Conversation
code for harness
…uard v0.3) Major restructuring of AgentGuard into a production-ready monorepo layout. Client (src/client/python/agentguard/): - Harness Runtime with LLM/Agent/Tool interception and lifecycle management - U-Guard dual-path engine: fast-path local policy + slow-path remote guard with PolicySnapshot sync, RemoteGuardClient, and offline fallback - Comprehensive checker suite (LLM input/output/thought, tool invoke/result, final response, memory) - Sandbox subsystem: NoopSandbox, LocalPermissionSandbox, SubprocessSandbox, SandboxExecutor with PermissionProfile - Tool registry, ToolWrapper, ToolDegradeManager; interceptor chain; client plugins with AgentDoGProxyPlugin as paired plugin example - Adapters for major LLM/agent frameworks (offline-safe defaults) - Root-level skills: DSLWriter, RuleLinter, PolicyExplainer, RuleTester, PolicySnapshotBuilder, TraceToRule, PolicyGapAnalyzer, RegressionTestGenerator and runtime skills (SafeRewrite, ToolRepair, ThoughtAlign, ObservationSanitize, ArgumentDegrade) - Structured audit: redactor, JSONL trace logger, recorder; CLI subcommands Server (src/server/backend/): - RuntimeManager with observer hook for real-time console data - PolicyEngine deny-overrides, PolicyStore, snapshot builder - AgentDoG server plugin: real trajectory judge using genuine AgentDoG prompt + OpenAI-compatible model endpoint (e.g. vLLM + AgentDoG checkpoint); HeuristicAgentDoGAdapter as offline fallback (real deterministic detector) - AgentDoGModelAdapter with graceful fallback on network/parse errors - LLM provider: real OpenAI-compatible HTTP + HeuristicProvider offline fallback (replaces former mock implementation) - Management console API (console_router): tools/labels CRUD, rules CRUD + check/reload, stats/traffic/audit/approvals backed by live decision data - Enriched /health endpoint; FastAPI app with CORS Shared, rules, plugins, examples: - Protocol contracts (RemoteGuardRequest/Response, PluginManifest) - Built-in JSON rule files; plugin manifests (agentdog paired plugin) - 7 runnable examples + cross-container e2e client Frontend integration: - frontend/app.py: mock_backend dependency optional; defaults to real backend - Console fully wired to real server data Infrastructure: - Dockerfile: PYTHONPATH monorepo layout, frontend included - docker-compose.yml: server + frontend one-click stack; e2e client in profile - scripts/entrypoint.sh: added frontend command - conftest.py: sys.path wiring for all test environments Tests: 34 pytest tests (schemas, checkers, local engine, sandbox, parser, skills, server manager, e2e HTTP, real adapters, console state); all pass. Archive: original agentguard/ moved to legacy/agentguard/; AgentDoG added as git submodule under third_party/AgentDoG.
refactor: migrate to client-server runtime security framework
Add phased local/remote checker configuration across client and server, runtime config update APIs, server-to-client config propagation, attach-based agent integration, shared schema/rule utilities, optional rule-based server checker, and checker documentation.
…nd add key verification for interaction between client and server
wrap_llm for autogen & langchain
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Automated PR created from dev/v2.0 into main.\n\nNote: do not delete the source branch after merge.