Context search for people and AI agents

Your company’s knowledge. Your team’s knowledge. Your knowledge. One search away.

A context search engine for companies, teams, and people. Find code and saved knowledge. Give agents the same context.

The context your work needs.

Different work. Different questions. See the sources, exceptions, and next steps that help people and agents move forward.

For engineering teams and their agents

Understand the code before you change it.

See the code, its original rationale, and the tests still to write.

Why do older orders skip refund fees?

Evaluate your own question

What the sources tell you

Orders created before the billing migration keep their original no-fee refund terms.

Example saved finding: billing migration

The refund entry point
if (order.legacy) return refundLegacyOrder(order);
src/billing/refund.ts
The original migration decision
Pre-migration orders retain the original no-fee refund policy.
notes/billing-decisions.md
The boundary tests to write next
Both paths are mapped. Boundary and repeated-request tests are next.
handoffs/refund-review.md

Condition to check: Only orders created before the migration
Suggested next step: Test both refund paths and repeated requests.

Illustrative saved knowledge, not live results or available connectors. Your trial confirms the sources and access rules for your deployment.

From a question to the source.

Choose a question, then inspect the code, saved decision, or handoff behind a result. Illustrative data, not a live search.

Where does our refund logic live?

Explore the context 3 example sources

Source previewL24–30

The refund entry point

src/billing/refund.ts

export function refund(order: Order) {
if (order.legacy) {
return refundLegacyOrder(order);
}
const fee = calculateFee(order);
return issueRefund(order, fee);
}

Source references keep the evidence close.
Inspect the original context before you act.

Illustrative data · Choose a question, then select a result to inspect its source. Not a live search.

Start with what you know.

Start with code and saved knowledge. Confirm additional sources for your deployment.

Codebases

Registered repositories, symbols, and source references.

Trial starting point

Saved knowledge

Findings, decisions, and session handoffs with their source context.

Trial starting point

Documents & knowledge bases

Scope the systems, content, and access rules you need.

Integration to confirm

Slack & email

Message ingestion is implemented; connection setup and access need confirmation.

Setup to confirm

Business systems

Agree on the APIs, data boundaries, and workflows to evaluate.

Integration to confirm

Read the source guide

A source listed for discussion is not an available connector. Supported sources and permission behavior are confirmed for your deployment.

Set the boundaries before connecting data.

Hosting, access, and data handling are agreed for your evaluation.

Review trust and access

Sources: What to index
Audience: Who can retrieve it
Deployment: Where it runs

Your agents.
Your context.

Use the configured CLI or MCP to bring retrieval into the workflow you already have.

Search: Retrieve code and saved knowledge with source references.
Focus: Use a ranked, budgeted context pack for the task.
Remember: Keep findings, decisions, and handoffs with their origin. Recall useful knowledge with its freshness in view.
Coordinate: See shared findings, handoffs, and work in progress within the agreed audience.

Explore developer access

How the platform works

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 after your evaluation environment is configured. Replace the organization and actor placeholders with your assigned identities.

Your company’s knowledge. Your team’s knowledge. Your knowledge. One search away.Your company’sknowledge.One search away.