Your workflow.
A better starting point.

Bring code and saved knowledge to people and AI agents through the configured CLI or MCP. A context layer for companies, teams, and people, inside the workflow you already use.

Access is configured for your evaluation. Confirm source coverage, identity, and host compatibility before running a request.

Configured access surfaces. Source coverage and host compatibility are confirmed for your evaluation.

Ask for context.
Keep the source in view.

This example calls context.search in an already configured evaluation. Replace the organization and actor placeholders with your assigned identities.

The request sets a result limit and context budget, and asks for an explanation. Inspect the available references before using the result in your next task.

CLI / context.searchA query in a configured evaluation
shepaird call context.search --input '{
  "query": "Where is refund eligibility checked?",
  "org_id": "YOUR_ORG_ID",
  "actor_id": "YOUR_ACTOR_ID",
  "budget_tokens": 2000,
  "limit": 10,
  "explain": true
}'

Example request for a configured environment. Your evaluation contact confirms the supported setup.

Choose the access surface that fits your work.

CLI

Make an explicit capability call from a configured environment. A useful starting point for inspecting a request and its returned context.

MCP

Give a compatible agent host access to the configured context tools. Confirm the profile, identity, and source boundaries supported in your trial.

Provider-free retrieval is a starting point.

Keyword and fuzzy retrieval work without a model provider. Semantic retrieval depends on configured embedding coverage. Evaluate a ranked, budgeted context pack against your existing workflow. Measure answer quality, latency, and model usage in your host.

Explore the retrieval path

Continue from
what was learned.

Recall reusable findings with their origin and freshness, then inspect the source when you need to verify the evidence. Session fabric connects shared findings, handoffs, and work in progress within an explicit audience. A handoff gives the next session a place to begin and preserves the checks that are still open.

Evaluate source freshness, recall, and session continuity in your actual workflow. Keep execution approvals and tool permissions explicit in your workflow.

Review trust and access
Illustrative session continuity, with an explicit audience and the open check.

Evaluate one
complete workflow.

Bring a real question, the sources that answer it, and the host where you want to use the context.

Plan your trial