by GTDataworks
Six skills to compare agent runs, turn failures into regression tests, build eval datasets, gate provenance, inspect untrusted content, and write audit reports whose proof a stranger can re-check.
by GTDataworks
Convert a repeatable workflow into a reusable agent skill or staged skillchain with contracts and a build receipt.
by GTDataworks
Run a buyer-readiness check before you publish an AI agent skill. PASS/WARN/FAIL on structure, claims, secrets, and listing clarity.
by GTDataworks
Generate a source-safe repository audit and repair handoff without mutating your code. Secret scan, risk register, Map of Change, Codex/GPT prompts.
by GTDataworks
Mechanical eval gates for agents: 12 gate contracts, strict schemas, dependency-free validators, pass/fail runner with no LLM judge.
by GTDataworks
Capture tool-trace receipts, label failures, and emit replay prompts so you can debug an agent run after it finishes.
by GTDataworks
10 read-only incident response skills: ransomware, M365/Google Workspace account takeover, phishing/BEC and supply-chain incident command, DFIR casebuilding, Sigma-compatible detection rules, threat hunting, CISA KEV patch prioritization, and Kubernetes and AI-agent security review.
by GTDataworks
7 skills for securing and testing tool-using AI agents: guardrails that stop prompt injection from becoming action authority, runaway-agent containment, agent memory integrity, multi-agent orchestration, tool and MCP connector error recovery, trajectory grading and adversarial eval design.
by GTDataworks
Four small skills, tested on small models with graded runs, that stop coding agents from faking a passing check, claiming tests they never ran, widening the scope of a request, changing locked files, or taking irreversible steps without approval.
by GTDataworks
Coordinates evidence-preserving containment of compromised or runaway agents across sessions, tools, connectors, memory, child workers, and downstream mutations, then gates re-enablement.
by GTDataworks
Creates adversarial, lineage-safe, blind agent evaluation campaigns with negative controls, deterministic external grading, and predeclared Luna versus Luna Cyber experiments.
by GTDataworks
Compiles least-authority, target-bound, expiring grants and intercepts agent actions so prompts, retrieved content, tool output, and connectors cannot mint or expand authority.
by GTDataworks
An evidence-backed agent authority graph, adversarial test plan, control map, and release decision with measurable residual risk.
by GTDataworks
A defensible evidence index and normalized timeline whose conclusions can be independently reproduced.
by GTDataworks
Coordinates multi-agent work through authority-bounded delegation, dependency graphs, independent verification, dissent-preserving fan-in, merge gates, and termination receipts.
by GTDataworks
Recovers tool and connector workflows after contract errors, timeouts, permission denial, stale state, and partial writes using bounded, mechanically checked state transitions.
by GTDataworks
A campaign-scoped response that protects identities and payment processes while preserving evidence and minimizing destructive mailbox actions.
by GTDataworks
A closed provenance graph that identifies affected artifacts and produces bounded quarantine, rebuild, and customer-notification decisions.
by GTDataworks
A prioritized cluster defense report with evidence-backed attack paths, safe remediation order, and verification commands proposed but not silently executed.
by GTDataworks
An evidence-backed identity attack graph, scoped containment plan, and verified tenant recovery without treating password reset as complete remediation.
by GTDataworks
Controls memory admission, retrieval, conflict, expiry, correction, and deletion while preserving source lineage and preventing stale or untrusted content from becoming authority.
by GTDataworks
A production-candidate detection package that proves syntax, behavior coverage, expected noise, and deployment safety.
by GTDataworks
A sequenced containment, identity, communications, evidence, restoration, and verification plan that protects critical services.
by GTDataworks
Externally adjudicates agent traces using blind projections, frozen rubrics, trace and replay integrity, independent graders, disagreement handling, and bounded verdicts.