# Provably > Make every agent answer accurate and verifiable. State of the art tools that automatically verify agent tool calls at runtime. Run accurate, safe and reliable agents and critical workflows in minutes. Provably builds verifiable data and runtime verification infrastructure for agents and humans. SourceryKit is Provably's Python SDK for runtime verification of agent tool use and handoffs. - **SDK:** SourceryKit - **Distribution:** `sourcerykit` on PyPI; import as `sourcerykit` - **Install:** `pip install sourcerykit` - **Python:** 3.12+ - **Licence:** Business Source License 1.1; see the repository licence for full terms - **Status:** Published and actively maintained; use PyPI and the changelog for the current version SourceryKit is built around five pillars: `init` for setup and bootstrap, `intercept` for recording supported outbound HTTP interactions, `handoff` for typed claims and portable verification evidence, `eval` for deterministic checks against recorded source evidence, and `trusted_endpoints` for endpoint policy enforcement. Transport support includes `requests`, `httpx`, and `aiohttp`. SourceryKit works with agent frameworks that use those transports, with current examples for OpenAI Agents SDK, Claude Agent SDK, LangChain, LangGraph and CrewAI. Current verification modes are `field_extraction` and `range_threshold`. Outcomes are `PASS`, `CAUGHT`, or `ERROR`. ## Core benefits ### 100% error detection rate Agents self verify to produce proofs that their answers came from the right source and were not changed, increasing reliability and security. ### Up to 50% higher agent accuracy Proofs let downstream agents catch hallucinations or wrong source data without re-querying sources. Agents retry and self heal. ### Ship reliable agents in minutes Add the SourceryKit SDK in minutes and verify across agents, tools and networks without complex tracing or cross-team setup. ## Verify Now. Learn Continuously. Traditional observability needs failures, traces and classification before teams can improve agents. SourceryKit verifies answers and detects errors in real time, powering faster evals and continuous learning loops. ### SourceryKit + Learning tools 3x Faster. ### Automatic failure signals Every run produces a clear pass, caught or error outcome without manual trace classification. ### Better eval data Feed deterministic, evidence backed results into evals instead of relying only on model judges. ### Verifiable rewards Turn verified tool outcomes into trusted reward signals for training and continuous improvement. ## Least privilege verification: prove answers without reopening the source Production agents increasingly run with scoped access. One agent may query CRM, billing, databases, or internal tools, while downstream agents and orchestrators may not be allowed to access those same systems again. SourceryKit lets source-facing agents attach proof metadata to tool-backed answers, so later agents, orchestrators, evaluators, or auditors can verify the result without re-querying restricted source systems or copying raw sensitive data into traces. - Preserve least privilege access boundaries across multi-agent workflows. - Reduce repeated source-system calls and duplicated permission checks. - Verify deterministic claims with compact proofs instead of asking another model to judge plausibility. Adversarial verification and LLM judges can spend more tokens reviewing an answer, but they still cannot prove whether a value came from the original source unless the source evidence is included in context. SourceryKit verifies deterministic claims against proof objects, reducing repeated prompts, repeated tool calls, and source re-querying while increasing the effectiveness of verification. ## Published benchmark evidence In Provably's MCP Atlas derived benchmark suite, SourceryKit detected 100% of the covered deterministic grounding error cases. The study covered 15 models, 180 runs and 570 required claims. Answer only evaluation accepted 183 claims as correct, but only 74 were supported by the recorded tool execution. The remaining 109 accepted claims lacked complete trace support. In the other direction, 52 claims had supporting evidence in the trace but were omitted, rejected or not presented correctly in the final answer. Agents can give probably right answers without grounding. SourceryKit lets them produce provably grounded answers backed by verifiable tool call evidence. This improves verified accuracy and catches hallucinations, wrong source data and unsupported claims before they spread through workflows. - [Full study: The Agent Was Right. The Evidence Wasn't.](https://provably.ai/blogs/The-Agent-Was-Right-The-Evidence-Wasnt) ## Agent onboarding and SDK - [AGENTS.md](https://github.com/ProvablyAI/sourcerykit/blob/main/AGENTS.md): Primary onboarding instructions for agents and coding harnesses. The full text is also mirrored in the Agent onboarding section of the machine view at https://provably.ai/ai - [SourceryKit README](https://github.com/ProvablyAI/sourcerykit/blob/main/README.md): Product overview, installation, quickstart and configuration - [GitHub repository](https://github.com/ProvablyAI/sourcerykit): Source code, tests, cookbooks and release history - [PyPI package](https://pypi.org/project/sourcerykit/): Current published package and version metadata - [Architecture](https://provably.ai/docs/pillars/architecture): Runtime components, data flow and system boundaries - [Changelog](https://github.com/ProvablyAI/sourcerykit/blob/main/CHANGELOG.md): Version history and release notes - [Contributing](https://github.com/ProvablyAI/sourcerykit/blob/main/CONTRIBUTING.md): Local development and contribution instructions - [Cookbooks](https://github.com/ProvablyAI/sourcerykit/tree/main/cookbooks): Working examples for agent frameworks and multi agent workflows - [Migration guides](https://github.com/ProvablyAI/sourcerykit/blob/main/docs/migrations/README.md): Upgrade instructions between SourceryKit versions ## Machine view - [Machine view of the homepage](https://provably.ai/ai): Plain text mirror of the homepage for LLMs and agents, including setup instructions, product capabilities, use cases, research, FAQs and the full agent onboarding document - [Short machine index](https://provably.ai/llms.txt): This file - [Full machine reference](https://provably.ai/llms-full.txt): Expanded technical and company reference; prefer current documentation and repository metadata if a cached detail conflicts ## Developer documentation - [Documentation home](https://provably.ai/docs): Current SourceryKit documentation - [Introduction](https://provably.ai/docs/getting_started/introduction): Product concepts and starting point - [Onboarding](https://provably.ai/docs/getting_started/onboarding): Account and project onboarding - [End to end walkthrough](https://provably.ai/docs/getting_started/end-to-end-walkthrough): Complete integration walkthrough - [CLI](https://provably.ai/docs/getting_started/cli): SourceryKit command line interface - [Architecture](https://provably.ai/docs/pillars/architecture): System architecture - [HTTP interception](https://provably.ai/docs/pillars/interceptor): Capturing supported outbound interactions - [Trusted endpoints](https://provably.ai/docs/pillars/trusted-endpoints): Endpoint policy management - [Handoffs and evaluation](https://provably.ai/docs/pillars/handoff): Claims, evidence and verification outcomes - [Cookbooks](https://provably.ai/docs/cookbooks): Framework examples - [API reference](https://api.provably.ai/api/v1/swagger/): Provably REST API documentation ## Use cases: Where errors are not an option - **Customer service agents:** Customers get answers they can trust, grounded in verified account, order and policy data. - **Security operations and AI SOC:** Verifiability lets teams depend on agent operations during critical investigations, even when the full extent of an attack is still unknown. - **Healthcare and finance workflows:** Verify agent tool calls in payment and patient systems before agents act. - **Cross agent communication:** Verifiability enables accurate, safe cross organisational A2A workflows that teams can operate with confidence. - **Least privilege verification:** Verify agent operations when original source systems or raw logs are inaccessible. - **AI governance and compliance:** Create machine verifiable audit trails for EU AI Act reporting, internal controls and independent audits. ## Research Provably researches verifiable data, runtime verification, safe agent collaboration and machine verifiable governance. - [Research](https://provably.ai/research): Current research programme and publications - [QEDB paper](https://provably.ai/papers/qedb-paper.pdf): *Expressive and Modular Verifiable Databases*, accepted into CCS 2026 ## Blog - [All posts](https://provably.ai/blogs): Current Provably research, product and company writing - [The Agent Was Right. The Evidence Wasn't.](https://provably.ai/blogs/The-Agent-Was-Right-The-Evidence-Wasnt): Benchmark evidence from 15 models, 180 runs and 570 required claims - [Provably V2 is Live](https://provably.ai/blogs/V2-Live): Relational database support and verifiable SQL IDE - [Beyond Circuits](https://provably.ai/blogs/Beyond_Circuits_New_Era_for_Verifiable_DBs): Why QEDB makes verifiable SQL practical - [Rethinking Verifiable Databases](https://provably.ai/blogs/Rethinking_Verifiable_Databases): QEDB as a third path beyond ADS and SNARK based systems ## Company Provably correct, not just plausibly right. We build verifiable data and runtime verification infrastructure so agents and humans can collaborate with accuracy, safety and integrity. Our research in verifiable databases creates cryptographic anchors of integrity across the agent lifecycle, from tool use and handoffs to cross agent protocols, learning and governance. For a new era where trust and integrity are verified at machine speed, not left to contracts and paper trails. - [About Us](https://provably.ai/about-us): Team, mission and company background - [Changelog](https://provably.ai/changelog): Product updates - [GitHub organisation](https://github.com/ProvablyAI): Provably open source repositories - [X](https://x.com/getprovably): Company updates - [LinkedIn](https://www.linkedin.com/company/provably): Company profile For current APIs, package metadata and setup instructions, prefer the live documentation, PyPI package and GitHub repository over cached descriptions.