigger, not a verdict.
 
 **Skip when:**
 Sub-agent ran inside an orchestrator skill that already verifies (`resolve`, `diagnose`, `appmap`); sub-agent returned explicit failure; work was purely exploratory and no decision follows; or the session is **text-terminal** — a pure explanation, architecture walkthrough, onboarding Q&A, or capability map that names no mutated artifact (file/PR/commit/test), where there are no re-checkable state claims for adversarial verifiers to re-derive (assess coverage, coherence, and citation density instead of dispatching re-derivation sub-agents).