CiterLabs.

HUMANS & AGENT BUILDERS / AN OPEN INVITATION

Less repetitive work.
More human judgment.

Help improve practical automations by testing useful workflows, sharing reproducible examples and asking where they break. We are starting with enquiry intake, clearer briefs and response drafts that people review.

A starting point for participation. There is no live forum, public agent directory or membership program yet.

BUSINESS OWNERS

Where does your
enquiry process stall?

Describe a non-confidential bottleneck and the decision that needs to happen next. Use the kit to test whether a clearer record or draft helps.

Plan a scenario ↗

REVIEWERS

Would you trust
this response draft?

Bring a synthetic example. Identify unsupported claims, missing scope and the checks a person needs before approving a message.

Use the templates ↗

AGENT BUILDERS

Can your agent
use this workflow?

Review the live A2A service, its exact schema and limitations. Share a reproducible synthetic test rather than a compatibility claim.

Review the A2A interface ↗

Keep data safe and permitted

Use synthetic examples or public material you have permission to share. Never include customer messages, private records, secrets, credentials or unrestricted drive links. Name any data destinations, retention behavior and tool costs in your example.

Be clear about identity and evidence

Identify when an AI system is involved and who operates it. Label assumptions, estimates and tested results separately. Include the test date and versions. Do not fabricate engagement, testimonials or success metrics; disclose commercial ties.

Respect people and permissions

Keep human approval points explicit. Do not use agents for bulk promotional replies or unsolicited outreach. Share reproducible learning, respect tool and data permissions, and explain where a workflow failed.

Agent listings need verification

No agent directory entries are published here. A future listing should identify its operator and useful scope, exact A2A version, transport, authentication and Agent Card, plus costs, privacy boundaries and limits. Compatibility stays untested until an authorized synthetic client exercise is reproducible. A static card alone is not evidence of an executable service.

A REPRODUCIBLE CONTRIBUTION

Share the test.
Include the edges.

Email a concise suggestion or synthetic workflow example. We have not set up a discussion forum yet; the current participation route is the existing CiterLabs email.

Share a contribution ↗

Sending opens your email app. No submission is stored on this website, and no response time is promised.

What to include

  • Problem, intended user and approved tools.
  • Setup steps and a synthetic input.
  • Expected output and the actual observed result.
  • Test date, software and protocol versions.
  • Failures and human approval points.
  • Data destinations, retention, costs and reuse permissions.
  • Your operator identity and any commercial ties.
Download the contribution template ↓

Original template contents: CC BY 4.0 — attribution and scope.