{"id":51721125,"url":"https://github.com/danieljustus/symaira-guard","last_synced_at":"2026-07-17T11:40:58.708Z","repository":{"id":367596416,"uuid":"1281481116","full_name":"danieljustus/symaira-guard","owner":"danieljustus","description":"A local-first security gateway for AI agents, MCP servers, and Symaira toolchains.","archived":false,"fork":false,"pushed_at":"2026-06-26T18:03:29.000Z","size":41,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-06-26T18:09:08.563Z","etag":null,"topics":["ai-security","go","llm","mcp","policy-engine"],"latest_commit_sha":null,"homepage":"https://symaira.com/","language":"Go","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/danieljustus.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":".github/CODEOWNERS","security":".github/SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2026-06-26T15:44:24.000Z","updated_at":"2026-06-26T17:49:02.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/danieljustus/symaira-guard","commit_stats":null,"previous_names":["danieljustus/symaira-guard"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/danieljustus/symaira-guard","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/danieljustus%2Fsymaira-guard","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/danieljustus%2Fsymaira-guard/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/danieljustus%2Fsymaira-guard/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/danieljustus%2Fsymaira-guard/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/danieljustus","download_url":"https://codeload.github.com/danieljustus/symaira-guard/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/danieljustus%2Fsymaira-guard/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35580669,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-07-17T02:00:06.162Z","response_time":116,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai-security","go","llm","mcp","policy-engine"],"created_at":"2026-07-17T11:40:57.994Z","updated_at":"2026-07-17T11:40:58.702Z","avatar_url":"https://github.com/danieljustus.png","language":"Go","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Symaira Guard (`symguard`)\n\n\u003e A local-first security gateway for AI agents, MCP servers, and Symaira toolchains.\n\n**Human control for agent autonomy.**\n\n[![CI](https://github.com/danieljustus/symaira-guard/actions/workflows/ci.yml/badge.svg)](https://github.com/danieljustus/symaira-guard/actions/workflows/ci.yml)\n[![Go](https://img.shields.io/github/go-mod/go-version/danieljustus/symaira-guard)](https://go.dev/)\n[![License](https://img.shields.io/github/license/danieljustus/symaira-guard)](https://github.com/danieljustus/symaira-guard/blob/main/LICENSE)\n\n---\n\n## What\n\n`symguard` sits between AI clients and the tools they call. It inspects every MCP tool call, classifies risk, enforces local policy, asks for human approval when needed, and records tamper-evident audit trails.\n\n```\nAI client / agent  →  symguard  →  MCP servers / CLIs / APIs / Symaira tools\n```\n\nThe agent still gets useful tools. The human keeps enforceable boundaries.\n\n## Why\n\nMCP solved interoperability between AI clients and tool servers. It did not solve:\n\n- Tool poisoning and rug-pull attacks (changed tool descriptions)\n- Prompt injection escalating into tool calls\n- Unbounded shell / filesystem / network / secret access\n- Missing human approval for risky operations\n- Secret exfiltration through tool output\n- Cross-agent delegation risk\n\n`symguard` is the missing local control layer.\n\n## Quick demo\n\n```bash\n$ symguard version\nsymguard v0.1.0 (commit abc1234, built with Go 1.26)\n\n$ symguard doctor\n✓ CLI config found at ~/.config/symguard/config.toml\n✓ XDG paths OK\n✓ Go runtime 1.26+ detected\n✓ MCP clients found: hermes, claude, cursor\n→ For more detail: symguard scan\n\n$ symguard scan --client hermes\n✓ Scanned Hermes (3 MCP servers, 17 tools)\n\n  symmemory     memory_search, memory_set, entity_list    1 tool changed\n  symvault      get, set, search, health                  4 tools, stable\n  filesystem    read, write, grep, glob, ...              7 new tools\n\nPolicy summary: 12 tools allow, 4 tools ask, 1 tool deny\n```\n\n## What it does\n\n### 1. Scan\n\nDiscover MCP servers configured across local AI clients and classify their tools by risk.\n\n```bash\nsymguard scan                        # scan all clients\nsymguard scan --client hermes         # scan one client\nsymguard scan --format json           # machine-readable output\n```\n\n### 2. Policy\n\nDefine local rules that decide what gets through:\n\n```toml\n[defaults]\nshell = \"ask\"\nread_secret = \"deny\"\nwrite_file = \"ask\"\n\n[[rules]]\nmatch.server = \"symmemory\"\nmatch.tool = \"memory_search\"\ndecision = \"allow\"\n\n[[rules]]\nmatch.command_contains = [\"rm -rf\", \"curl | sh\"]\ndecision = \"deny\"\n```\n\nDecisions: `allow`, `ask`, `deny`, `redact`, `readonly`, `sandbox`.\n\n### 3. Proxy\n\nRun as an MCP proxy that enforces policy per tool call:\n\n```bash\nsymguard proxy --config ~/.config/symguard/config.toml\n```\n\nEach tool call is classified, policy-checked, optionally approved by a human, then forwarded upstream. Sensitive output can be redacted before it reaches the agent.\n\n### 4. Pin\n\nStore hashes of MCP tool descriptions and schemas. If a tool's description changes (hidden instructions, scope expansion), `symguard` flags it:\n\n```\nWARNING: Tool schema changed for server \"filesystem\" tool \"read_file\".\nPolicy: require re-approval\n```\n\n### 5. Audit\n\nAppend-only local audit log with hash chaining. Records what was requested, which policy matched, who approved, what executed, and what came back.\n\n### 6. Remote access\n\nLater phases add agent-aware remote MCP access over existing transports (SSH, Tailscale, LAN/mDNS) — not a new VPN, but policy and audit on top of tools you already trust.\n\n## Risk classes\n\n| Risk | Examples | Default |\n|------|----------|---------|\n| `read_public` | docs, README, public web | allow |\n| `read_private` | repo files, notes, local docs | allow or ask |\n| `read_secret` | `.env`, SSH keys, vault entries | ask / deny |\n| `write_file` | patch, overwrite, create file | ask |\n| `shell` | command execution | ask |\n| `network` | outbound API / web requests | ask |\n| `browser` | cookies, sessions, web automation | ask |\n| `credential_use` | using secrets without revealing them | ask once / scoped |\n| `deploy` | release, push, infra mutation | ask every time |\n| `destructive` | delete, wipe, reset, revoke | ask / deny |\n\n## Symaira ecosystem position\n\n`symguard` is a **public, self-hosted core** tool. No Pro, tenant, or billing code.\n\n```\n┌─────────────────────────────────────────┐\n│ AI clients / agents                     │\n│ Hermes · Claude · Cursor · OpenCode ... │\n└───────────────────┬─────────────────────┘\n                    ▼\n            ┌──────────────┐\n            │  symguard    │  ← trust boundary\n            └──────┬───────┘\n                   ▼\n    symvault · symmemory · symscope · symseek · ...\n```\n\nOptional runtime integrations, no compile-time dependencies on siblings.\n\n## Principles\n\n- **Local-first.** Policy decisions happen on your machine. No mandatory cloud account.\n- **Boring is good.** No custom VPN, no NAT traversal, no WireGuard daemon. Reuse existing transports.\n- **Discovery ≠ trust.** Finding a remote MCP server never auto-implies permission.\n- **Agent identities.** Agents and runs are first-class identities with TTL and scoped grants.\n- **Explainable.** Every decision has a reason. Simulate before acting. Diagnose after failing.\n\n## Non-goals\n\nNot a chat frontend, not a SIEM, not a cloud-only SaaS, not a VPN replacement, not a full endpoint protection platform.\n\n\u003e Classify agent tool calls, enforce local policy, ask the human when needed, and record what happened.\n\n---\n\n## Build\n\nRequires Go 1.26+. No external dependencies — only the Go standard library.\n\n```bash\n# Build the binary\nmake build\n\n# Run tests\nmake test\n\n# Lint (golangci-lint or go vet fallback)\nmake lint\n\n# Set a version string at build time\nmake build VERSION=v1.0.0\n```\n\nOr directly with `go`:\n\n```bash\ngo build -ldflags \"-X main.version=dev\" -o symguard ./cmd/symguard\ngo vet ./...\ngo test ./...\n```\n\n### Quick start\n\n```bash\n./symguard version    # print version and build info\n./symguard doctor     # check system health\n```\n\n## Status\n\nEarly development. See [docs/intern/IDEA.md](docs/intern/IDEA.md) for the full design document.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdanieljustus%2Fsymaira-guard","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fdanieljustus%2Fsymaira-guard","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdanieljustus%2Fsymaira-guard/lists"}