Assessments and reports

Running your first assessment

What happened

You have access and an agent to document (via the connector or added manually), and you want to produce a result. AgentProof applies only the question packs relevant to what your agent can do, then produces a readiness score and band, an evidence, controls, and gaps summary, and a compliance-readiness record.

Why it matters

The compliance-readiness record is the core deliverable: a decision-support aid for getting an agent ready before your compliance review. Older compliance-readiness records stay immutable, so your progress is a real, dated record rather than a retroactive edit. AgentProof is not a certification, regulatory approval, audit, or insurance.

What to do next

1. Describe one agent: what it does, who it talks to, and what it can act on.
2. Answer the matching question packs and provide supporting evidence.
3. Review the generated compliance-readiness record; work through the listed evidence gaps and blockers, then reassess.
4. To see the end result before committing, explore the public 'how an agent gets documented' walkthrough; no sign-in is required.

What information to include

1. Your workspace name and which agent you assessed.
2. The step where the assessment stopped, if it did.
3. A screenshot of any on-screen message.
4. What looked incomplete or inconsistent compared with your inputs.

When to contact support

If an assessment will not generate, or a compliance-readiness record looks incomplete or inconsistent with your inputs, contact [email protected] with the workspace name and the agent. If you need help interpreting a result for stakeholders, we are happy to assist.

Reading your compliance-readiness record

What happened

You finished answering the question packs for an agent and AgentProof generated a compliance-readiness record. It contains a 0 to 100 readiness score, a readiness band (Ready with controls, Needs review, Not ready, or Insufficient evidence), a risk view, your top control priorities, the evidence basis behind each finding, an evidence-gap and blocker summary, and a single 'next best action'.

Why it matters

The band and risk posture are the headline, not the raw number. The score is the one figure you can paste into a status update, but the band tells stakeholders whether the agent is genuinely ready, needs work, or simply has not been evidenced yet. 'Insufficient evidence' means AgentProof could not confirm enough to judge readiness; it is not the same as a low score.

What to do next

1. Read the band and risk view first, then the 'next best action' card; it identifies the single improvement that would most lift the score.
2. Work through the Top 5 priorities; these are the focused fixes that close your largest evidence gaps and blockers.
3. Apply the controls, then reassess the agent. Your previously confirmed answers are shown again and remain editable, so you update only what changed; the new compliance-readiness record is saved alongside the prior record, which stays unchanged in history.
4. Print or save to PDF directly from your browser when you need to share it; AgentProof formats its own compliance-readiness record pages.

What information to include

1. Your workspace name and which agent the record is for.
2. The readiness band and score you expected versus what you saw.
3. The approximate date the compliance-readiness record was generated.
4. A screenshot of any figure that does not match your answers.

When to contact support

If the compliance-readiness record fails to generate, shows figures that do not match the answers you submitted, or a saved record is missing from history, email [email protected] with the workspace name and the agent. Do not rely on a re-score to recover a missing record; contact us so the original can be checked.

Why an older record did not change after I re-scored

What happened

You re-scored an agent and noticed the earlier record is unchanged, with a new compliance-readiness record sitting alongside it in history.

Why it matters

Compliance-readiness records are immutable: once generated, a compliance-readiness record is never edited. A re-score creates a new record and leaves the old one in your history unchanged, so each record is a permanent, dated snapshot and your progress is real rather than retroactive.

What to do next

1. Check your record history for the assessment date you expect; new assessments produce new compliance-readiness records.
2. Make sure you are signed in to the correct workspace; you may need to sign in to see your own records.
3. Compare the new record against the prior one to see how applied controls moved the band and score.

What information to include

1. Your workspace name and which agent the record was for.
2. The approximate date the compliance-readiness record was produced.
3. Whether you have re-scored that agent since.
4. Whether you are signed in to the same workspace where it was created.

When to contact support

If a record you know was generated is genuinely missing from your workspace history, email [email protected] with the workspace and the approximate date it was produced.

I cannot complete an assessment or the compliance-readiness record did not generate

What happened

An assessment did not finish, or the compliance-readiness record failed to generate.

Why it matters

Your answers are preserved when a compliance-readiness record could not be generated, so you will not lose your work; but you do need a successful generation to get the readiness score and the evidence, controls, and gaps summary.

What to do next

1. Make sure you are signed in to the correct workspace.
2. Refresh the page and try generating the compliance-readiness record again; your saved answers are retained.
3. If a page shows a temporarily-unavailable message, your work continues locally and is not lost; retry in a few minutes.

What information to include

1. Your workspace name and which agent you were assessing.
2. The exact step where the assessment or compliance-readiness record generation stops.
3. A screenshot of any on-screen message, never any token or secret.
4. Whether your saved answers are still present after a refresh.

When to contact support

If generation fails repeatedly after a retry, email [email protected] describing the step where it stops and including a screenshot of any on-screen message. Never include tokens or secrets. Your data is stored in the EU.

When and why to reassess an agent

What happened

You already have a compliance-readiness record for an agent and you are deciding whether it is worth running the assessment again. Reassessment re-scores the same agent and produces a fresh compliance-readiness record, while the earlier record stays in your history unchanged.

Why it matters

A readiness score is a point-in-time view, not a permanent verdict. After you act on the findings, or after the agent or the wider picture changes, the old record can understate or overstate how ready the agent really is now. Reassessing keeps your decision based on the current state, and because records are immutable you also get a dated snapshot of the improvement over time.

What to do next

1. Reassess after you have applied controls or closed the evidence gaps and blockers the last record listed, so the new score reflects the work you did.
2. Reassess when the agent itself changes: new capabilities, new data access, a new system it can act on, or a new audience it talks to.
3. Reassess when Radar intelligence updates the picture, since new guidance can change which controls matter for your agent.
4. Compare the new record against the prior one in history to see how the band and score moved; the earlier record is never overwritten.

What information to include

1. Your workspace name and which agent you reassessed.
2. What changed since the last record (controls applied, capability change, or a Radar update).
3. The readiness band and score before and after, if you have them.
4. The approximate dates of the two records you are comparing.

When to contact support

If a reassessment will not generate, or the new compliance-readiness record does not reflect changes you know you made, email [email protected] with your workspace name, the agent, and what changed since the previous record. Never include tokens or secrets.