Agents
Register your agent. Power the network.
Every agent that joins makes the platform more valuable for everyone. Contribute anonymized execution data and get analytics, benchmarks and evaluations built on a corpus no single team could assemble.
How it works.
Register your agent
Create an agent profile in minutes. No minimum scale, works with any framework.
Connect the SDK
A few lines of code capture traces, tool calls and outcomes automatically.
Share anonymized traces
Choose exactly what to share with granular, per-agent opt-in controls.
Unlock analytics & benchmarks
Get execution analytics, benchmark reports and evaluations in return.
Why contribute.
Clear incentives, designed so participation pays for itself.
Execution analytics
Understand your agents' latency, cost and success trends against the broader network.
Benchmark reports
See how your agents compare on production-derived benchmarks.
Evaluations on your stack
Run StratScope evaluations against your own tasks and tooling.
Priority dataset access
Early access to new datasets and private corpus releases.
Contribution incentives
API credits and dataset allowances that scale with the data you share.
Research previews
First look at findings from millions of anonymized executions.
Privacy and control by design.
Explicit opt-in, always
Nothing is shared without explicit consent. Participation is per-agent, per-dataset, and reversible at any time.
- Granular per-agent sharing controls
- Per-dataset opt-in choices
- Audit log of everything shared
- No retroactive use of withdrawn data
Anonymized before it leaves
Traces are normalized and anonymized before entering the corpus. Payloads, identifiers and sensitive content are stripped.
- Payload and PII redaction
- Identifier replacement and tokenization
- Raw conversations never stored
- Opt-out with a single switch
Two lines to start capturing.
from stratscope import StratScope
client = StratScope(api_key="sk_live_...")
@client.trace(agent="support-bot-v3")
def handle_ticket(ticket_id):
return escalate(ticket_id) # every call, retry and outcome is captured
Frequently asked questions.
Only the execution metadata you opt in to: tool calls, outcomes, latency and cost. Raw payloads and conversations are never stored.
Yes. Contributing agents get a free tier with analytics, benchmarks and evals. Premium access scales with contributions.
Yes. Opt out anytime; withdrawn data is excluded from future dataset releases and not retroactively extended.
Identifiers and payloads are stripped or replaced during normalization, before anything enters the corpus.
Only aggregated, anonymized datasets reach licensed customers. Contributor identities are never disclosed.
We support any agent built on Python, TypeScript or REST — from LangChain-style frameworks to bespoke systems.
Analytics, benchmark reports, evaluations on your stack, priority dataset access and research previews.
Bring your agent into the network.
Register your agent, contribute anonymized execution data, and unlock analytics, benchmarks and evaluations that no vendor can give you alone.