The core ShipFoundry artifact that turns a relevant update into scoped agent-ready work.

An Investigation Brief explains what changed, why it matched a project, what evidence exists, what is still missing, and which output lane is safe next.

Core Artifact

If a user reads only one artifact page, make it this one. The brief is the trust boundary between upstream change and project-specific action.

When To Use This

Use an Investigation Brief when an update looks relevant but still needs scoped product or repo review before action.

Use it before asking an agent to implement anything. The brief should make the investigation bounded: what changed, where to look, what not to touch, and how to verify the outcome.

How It Works

ShipFoundry combines the public update, project context, tracked stack, source evidence, repo evidence if available, and prior decisions into a structured brief.

The brief is not just summary text. It is an operational artifact that should help you decide:

whether this is worth acting on
whether local repo proof exists
whether more inspection is needed
whether the work is safe to hand to an agent
what acceptance checks should prove success

Output Lanes

The lane says what the next human or agent should do. It is not a cosmetic label.

Lane	Use When	Required Shape
Hand To Agent	Existing implementation is proven and bounded enough for direct work.	Executable work order with concrete files, constraints, and verification.
Engineering Spike	The opportunity is real but implementation or adoption proof is incomplete.	Bounded investigation with concrete starting surfaces and an expected decision.
Workflow Brief	The unlock changes an operating process.	Decision memo with workflow options, evidence, and a trigger.
Strategy Brief	The unlock changes product or commercial direction.	Decision memo with options, tradeoffs, and a recommendation.
Research Brief	The key unknown needs external or technical investigation.	Research plan with evidence, unknowns, and a stop condition.
No Packet	No sufficiently useful action artifact exists yet.	The judgment remains visible without generating a weak handoff.

Net-new product work stays an Engineering Spike until the repository and product path establish enough proof for implementation.

Brief Template

Every strong brief should answer:

What changed
Why it matched
Evidence available
Evidence missing
What to inspect
What not to touch
Risks
Suggested path
Acceptance checks
Agent prompt / handoff
Outcome options

Step-By-Step

Open a relevant update.
Check the trust and readiness state.
Read what changed and why it matched.
Review source evidence and repo evidence separately.
Confirm what evidence is missing.
Confirm that the selected output lane matches the evidence and unresolved questions.
Hand the brief to an agent only if the suggested path and acceptance checks are specific enough.
Record the outcome after review.

What ShipFoundry Does

Separates evidence from recommendation.
Names missing proof instead of hiding uncertainty.
Gives the agent a bounded path and non-goals.
Provides acceptance checks so the work has a finish line.
Keeps the brief tied to a specific product, update match, and outcome history.

What ShipFoundry Does Not Do

It does not guarantee the update applies to your repo without repo evidence.
It does not replace local inspection by a user-owned coding agent.
It does not authorize broad refactors.
It does not automatically export every brief to Linear.
It does not treat every relevant update as implementation-ready.

Investigation Vs Implementation

An Engineering Spike asks an agent to resolve a bounded uncertainty. Hand To Agent authorizes a bounded implementation only after the relevant code path is proven. Research, workflow, and strategy briefs route the opportunity away from premature coding.

If ShipFoundry shows No Packet, improve the project context or evidence instead of forcing a generic prompt.

Troubleshooting

If a brief feels too broad, do not export it. Improve product context or wait for stronger repo evidence.

If evidence is missing, ask the agent to inspect only the named uncertainty.

If the agent finds the update does not apply, record a false-positive or not-relevant outcome.

Example artifact: Sample Engineering Spike
Safety states: Trust And Readiness
Per-match repo proof: GitHub Repo Evidence
End-to-end agent path: Agent Workflows
Agent workflow: MCP
Product CTA: Start 7-Day Trial

Was This Page Useful?

Send a short note if this page is unclear, overclaims a capability, or is missing the proof you expected.

Send Docs Feedback

Investigation Briefs