Gadriel AI Security Harness

Plugin gadriel in Gadriel-ai/gadriel-codex-plugin. Commands read from the repo's manifest — 12 days ago.

Skills (22)

• gadriel-a2a-contracts

  Agent-to-agent contract patterns — ack/timeout, idempotency keys, message versioning, signed envelopes. Auto-invoke for findings tagged `teamwork`, `a2a`, `agent-contract`, or rule IDs `CODE-W1-AI-6**` where the AST shows agent-to-agent message passing.

• gadriel-ai-config-security

  LangChain, CrewAI, MCP, and direct-SDK configuration security — temperature/top_p sanity, system-prompt leak prevention, tool-exposure review. Auto-invoke for findings tagged `ai-config`, `model-config`, or rule IDs `CODE-W1-CONFIG-*`, and when the user asks about "safe defaults", "system prompt pro

• gadriel-ai-secrets-catalog

  AI-native secrets catalog — API key, token, and credential patterns for every major model provider and vector-DB. Auto-invoke for findings tagged `secret-leak`, `hardcoded-credential`, or rule IDs `CODE-W1-L1-*` when the leaked material is an LLM/AI provider key.

• gadriel-api-security-patterns

  REST, GraphQL, gRPC, and OpenAPI security patterns — authn, authz, rate limits, schema validation, mass assignment. Auto-invoke for findings tagged `api-security` or rule IDs `CODE-W1-API-*`, and when the user asks about "API hardening", "GraphQL query depth", "rate limit", "OpenAPI security".

• gadriel-bayesian-calibration

  Bayesian calibration for Gadriel verdicts — interpreting the {{bayes_prior}} prompt slot, reliability diagrams, ECE, post-hoc calibration (Platt, isotonic, temperature scaling). Auto-invokes for ALL agents on every finding; this is the cross-cutting skill shared by every pillar.

• gadriel-deadlock-resolution

  Multi-agent deadlock patterns — circular waits, supervisor-loop starvation, tool-contention. Auto-invoke for findings tagged `teamwork`, `deadlock`, `livelock`, `starvation`, or rule IDs `CODE-W1-AI-6**` where the graph contains cycles or shared mutexes.

• gadriel-diff-scan

  Use to security-scan only the files changed in the working tree or a diff — the fast pre-PR / pre-commit check. Scans the changed paths with Gadriel instead of the whole repo. Use when the user asks to scan changes, the diff, staged files, or 'before I open a PR'. For a full repository scan use gadr

• gadriel-dockerfile-best-practices

  Container security and efficiency patterns — CIS Docker Benchmark, root user, latest tag, secret-bake, image-bloat. Auto-invoke for findings tagged `container`, `dockerfile`, or rule IDs `CODE-W1-CONTAINER-*`, and when the user asks about "Dockerfile review", "Kubernetes hardening", "image scan".

• gadriel-eu-ai-act-mapper

  EU AI Act article-by-article mapping — translate Gadriel findings into the obligations they trigger (Art. 9 risk-mgmt, Art. 10 data governance, Art. 13 transparency, Art. 14 human oversight, Art. 15 accuracy/robustness/cybersec). Auto-invoke for compliance findings or when the user asks about "EU AI

• gadriel-fix

  Use to remediate ONE specific Gadriel finding identified by its rule ID (e.g. CODE-W1-L3-017). Applies or proposes a fix and re-scans the affected file to confirm the finding is resolved. Use when the user asks to fix, remediate, or resolve a named Gadriel finding. To discover findings first, use ga

• gadriel-graph-attack-patterns

  Multi-agent graph-attack catalog — prompt injection across hand-offs, cross-agent context smuggling, tool-chain hijack. Auto-invoke for findings tagged `graph-attack`, `cross-agent`, or rule IDs `CODE-W1-AI-6**` when the AST shows agent-to-agent edges (CrewAI Task→Agent, LangGraph nodes, Anthropic A

• gadriel-hitl-patterns

  Human-in-the-loop gate patterns — autonomy boundaries, approval gates, audit trails for high-impact actions. Auto-invoke for findings tagged `safety`, `autonomy-gate`, `hitl`, or rule IDs `CODE-W1-AI-6**` where the agent has high-impact tools (shell, financial, deletion).

• gadriel-license-compatibility

  Open-source license compatibility matrix — copyleft propagation, viral GPL detection, license-incompatible imports. Auto-invoke for findings tagged `license`, `license-incompat`, or rule IDs `CODE-W1-L2-*` when the matched dep has a flagged license.

• gadriel-nist-ai-rmf-mapper

  NIST AI Risk Management Framework (AI RMF 1.0) function-by-function mapping — Govern, Map, Measure, Manage. Auto-invoke for compliance findings, or when the user asks about "NIST AI RMF", "AI governance", "AI risk assessment".

• gadriel-output-schema-library

  JSON Schema patterns for structured LLM output — strict schemas, refusal handling, retry-on-parse-fail, common gotchas. Auto-invoke for findings tagged `coherence`, `schema`, `structured-output`, or rule IDs `CODE-W1-AI-6**` where the agent emits machine-parsable output.

• gadriel-owasp-llm-top10

  OWASP LLM Top 10 (2025) catalog with detection patterns and remediations. Auto-invoke for findings tagged `owasp-llm-*`, `prompt-injection`, `insecure-output-handling`, or rule IDs `CODE-W1-AI-*`, when the user asks about "LLM vulnerabilities", "prompt injection", or "model security".

• gadriel-owasp-web-top10

  OWASP Web Top 10 (A01-A10, 2021) detection patterns and language-specific remediations. Auto-invoke for findings tagged `owasp-web-*` or rule IDs `CODE-W1-L3-*` (injection/auth/access-control), and when the user asks about "SQL injection", "broken access control", "XSS", or general web vulnerabiliti

• gadriel-reports

  Use to generate Gadriel's compliance reports (PDF) from the latest scan — one per pillar plus an umbrella report, mapped to frameworks such as OWASP LLM Top 10, SOC 2, HIPAA, EU AI Act and NIST AI RMF. Use when the user asks for a compliance report, audit report, or framework-mapped evidence. Requir

• gadriel-sbom-guidance

  SBOM generation guidance — SPDX vs CycloneDX, when each is required, common gotchas (transitive deps, base-image layers, AIBOM). Auto-invoke for compliance findings tagged `sbom`, supply-chain findings, or when the user asks about "SBOM", "SPDX", "CycloneDX", "AIBOM", "Executive Order 14028".

• gadriel-scan

  Use to run a Gadriel security scan of the whole repository or a scoped path/package/folder/file. This is the default full scan: SAST, secrets, dependencies (SCA), containers and configuration, including AI-specific risks (OWASP LLM Top 10). It writes results to .security/ and arms Gadriel's edit gua

• gadriel-status

  Use to show the open Gadriel findings from the last scan without re-scanning — counts by severity and the top findings. Use when the user asks 'is this repo clean?', for the current security status, or the open findings. Reads .security/; if there is no scan yet, suggest gadriel-scan.

• gadriel-token-cost-estimator

  Token-cost math, per-call pricing, ceiling enforcement, model-tier selection. Auto-invoke for findings tagged `finops`, `cost-anomaly`, `token-budget`, or rule IDs `CODE-W1-COST-*`, and when the user asks about "token cost", "pricing", "context-window optimization", "model tier".

MCP servers (1)

• gadriel

  runs: ./bin/gadriel code mcp

Similar plugins