tollgate 1.0.5__py3-none-any.whl → 1.4.0__py3-none-any.whl

tollgate-1.4.0.dist-info/METADATA

@@ -0,0 +1,393 @@
+Metadata-Version: 2.4
+Name: tollgate
+Version: 1.4.0
+Summary: Runtime enforcement layer for AI agent tool calls using Identity + Intent + Policy
+Project-URL: Homepage, https://github.com/ravi-labs/tollgate
+Project-URL: Repository, https://github.com/ravi-labs/tollgate.git
+Project-URL: Documentation, https://github.com/ravi-labs/tollgate#readme
+Project-URL: Issues, https://github.com/ravi-labs/tollgate/issues
+Project-URL: Changelog, https://github.com/ravi-labs/tollgate/blob/main/CHANGELOG.md
+Author: Tollgate Maintainers
+License-Expression: Apache-2.0
+License-File: LICENSE
+Keywords: agents,ai,enforcement,llm,policy,security
+Classifier: Development Status :: 4 - Beta
+Classifier: Framework :: AsyncIO
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Security
+Requires-Python: >=3.10
+Requires-Dist: pyyaml>=6.0.0
+Provides-Extra: dev
+Requires-Dist: build>=1.0.0; extra == 'dev'
+Requires-Dist: pytest-asyncio>=0.21.0; extra == 'dev'
+Requires-Dist: pytest>=7.0.0; extra == 'dev'
+Requires-Dist: ruff>=0.1.0; extra == 'dev'
+Requires-Dist: twine>=4.0.0; extra == 'dev'
+Provides-Extra: redis
+Requires-Dist: redis[hiredis]>=5.0.0; extra == 'redis'
+Description-Content-Type: text/markdown
+
+# tollgate
+
+Runtime enforcement layer for AI agent tool calls using **Identity + Intent + Policy**.
+
+`tollgate` provides a deterministic safety boundary for AI agents. It ensures every tool call is validated against a policy before execution, with support for async human-in-the-loop approvals, framework interception (MCP, Strands, LangChain, OpenAI), and structured audit logging.
+
+> [!CAUTION]
+> **Disclaimer**: This project is an exploratory implementation intended for learning and discussion. It is not production-hardened and comes with no guarantees.
+
+**[Architecture](./ARCHITECTURE.md) | [Quickstart Guide](./QUICKSTART.md) | [Feature Guide](./FEATURES.md) | [Integration Comparison](./COMPARISON.md) | [Security](./SECURITY.md)**
+
+```
+                          +-----------------+
+                          |    AI Agent     |
+                          +--------+--------+
+                                   |
+                    +--------------v---------------+
+                    |     Tollgate ControlTower    |
+                    |------------------------------|
+                    | 1. Identity Verification     |
+                    | 2. Circuit Breaker Check     |
+                    | 3. Rate Limiting             |
+                    | 4. Policy Evaluation         |
+                    | 5. Network Guard             |
+                    | 6. Parameter Validation      |
+                    | 7. Constraint Checking       |
+                    +--------------+---------------+
+                                   |
+              +--------------------+--------------------+
+              |                    |                    |
+              v                    v                    v
+         +--------+           +--------+          +--------+
+         | ALLOW  |           |  ASK   |          |  DENY  |
+         +---+----+           +---+----+          +---+----+
+             |                    |                    |
+             v                    v                    v
+        +--------+        +-------------+        +--------+
+        |Execute |        |Human/Grant  |        | Block  |
+        | Tool   |        | Approval    |        | & Log  |
+        +---+----+        +------+------+        +---+----+
+             \                   |                   /
+              \                  |                  /
+               +--------+--------+--------+--------+
+                        |
+                        v
+                  +-----------+
+                  | Audit Log | --> Anomaly Detection
+                  +-----------+
+```
+
+## Installation
+
+```bash
+pip install tollgate
+
+# Optional: For Redis-backed persistent stores
+pip install tollgate[redis]
+```
+
+## Core Concepts
+
+| Concept | Description |
+|---------|-------------|
+| **AgentContext** | Who is asking? (`agent_id`, `version`, `owner`, `delegated_by`) |
+| **Intent** | What is the goal? (`action`, `reason`, `confidence`) |
+| **ToolRequest** | What tool and parameters? (`tool`, `action`, `effect`, `params`) |
+| **Decision** | Policy result: `ALLOW`, `ASK`, or `DENY` |
+| **Grant** | Pre-authorization that bypasses ASK for specific patterns |
+
+## Core Principles
+
+1. **Interception-First**: Enforcement at the tool execution boundary via adapters
+2. **Safe Defaults**: Unknown effects or resources default to **DENY**
+3. **Trust Model**: Tool metadata trusted only from developer-controlled **Tool Registry**
+4. **Approval Integrity**: Approvals bound to request hash with replay protection
+5. **Async-First**: Native async support with non-blocking approvals
+6. **Audit Integrity**: Every decision recorded with cryptographic context
+7. **Defense in Depth**: Multiple security layers (rate limiting, circuit breaker, network guard)
+
+## Quick Example
+
+```python
+import asyncio
+from tollgate import (
+    ControlTower, YamlPolicyEvaluator, ToolRegistry,
+    AutoApprover, JsonlAuditSink, AgentContext, Intent,
+    ToolRequest, Effect
+)
+
+async def main():
+    # Setup
+    policy = YamlPolicyEvaluator("policy.yaml")
+    registry = ToolRegistry("manifest.yaml")
+    audit = JsonlAuditSink("audit.jsonl")
+
+    tower = ControlTower(policy, AutoApprover(), audit, registry=registry)
+
+    # Define context
+    agent = AgentContext(agent_id="my-agent", version="1.0", owner="my-team")
+    intent = Intent(action="fetch_data", reason="User requested weather info")
+    request = ToolRequest(
+        tool="api:weather",
+        action="get",
+        resource_type="weather",
+        effect=Effect.READ,
+        params={"city": "London"},
+        manifest_version="1.0.0"
+    )
+
+    # Execute with enforcement
+    result = await tower.execute_async(
+        agent, intent, request,
+        lambda: fetch_weather("London")  # Your tool function
+    )
+    print(result)
+
+asyncio.run(main())
+```
+
+## Feature Highlights
+
+### Security Hardening
+
+| Feature | Description |
+|---------|-------------|
+| **Parameter Validation** | JSON Schema validation for tool parameters |
+| **Rate Limiting** | Per-agent, per-tool, per-effect rate limits |
+| **Agent Identity Signing** | HMAC-SHA256 verification of agent contexts |
+| **URL Constraints** | Per-tool URL allowlisting/blocklisting |
+| **Webhook Alerts** | Real-time alerts for blocked/denied events |
+| **Audit Schema Versioning** | Forward-compatible audit event schemas |
+
+```python
+from tollgate import InMemoryRateLimiter, sign_agent_context, make_verifier
+
+# Rate limiting
+limiter = InMemoryRateLimiter([
+    {"agent_id": "*", "tool": "api:*", "max_calls": 100, "window_seconds": 60},
+    {"agent_id": "*", "effect": "write", "max_calls": 10, "window_seconds": 60},
+])
+
+# Agent identity signing
+secret = b"your-secret-key"
+signed_agent = sign_agent_context(agent, secret)
+tower = ControlTower(..., verify_fn=make_verifier(secret), rate_limiter=limiter)
+```
+
+### Resilience & Protection
+
+| Feature | Description |
+|---------|-------------|
+| **Circuit Breaker** | Auto-disable failing tools after threshold |
+| **Manifest Signing** | HMAC-SHA256 integrity verification for manifests |
+| **NetworkGuard** | Global URL policy enforcement |
+| **Persistent Backends** | SQLite and Redis stores for grants/approvals |
+
+```python
+from tollgate import InMemoryCircuitBreaker, NetworkGuard, sign_manifest
+from tollgate.backends import SQLiteGrantStore
+
+# Circuit breaker - opens after 5 failures, 60s cooldown
+breaker = InMemoryCircuitBreaker(failure_threshold=5, cooldown_seconds=60)
+
+# Global network policy
+guard = NetworkGuard(
+    default="deny",
+    allowlist=[
+        {"pattern": "https://api.github.com/*"},
+        {"pattern": "https://api.openai.com/*"},
+    ],
+    blocklist=[
+        {"pattern": "http://*"},  # Block insecure
+    ]
+)
+
+# Sign your manifest at build time
+sign_manifest("manifest.yaml", secret_key=b"build-secret")
+
+# Load with signature verification
+registry = ToolRegistry("manifest.yaml", signing_key=b"build-secret")
+
+tower = ControlTower(..., circuit_breaker=breaker, network_guard=guard)
+```
+
+### Defense in Depth
+
+| Feature | Description |
+|---------|-------------|
+| **Multi-Agent Delegation** | Track and control delegation chains |
+| **Policy Testing Framework** | CI-friendly declarative policy testing |
+| **Context Integrity Monitor** | Detect memory/context poisoning |
+| **Anomaly Detection** | Z-score based rate spike detection |
+
+```python
+from tollgate import (
+    AgentContext, PolicyTestRunner,
+    ContextIntegrityMonitor, AnomalyDetector
+)
+
+# Delegation chain tracking
+sub_agent = AgentContext(
+    agent_id="sub-agent",
+    version="1.0",
+    owner="team-a",
+    delegated_by=("orchestrator", "router"),  # Delegation chain
+)
+print(sub_agent.delegation_depth)  # 2
+print(sub_agent.root_agent)        # "orchestrator"
+
+# Policy testing (CI integration)
+runner = PolicyTestRunner("policy.yaml", "test_scenarios.yaml")
+results = runner.run()
+assert results.all_passed  # Fail CI if policy regressed
+
+# Context integrity monitoring
+monitor = ContextIntegrityMonitor()
+monitor.snapshot("agent-1", "turn-1", {"system_prompt": "You are helpful..."})
+# Later...
+result = monitor.verify("agent-1", "turn-1", current_context)
+if not result.is_valid:
+    print(f"Context tampered! Changed: {result.changed_fields}")
+
+# Anomaly detection (plugs into audit pipeline)
+detector = AnomalyDetector(z_score_threshold=3.0)
+composite_sink = CompositeAuditSink([jsonl_sink, detector])
+```
+
+### CLI Tools
+
+```bash
+# Test policies against scenarios (for CI)
+tollgate test-policy policy.yaml --scenarios test_scenarios.yaml
+```
+
+## Framework Integrations
+
+### Session Grants
+Pre-authorize specific actions to bypass human approval:
+
+```python
+from tollgate import Grant, InMemoryGrantStore, Effect
+
+grant_store = InMemoryGrantStore()
+tower = ControlTower(..., grant_store=grant_store)
+
+# Issue a grant (after initial approval)
+grant = Grant(
+    agent_id="my-agent",
+    effect=Effect.WRITE,
+    tool="mcp:*",  # Wildcard: any MCP tool
+    action=None,   # Wildcard: any action
+    resource_type=None,
+    expires_at=time.time() + 3600,
+    granted_by="admin",
+    created_at=time.time()
+)
+await grant_store.create_grant(grant)
+```
+
+### MCP (Model Context Protocol)
+```python
+from tollgate.integrations.mcp import TollgateMCPClient
+
+registry = ToolRegistry("manifest.yaml")
+tower = ControlTower(...)
+client = TollgateMCPClient(base_client, "my_server", tower=tower, registry=registry)
+
+await client.call_tool("read_data", {"id": 1}, agent_ctx=ctx, intent=intent)
+```
+
+### Strands / LangChain
+```python
+from tollgate.integrations.strands import guard_tools
+
+guarded = guard_tools(my_tools, tower, registry)
+await guarded[0]("input", agent_ctx=ctx, intent=intent)
+```
+
+## Policy Example
+
+```yaml
+# policy.yaml
+version: "1.0"
+rules:
+  # Allow reads for verified agents
+  - id: allow_reads
+    effect: read
+    decision: ALLOW
+    reason: "Read operations are safe"
+
+  # Block deep delegation chains
+  - id: block_deep_delegation
+    effect: write
+    decision: DENY
+    reason: "Delegation too deep"
+    agent:
+      max_delegation_depth: 2
+
+  # Require approval for writes from untrusted delegators
+  - id: ask_untrusted_writes
+    effect: write
+    decision: ASK
+    reason: "Write requires human approval"
+    agent:
+      blocked_delegators:
+        - "untrusted-agent"
+
+  # Default deny
+  - id: deny_default
+    decision: DENY
+    reason: "No matching rule"
+```
+
+## Persistent Backends
+
+For production deployments, use persistent stores:
+
+```python
+# SQLite (zero dependencies, single-process)
+from tollgate.backends import SQLiteGrantStore, SQLiteApprovalStore
+
+grant_store = SQLiteGrantStore("tollgate.db")
+approval_store = SQLiteApprovalStore("tollgate.db")
+
+# Redis (multi-process, distributed)
+from tollgate.backends import RedisGrantStore, RedisApprovalStore
+
+grant_store = RedisGrantStore(redis_url="redis://localhost:6379/0")
+approval_store = RedisApprovalStore(redis_url="redis://localhost:6379/0")
+```
+
+## Development
+
+```bash
+# Install
+make install
+
+# Run Tests (188 tests)
+make test
+
+# Run Examples
+python examples/mcp_minimal/demo.py
+python examples/strands_minimal/demo.py
+python examples/mock_tickets/demo.py
+```
+
+## Documentation
+
+- **[QUICKSTART.md](./QUICKSTART.md)** - Get started in 5 minutes
+- **[FEATURES.md](./FEATURES.md)** - Complete feature guide
+- **[COMPARISON.md](./COMPARISON.md)** - Integration effort comparison
+- **[SECURITY.md](./SECURITY.md)** - Security model and guarantees
+- **[CHANGELOG.md](./CHANGELOG.md)** - Version history
+- **[CONTRIBUTING.md](./CONTRIBUTING.md)** - How to contribute
+
+## License
+
+Apache-2.0

tollgate-1.4.0.dist-info/RECORD

@@ -0,0 +1,33 @@
+tollgate/__init__.py,sha256=BFlGuAW4QPHeFkVKuVLAQAgUCYj6JBGIQLFJVoXV3oU,2624
+tollgate/anomaly_detector.py,sha256=3ODBwFNbRK4JpflrH94fgQXchI0X6LoMoZeWJBNN9NQ,13569
+tollgate/approvals.py,sha256=lalvun6i5yaZ2EGZ-apQJBrLVmht5OfwgiUFsImPdYE,7814
+tollgate/audit.py,sha256=ej99uu0lCRe1VRnY0Dkwpk0FE2F7gNIW5NfdpM0KcxA,4444
+tollgate/circuit_breaker.py,sha256=JEtNZLDnw38pRR1ZAA4Scig6ulDwbcACMyEizN6o6Kc,7014
+tollgate/context_monitor.py,sha256=sg13QA_UZTjsvDhZzvXWgqYVzpI1vc4PEJXwFh222bY,10013
+tollgate/exceptions.py,sha256=s8VoAZYcjqnSx56cqw4FmDEbuXaHLTxyzwAGWVjVz0A,1571
+tollgate/grants.py,sha256=H5soShiNcyQiY3gLJRqqP7ffAo2Udmnj3iU15sGi3os,5177
+tollgate/helpers.py,sha256=ZFMi19_ogpTlV_svFtgJ9kkmA1sPte1dmDnHYVdWGyo,1657
+tollgate/manifest_signing.py,sha256=PvJMqHX-iwhXf3a4vveVyFhsGRCfWpBJ7cHdU1ASQzo,2936
+tollgate/network_guard.py,sha256=pslR2utlRbrIcKfJt7A9mZE_vFQpS2HqRTWUL8bSCPY,4388
+tollgate/policy.py,sha256=wP-qfIoKdVsZY5RiXS6Yn9ESgSGFx0EmEVAtZz8OaAs,7757
+tollgate/policy_testing.py,sha256=lwSvm56UAoZwrWG7eKO-BUJ-pfEJHkICKHRGjBJRAzU,11652
+tollgate/rate_limiter.py,sha256=dTjAeOis2nJKpaECT9xLo5ErWABfzmJjf1OTdna2Shg,5470
+tollgate/registry.py,sha256=9TG65057NzA2bN9pG5bBTmEp_Jl68vVihcbJSn_XJJc,11370
+tollgate/tower.py,sha256=k2wGQPw7ojU4k_suptowJ2KEGFfRsbgL9sD2l68A2RY,14931
+tollgate/types.py,sha256=ZL-gmsG9dVi28U2zUCdHCVl73xYreqyuO1HrYCYafEQ,4551
+tollgate/verification.py,sha256=EBncuJ834ILquUCOdbdZ0X0nJABblnCJHxRNYT7MdnY,2437
+tollgate/backends/__init__.py,sha256=IbsabqDZ9dlq9XeM9twMrPBREElg4eTOxzZ0YqxpQcA,1164
+tollgate/backends/redis_store.py,sha256=vKhqBbGgCknnKqklZ8d_E_eMUNubu6bdf4gO7DIiHY8,13445
+tollgate/backends/sqlite_store.py,sha256=LQxVuF_LnQXobEP_gBZIBWINX4mT7gfs3sjwEMb4jKE,14879
+tollgate/integrations/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+tollgate/integrations/mcp.py,sha256=jBy1d64Yc55fO4JSVfhKqiffVCZdBMmmNLRta9SVfDE,1750
+tollgate/integrations/strands.py,sha256=_AWsIyva_iwLFIQeTMxxbEzLSjneo0YumWfY9oayKeM,3003
+tollgate/interceptors/__init__.py,sha256=0c3MYyVKGYrBOZ1mMolgrAowrqZrULzAl0vv-s8IFFg,305
+tollgate/interceptors/base.py,sha256=uJxHzH0eurcGEVknbk1kQkk_2u2qNtnM--ZRCp52Wyo,1390
+tollgate/interceptors/langchain.py,sha256=_8vXCjWkRKeTlxtXm33a67Gf4po9YiHS7faThMJLohc,2963
+tollgate/interceptors/openai.py,sha256=cHnOQ8keQJRYBm_1RpohKWk3D9Ask22hpuQghm9iahU,3337
+tollgate-1.4.0.dist-info/METADATA,sha256=eo3_xCfyEb35-jv61Cpi2TrSmDLGG7AAKr-ifNtRgr4,13070
+tollgate-1.4.0.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+tollgate-1.4.0.dist-info/entry_points.txt,sha256=trBcUVx2EZAAVORkTpGcoUXtoal-_GeCp8BZshA3gm0,62
+tollgate-1.4.0.dist-info/licenses/LICENSE,sha256=EZ9SehMCkcatlggcoT7WV0tx-ku4OsAoQf9LmJZvG1g,10806
+tollgate-1.4.0.dist-info/RECORD,,

tollgate-1.4.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+tollgate = tollgate.policy_testing:cli_main

tollgate-1.0.5.dist-info/METADATA

@@ -1,144 +0,0 @@
-Metadata-Version: 2.4
-Name: tollgate
-Version: 1.0.5
-Summary: Runtime enforcement layer for AI agent tool calls using Identity + Intent + Policy
-Author: Tollgate Maintainers
-License-Expression: Apache-2.0
-License-File: LICENSE
-Keywords: agents,ai,enforcement,llm,policy,security
-Classifier: Development Status :: 3 - Alpha
-Classifier: Intended Audience :: Developers
-Classifier: License :: OSI Approved :: Apache Software License
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3.12
-Requires-Python: >=3.10
-Requires-Dist: pyyaml>=6.0.0
-Provides-Extra: dev
-Requires-Dist: build>=1.0.0; extra == 'dev'
-Requires-Dist: pytest-asyncio>=0.21.0; extra == 'dev'
-Requires-Dist: pytest>=7.0.0; extra == 'dev'
-Requires-Dist: ruff>=0.1.0; extra == 'dev'
-Requires-Dist: twine>=4.0.0; extra == 'dev'
-Description-Content-Type: text/markdown
-
-# tollgate 🚪
-
-Runtime enforcement layer for AI agent tool calls using **Identity + Intent + Policy**.
-
-`tollgate` provides a deterministic safety boundary for AI agents. It ensures every tool call is validated against a policy before execution, with support for async human-in-the-loop approvals, framework interception (MCP, Strands, LangChain, OpenAI), and structured audit logging.
-
-> [!CAUTION]
-> **Disclaimer**: This project is an exploratory implementation intended for learning and discussion. It is not production-hardened and comes with no guarantees.
-
-**[🚀 Quickstart Guide](https://github.com/ravi-labs/tollgate/blob/main/QUICKSTART.md) | [📊 Integration Comparison](https://github.com/ravi-labs/tollgate/blob/main/COMPARISON.md)**
-
-```
-┌────────────┐     ┌─────────────────────┐     ┌──────────────────┐
-│  AI Agent  │────▶│ Tollgate Interceptor│────▶│ Policy + Registry│
-└────────────┘     └─────────────────────┘     └────────┬─────────┘
-                                                        │
-                   ┌────────────────────────────────────┼────────────────────────────────────┐
-                   │                                    │                                    │
-                   ▼                                    ▼                                    ▼
-            ┌───────────┐                        ┌───────────┐                        ┌───────────┐
-            │   ALLOW   │                        │    ASK    │                        │   DENY    │
-            └─────┬─────┘                        └─────┬─────┘                        └─────┬─────┘
-                  │                                    │                                    │
-                  ▼                                    ▼                                    ▼
-            ┌───────────┐                     ┌───────────────┐                      ┌───────────┐
-            │  Execute  │                     │Human Approval │                      │  Block &  │
-            │   Tool    │                     │  (Approved?)  │                      │   Audit   │
-            └─────┬─────┘                     └───────┬───────┘                      └─────┬─────┘
-                  │                              ▼         ▼                               │
-                  │                          Yes │         │ No                            │
-                  │                              ▼         ▼                               │
-                  │                        ┌─────────┐ ┌─────────┐                         │
-                  │                        │ Execute │ │  Block  │                         │
-                  │                        └────┬────┘ └────┬────┘                         │
-                  │                             │           │                              │
-                  └─────────────────────────────┴───────────┴──────────────────────────────┘
-                                                        │
-                                                        ▼
-                                                 ┌───────────┐
-                                                 │ Audit Log │
-                                                 └───────────┘
-```
-
-## ✨ v1 Core Principles
-
-1. **Interception-First**: Enforcement happens at the tool execution boundary via adapters.
-2. **Safe Defaults**: Any unknown tool effect or resource defaults to **DENY**.
-3. **Trust Model**: Tool metadata is trusted only if it comes from a developer-controlled **Tool Registry**.
-4. **Approval Integrity**: Approvals are bound to a request hash and correlation ID with replay protection.
-5. **Async-First**: Native support for asynchronous agent loops and non-blocking approvals.
-6. **Audit Integrity**: Every decision, approval, and outcome is recorded with full cryptographic context.
-
-## 🚀 v1 Integrations
-
-### 🎟️ Session Grants
-Grants allow you to pre-authorize specific actions for an agent session, bypassing human-in-the-loop approvals for repetitive or low-risk tasks.
-
-```python
-from tollgate import Grant, InMemoryGrantStore, Effect
-
-# 1. Setup a grant store
-grant_store = InMemoryGrantStore()
-tower = ControlTower(..., grant_store=grant_store)
-
-# 2. Issue a grant (e.g., after initial human approval)
-grant = Grant(
-    agent_id="my-agent",
-    effect=Effect.WRITE,
-    tool="mcp:*", # Wildcard prefix: matches any MCP tool (e.g., "mcp:server.write")
-    action=None,  # Wildcard: matches any action
-    resource_type=None,
-    expires_at=time.time() + 3600, # Valid for 1 hour
-    granted_by="admin-user",
-    created_at=time.time()
-)
-await grant_store.create_grant(grant)
-```
-
-### MCP (Model Context Protocol)
-Wrap an MCP client to gate all tool calls:
-```python
-from tollgate import ControlTower, ToolRegistry
-from tollgate.integrations.mcp import TollgateMCPClient
-
-registry = ToolRegistry("manifest.yaml")
-tower = ControlTower(...)
-client = TollgateMCPClient(base_client, server_name="my_server", tower=tower, registry=registry)
-
-# Calls are now gated!
-await client.call_tool("read_data", {"id": 1}, agent_ctx=ctx, intent=intent)
-```
-
-### Strands Agents
-Gate Strands tools with minimal friction:
-```python
-from tollgate.integrations.strands import guard_tools
-
-guarded = guard_tools(my_strands_tools, tower, registry)
-
-# Use guarded tools in your agent
-await guarded[0]("input", agent_ctx=ctx, intent=intent)
-```
-
-## 📜 Development
-
-```bash
-# Install
-make install
-
-# Run Tests
-make test
-
-# Run Examples (non-interactive)
-python examples/mcp_minimal/demo.py
-python examples/strands_minimal/demo.py
-```
-
-## ⚖️ License
-Apache-2.0

tollgate-1.0.5.dist-info/RECORD

@@ -1,21 +0,0 @@
-tollgate/__init__.py,sha256=KWxtUI6D0Cm4QHCAaVv5VdEiQIo_krw7jzbXofxW_FM,1441
-tollgate/approvals.py,sha256=lalvun6i5yaZ2EGZ-apQJBrLVmht5OfwgiUFsImPdYE,7814
-tollgate/audit.py,sha256=ugMhuuLoyBNdYD2S_MjGN_ac4nHdtRz15MElqElREIs,1279
-tollgate/exceptions.py,sha256=2yYY3esnHz26dyJlx_Cd_J64ryhexrgc4KliDmiVJSs,882
-tollgate/grants.py,sha256=H5soShiNcyQiY3gLJRqqP7ffAo2Udmnj3iU15sGi3os,5177
-tollgate/helpers.py,sha256=ZFMi19_ogpTlV_svFtgJ9kkmA1sPte1dmDnHYVdWGyo,1657
-tollgate/policy.py,sha256=1_6HcbdpGfZtwmgqapqeRzItc_wbEg_L-fOA-BVIWNg,6110
-tollgate/registry.py,sha256=ENKT-GnwFq0xXZj6qNlKFhAGYsxxBwtZsAdPZp2qlHM,2031
-tollgate/tower.py,sha256=rPG2sdVe2GpdGiyHwWz8K4k7EJCL2NqDAGsMUkKlt30,8574
-tollgate/types.py,sha256=Ks3HNawhPzEAggTCab7NLb8zl06777ehEusASJWSj_g,3811
-tollgate/integrations/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-tollgate/integrations/mcp.py,sha256=jBy1d64Yc55fO4JSVfhKqiffVCZdBMmmNLRta9SVfDE,1750
-tollgate/integrations/strands.py,sha256=_AWsIyva_iwLFIQeTMxxbEzLSjneo0YumWfY9oayKeM,3003
-tollgate/interceptors/__init__.py,sha256=0c3MYyVKGYrBOZ1mMolgrAowrqZrULzAl0vv-s8IFFg,305
-tollgate/interceptors/base.py,sha256=uJxHzH0eurcGEVknbk1kQkk_2u2qNtnM--ZRCp52Wyo,1390
-tollgate/interceptors/langchain.py,sha256=_8vXCjWkRKeTlxtXm33a67Gf4po9YiHS7faThMJLohc,2963
-tollgate/interceptors/openai.py,sha256=cHnOQ8keQJRYBm_1RpohKWk3D9Ask22hpuQghm9iahU,3337
-tollgate-1.0.5.dist-info/METADATA,sha256=zPDa8CQ7Qqx5GoQ6XojAZgur47ErP6KeWoUCfhTQOXs,7748
-tollgate-1.0.5.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-tollgate-1.0.5.dist-info/licenses/LICENSE,sha256=EZ9SehMCkcatlggcoT7WV0tx-ku4OsAoQf9LmJZvG1g,10806
-tollgate-1.0.5.dist-info/RECORD,,