> ## Documentation Index > Fetch the complete documentation index at: https://docs.corepillar.ai/llms.txt > Use this file to discover all available pages before exploring further. # Reports > Agent Audit Reports for a simulation run, an evaluation run or a single trace: what each report contains, how it is built, suggested fixes, and Markdown and PDF export. **Reports** in the sidebar opens **Agent Audit Reports**: "Analyse a simulation run, an evaluation run or a trace: what the agent did, how it graded, and the specific issues behind the score." A report is not a document you write or a file AgentHub stores. It is a view **derived from data your project already has**, recomputed every time you open it. There are three kinds of subject: | Source | Built from | Created by | | - | - | - | | **Simulation** | A Cost Calculator run: its scenarios, their executions and grader scores. | Running an agent in the [Cost Calculator](/agenthub/cost-calculator). | | **Evaluation** | A saved trace evaluation run and its per-evaluator results. | A [trace evaluation run](/agenthub/evaluation/runs#trace-evaluation-runs). | | **Trace** | One trace and its spans. | Any [trace](/agenthub/traces) in the project. | Every simulation run, saved evaluation run and trace in the project automatically has a report. The only thing a report ever saves is a simulation run's **Suggested Fixes**, which are stored on that run once generated. ## The reports list Under the heading, a line summarizes what's available, for example "Derived from this project's own data and recomputed each time a report is opened — 4 simulation runs · 2 evaluation runs · 120 traces." | Column | Contents | | - | - | | **Name** | The report name as a link, with a stable report ID under it such as `rpt_trace_8f2c1a…` (hover for the underlying subject ID). | | **Source** | **Simulation**, **Evaluation** or **Trace**. | | **Status** | **Completed**, **Pending** or **Failed** with a colored dot (hover for the subject's own status). Simulation reports that already have fixes also show **Fixes generated**. | | **Created** | When the subject was created or started. | Click a row or its name to open the report. **Filters** | Control | Options | | - | - | | **Search reports...** | Matches the name, report ID, subject ID and the secondary detail (model, judge, span count). Search looks through the 300 most recent subjects. | | Status | **All status**, **Completed**, **Pending**, **Failed**. | | Source | **All sources**, **Simulations**, **Evaluations**, **Traces**. | The list loads 20 at a time; **Load more** appends the next page. How status is bucketed: | Bucket | Simulation run | Evaluation run | Trace | | - | - | - | - | | **Completed** | `complete` | `completed` | `OK` | | **Pending** | `draft`, `generated`, `running` | `running` | (never) | | **Failed** | `failed` | `failed` | `ERROR` | Empty states: **Nothing to report on yet** (with **Go to Cost Calculator**) when the project has no subjects, and **Nothing matches these filters** (with **Clear filters**) when filters hide everything. ### Generate Report **+ Generate Report** opens a dialog: "Pick what to analyse — a simulation run, an evaluation run, or a single trace — and the report is built from that subject's own data." 1. **Report on**: Simulation, Evaluation or Trace, with counts. Kinds with none in the project are disabled. 2. **… to analyse**: pick the subject. Drafts that never ran aren't listed. 3. Click **+ Generate Report**. The report opens. The dialog tells you what generating costs: | Kind | Cost | | - | - | | Simulation | Scores and issues come from the run's own simulations. **Suggested fixes are written by a real model call using this project's configured key**, so generating this report spends against it. From this dialog, fixes are generated automatically when the report opens. | | Evaluation | "No model call, nothing is billed." | | Trace | "No model call, nothing is billed." | You can also reach reports directly: **Report** in the Traces list, **Run report** on a trace page. ## What a report contains Every report has the same layout: a header, a headline rate, a facts panel, a bar chart of results, **Top Issues**, and an evidence section. Sections that don't apply to a subject say so and explain why, often with a link to the page that would fill them. ### Header The kind ("Report · Simulation run", "Report · Evaluation run", "Report · Trace"), the report name, a status, the subject ID, the date, extra tags such as environment and source, and the **PDF** and **Markdown** export buttons. ### Simulation run reports | Section | Contents | | - | - | | **Scenario Pass Rate** | Executed scenarios that passed, out of those executed, with the number generated. A scenario passes when its execution completed without error; there are no per-scenario assertions beyond that. Empty until at least one scenario has executed. | | **Run facts** | Measured cost, estimated cost, tokens in / out, scenarios generated, scenarios executed, batch. | | **Grader Results by Category** | Offline grader scores, and pass rate by scenario category. | | **Top Issues** | Recurring failures grouped by scenario name and category, up to 8. | | **Suggested Fixes** | Targeted fixes to the agent's prompt and configuration, written by the run's own model. See below. | ### Evaluation run reports | Section | Contents | | - | - | | **Evaluator Pass Rate** | Evaluators that passed, out of those that ran, and how many the collection had. A rate over evaluators, not scenarios. | | **Run facts** | Collection, judge model, evaluators run (N of M), average score, the trace, and the error if the run failed. | | **Grader Results by Evaluator** | Each evaluator's score, and pass rate by evaluator category. | | **Top Issues** | Evaluators the judge marked as failed, worst score first. | | **Judge Reasoning** | "The judge's own written justification for each score, verbatim. These are explanations, not fixes." Each card has a **Copy** button. | ### Trace reports | Section | Contents | | - | - | | **Span Success Rate** | Share of the trace's spans with status `OK`, alongside the trace's own `OK` / `ERROR` outcome. A trace has no pass rate of its own. | | **Trace facts** | Outcome, duration, LLM time, HTTP time, spans, LLM calls, tool calls, tokens in / out, cost, environment, source. | | **Span Results by Kind** | Span outcomes per kind. If the trace was imported into the Cost Calculator, its offline grader scores too. If saved evaluation runs exist for the trace, they are listed as **Attached evaluation runs**. | | **Top Issues** | Spans with a non-OK status, worst first. | | **Failure Evidence** | The failing spans' recorded input and output, verbatim. If the trace was imported into the Cost Calculator and audited, the title becomes **Suggested Fixes & Failure Evidence** and includes those fixes. | Trace reports state plainly what they leave out: there is no scenario pass rate, no grader section unless an evaluation was run against the trace, and no suggested fixes for a trace on its own. ## Suggested fixes (simulation runs) On a simulation run report, **Generate suggested fixes** asks the run's configured model to diagnose the failing scenarios and propose targeted fixes. If nothing failed, the button reads **Suggest prompt improvements** and the model reads the passing transcripts and grader averages to draft improvements instead. * It is one real call, billed to the run's model on your key. It can take a minute; keep the page open. * The result is saved to the run and shown with "Generated … by `` and saved to this run." The list then shows **Fixes generated**. * Each fix card has a **Copy** button. It can't run when: | Message | Fix | | - | - | | "No scenario in this run has executed yet, so there is nothing for the audit to read." | **Run this agent's scenarios** in the Cost Calculator. | | "This run has no model attached, so the audit has nothing to ask." | Attach a model in the Cost Calculator. | | "This run's model has no API key, so the audit cannot call it." | **Open Model Manager** and add a key. | | "This run's model API key is missing or revoked." | Replace the key in Model Manager. | | "This run's model API key could not be decrypted." | Re-enter the key in Model Manager. | | "You need the run-agents permission to generate an audit." | Ask an admin for a role with run permission. | ## Export and sharing | Button | What you get | | - | - | | **Markdown** | Downloads the report as a `.md` file named after the report. | | **PDF** | Opens a print-ready version of the report in a new window and starts your browser's print dialog. Choose **Save as PDF** there. If nothing happens, allow pop-ups for the site ("Allow pop-ups for this site to export a PDF."). | Both exports include what is on screen, including generated fixes. There is no public share link. To share a report inside your organization, send its URL: anyone with access to the same project in AgentHub can open it, and it is recomputed from current data when they do. To share outside, export it. Because reports are recomputed on open, a report can change after you export it, for example when more scenarios run or new failures appear elsewhere in the project. The export is a snapshot. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.