Reports and evidence
This guide shows you how to map rules to compliance frameworks, collect evidence of automated check results, download reports for different readers, and put together what an auditor needs.
Note: LEA Platform reports show which automated technical checks passed or failed on which hosts. They support control-owner review and audit preparation. They are not an audit opinion, a certification, an attestation or a decision that you comply with a framework.
Before you start
- Anyone can open Reports. Changing framework settings, rotating the sealing key and taking snapshots need an administrator.
- Words used here:
- A framework is a compliance standard, such as CIS or ISO 27001. A requirement (also called a control) is one item in it, such as ISO 27001 A.8.2.
- A check is a rule: an automated test of one technical setting. A check can be mapped to one or more requirements, or only to the framework as a whole.
- A snapshot is a recorded copy of each in-scope host's latest result for every check of one framework. The results were observed at different times; the snapshot is taken afterwards.
- Evidence is one record in a snapshot: one check on one host for one mapped requirement, with its result.
- Drift is the number of host and control pairs that passed in the previous snapshot and fail in the latest one.
- See also the glossary.
Map rules to controls with std: tags
LEA Platform knows which rules belong to which framework through the rule's tags:
| Tag | Meaning | Example |
|---|---|---|
std:<FRAMEWORK> | The rule is a check associated with the framework, but not mapped to a specific requirement. Reports list it as an unmapped check. | std:CIS |
std:<FRAMEWORK>:<control> | The rule is mapped to this requirement. | std:ISO27001:A.8.2, std:SOC2:CC6.3, std:CMMC:AC.L2-3.1.5 |
evidence:sample | The rule reads sample or lab data, not live configuration. Reports label its results Sample data. | evidence:sample |
evidence:simulated | The rule's output is synthetic; no live control is tested. Reports label its results Simulated. | evidence:simulated |
A rule can carry tags for many frameworks. Library rules are already tagged, for example for CIS, CIS Controls v8, ISO 27001, SOC 2, CMMC, NIST SP 800-53, NIST CSF 2.0, PCI DSS 4.0, HIPAA, NIS2, DORA and the ASD Essential Eight.
Any framework works, including one that did not exist when LEA Platform was installed. The <FRAMEWORK> part of the tag is the framework's ID: tag your rules std:MYSTD:<control> (or add new rules with that tag) and MYSTD appears under Settings → Compliance frameworks, on the Dashboard and in the rule library. Enable it there to get sealed snapshots and reports. IDs are case-sensitive and use letters, digits, _, . and - (up to 32 characters).
Note:
std:CIS_Controls_v8:...is a different framework fromstd:CIS. Library rules with a CIS Controls v8 safeguard carry both tags.
The framework scores on the Dashboard are worked out live from the current results of the mapped rules, weighted by severity, and per-control views group the rules by the control part of the tag. A snapshot is a fixed copy: its pass rate counts passing and failing evidence records without weights.
To add tags to a rule:
- One rule: edit the rule → General → Tags, type the tag and press Enter, then Save rule.
- Many rules: on the Rules page, select the rules, click Bulk edit, tick Tags, choose Mode → Add, enter the tags and click Apply.
See Write and edit rules.
Set up a framework
-
Open Settings → Compliance frameworks.
-
Check the Evidence Sealing Key (HMAC) card. Key Fingerprint identifies the key that LEA Platform uses to compute the record MAC of every evidence record (the first 16 characters of the key's SHA-256; it is not a signature). If it says No key configured, click Rotate Key and then Generate & Rotate.
-
Find the framework. Frameworks are listed in three groups:
Group What is in it Enabled Frameworks that get snapshots and a card on Reports. Tagged in rules Active rules carry the framework's tags, but it is not enabled yet. New frameworks start here and stay off until you enable them. Known standards LEA Platform has report wording for them, but no active rule is tagged yet. Collapsed by default. Each card shows the framework's name, its ID, how many active rules and controls map to it, and Needs wording when reports cannot be made yet (see Describe a standard).
-
Set:
Field What it does Enabled / Disabled switch Only enabled frameworks get snapshots and appear with a card on Reports. Snapshot Interval (hours) How often a snapshot is taken automatically. Default 24. Report Interval (hours) Stored with the framework (default 168). Report Format JSON, PDF or Both. Stored with the framework. Endpoint Group Scope The groups whose hosts are included. All endpoints when none is chosen. -
Click Save in that framework's card.
You should see the framework on the Reports page, with Next Snapshot filled in.
Note: Only active hosts and active rules are included in a snapshot. Hosts are included through their groups, also through nested and dynamic groups.
Describe a standard
PDF reports print the standard's full name, a report title, a sentence on which checks the report covers and, in the GRC (Legal) report, its legal and commercial context. LEA Platform ships this wording for many standards, updated with rule packs. You can change it for any framework, and you write it once for a standard LEA Platform does not know yet.
To change the wording:
-
Open Settings → Compliance frameworks and click Report wording on the framework's card.
-
Change the fields you need:
Field Where it appears Full name (required) Report covers and the framework line of every report. Short name Cards, chips, charts and filters in the console. Published by, Version, Region, Reference URL Reference information for readers. Report title (required) Title of the evidence PDF and the stakeholder reports. Which checks the report covers (required) Cover subtitle; one sentence, for example "Selected automated technical checks associated with DORA ICT risk-management requirements." Legal and commercial context Callout of the GRC (Legal) report. Optional; the callout is left out when empty. Requirement is called What the standard calls a requirement, such as Requirement or Article. Header of the mapped requirements table. Evidence report columns Up to four extra columns of the evidence PDF, in order, with their headers: observed value, snapshot and record reference, record MAC, agent version. The check, requirement, host, result, observation time and run columns are always printed and cannot be renamed. -
Click Save wording.
A changed field shows Customized with Reset to default. Your wording is kept when a rule pack brings new default wording.
Results always print as Check passed or Check failed: a passing check does not show that a requirement is implemented or effective, so reports have no wording option for it. When your text claims verification, assurance, certification, implementation or compliance, the dialog shows a warning. Your text is still printed as written, but every report also prints its own summary, scope limits and the statement that the results are not audit assurance, and your text cannot replace them.
When you download a PDF, a stakeholder report or the ZIP for a framework without wording, the Report wording dialog opens with the missing fields marked. Fill them in and click Save and generate: the report downloads and later reports use the same wording. JSON downloads and snapshots never need wording. With the AI Assistant set up, Draft empty fields with AI proposes text for the fields that are still empty; check it before you save.
Take a snapshot and read the evidence
Snapshots are taken on the schedule above. To take one now:
- Open Reports, tab Frameworks & evidence.
- In the framework's card, click Snapshot (Trigger snapshot now).
For each host in scope and each check, the snapshot records the result (pass or fail), the observed value (the output the host reported), the agent version, when the host reported the result and when the snapshot was taken. The result and the observed value always come from the same run: the host's latest check, or its latest remediation when that ran after the check (after a remediation the agent runs the check again, and the result is that check's output). Each record also says which kind of run it came from, the ID of that result, the rule and rule version, the requirement it is mapped to (if any), whether the rule comes from a rule pack or was written or changed locally, and whether it is live, sample or simulated evidence.
LEA Platform computes a record MAC for each record (HMAC-SHA256 with the evidence sealing key) when it takes the snapshot. A matching MAC shows the record has not changed since then, unless someone used the key. It is made by the LEA server, not by the agent, so it does not prove when or where a result was observed, and only someone holding the key can check it. To check every record MAC of a snapshot, an administrator calls POST /lea/compliance/snapshots/<id>/verify (see the API reference); JSON downloads and the GRC (External Auditor) report include the result of this check.
Each framework card shows:
| Part | What it shows |
|---|---|
| Overall Posture | Passing and failing controls in the last snapshot. |
| Last Snapshot | When it was taken, how many hosts, and its ID. "Never" when none exists yet. |
| Next Snapshot | When the next scheduled snapshot runs. |
| Drift (PASS → FAIL) | "N controls regressed" since the previous snapshot, or "No drift detected". |
| Endpoint Group Breakdown | Pass, Fail and Pass Rate per group. |
To look at single records:
- Click Inspect (Drill down into evidence records).
- Choose By Control or By Host.
- Click a row to see the records behind it: Host (or Control ID), Result, Observed at, Observed Value, Agent Ver. and Record MAC.
Note: Rotating the sealing key does not invalidate existing records; new records are sealed with the new key. Rotate it only when your security policy asks for it, and note when you did.
Read the Inspect table
| Column | What it shows |
|---|---|
| Control ID / Rule | The framework requirement, built from the rule's std: tag (for example ISO27001-A.8.2), with the name of the rule that checks it. A rule with only the framework tag gets an internal ID like CIS-cmd-42: it is a check, not a requirement of the framework. A rule tagged with several requirements of the framework gives a record for each of them. |
| Pass | Number of passing records in the row: hosts that pass the control (By Control) or controls the host passes (By Host). |
| Fail | Number of failing records in the row: hosts that fail the control (By Control) or controls the host fails (By Host). |
| Pass Rate | Passing records out of all records in the row, without severity weights. INFO results are never part of a snapshot. |
Read the evidence records
| Column | What it shows |
|---|---|
| Control ID | The framework requirement the record is for, built from the rule's std: tag, with the rule that checks it below. <FRAMEWORK>-cmd-<rule> means the rule is not mapped to a requirement. |
| Result | Whether the host's latest result for this check, when the snapshot was taken, passed or failed. Only PASS and FAIL are recorded. |
| Observed at | When the host reported that result (UTC), not when the snapshot was taken. Below it: the kind of run (check run, remedy then validation, function remedy, action), the result ID and sample or simulated when the rule is marked so. "unknown (legacy record)" for records taken before observation times were stored. |
| Observed Value | The text the host printed (its stdout) in the run that decided the result: the rule's check, or its remediation when that ran last. Recorded and signed in the snapshot. At most 512 characters are kept. |
| Agent Ver. | Version of the LEA agent installed on the host when the snapshot was taken, or "unknown" when the agent did not report it. |
| Record MAC | First 12 characters of the record's MAC (HMAC-SHA256 with the evidence sealing key, computed when the snapshot was taken). The full MAC is in the JSON download. |
The JSON download (schema_version 2) separates three times: report_generated_at (when the file was made; the only value that changes when you download it again), snapshot_collected_at (when the snapshot was taken) and, per record, observed_at (when the host reported the result, full precision, UTC). Each record also has:
| Field | What it shows |
|---|---|
purpose | The kind of run the result and value come from: check, remedy (remediation followed by the check; the result is that check's output), function_remedy (a remediation released by a function) or action (a rule action chosen by a function). |
result_id | ID of that result in the execution history. |
requirement_id | The requirement the rule is mapped to, or null when the rule has only the framework tag. |
cmd_id, template_id | The rule and the exact rule version. |
rule_origin | pack (unchanged rule-pack rule), modified (changed locally), custom (written locally) or unknown. |
evidence_kind | live, sample or simulated, from the rule's evidence: tag. |
record_mac, payload_version | The record MAC and which fields it covers (1, 2 or 3). |
Records taken before these fields existed have them empty (observed_at, purpose and result_id from before October 2026, the rule and classification fields from before this release), and their MAC does not cover them. Empty times are never filled in. The file also has field_semantics, which describes every field, and integrity.verification, the result of the record MAC check.
Read the Endpoint Group Breakdown
| Column | What it shows |
|---|---|
| Pass | Passing records of the latest snapshot on hosts that are members of this group. A host in several groups counts in each of them. |
| Fail | Failing records of the latest snapshot on hosts that are members of this group. |
| Pass Rate | Passing records out of all records of this group's hosts in the latest snapshot, without severity weights. |
Download framework reports
All buttons are in the framework's card on Frameworks & evidence. They use the latest snapshot, so take a snapshot first if you need current data.
| Button | What you get |
|---|---|
| JSON | The snapshot's evidence records as a JSON file, for tools and archives. |
| A PDF report of the framework. | |
| Stakeholder Report | A branded PDF for one reader. Pick the reader from the menu (see below). |
| Download All (ZIP) | All five stakeholder reports in one ZIP file. |
| Reader in the menu | Focus |
|---|---|
| IT Manager | Detailed technical findings and remediation |
| CISO | Risk posture, drift and security priorities |
| GRC (Technical) | Check results with their requirement mapping, sample/simulated labels and rule origin |
| GRC (Legal) | Plain-language results, scope limits and the framework's legal context |
| GRC (External Auditor) | Every evidence record with observation time, run, result ID and record MAC, and how the MACs were checked |
Read a stakeholder report
Every report, for every framework, has the same core:
- Summary of Technical Check Results: how many evidence records passed and failed, which snapshot it is based on and the range of times the results were observed, how many checks are mapped to how many requirements, and how many have only the framework tag. Requirements without a check, and organizational measures, are listed as not assessed.
- Scope limitations: simulated and sample checks, custom or locally changed rules, legacy records and results older than the framework's snapshot interval.
- Times in This Report: Report generated at (when the PDF was made; downloading it again changes only this time), Snapshot collected at (when the snapshot was taken) and Result observed at (when the hosts reported each result). The score and the summary always use the snapshot time, never the time you downloaded the report. Times are in UTC; API callers can choose another time zone with
timezone(for exampleEurope/Berlin), and each time then shows its UTC offset. - How the Score Was Calculated: passing records out of all records in the snapshot, and what a record MAC does and does not prove.
Use the other report tabs
The other tabs of Reports show current data. Each has CSV and PDF buttons at the top right.
| Tab | What it shows |
|---|---|
| Hosts | Every host with status, worst severity, distribution, IP address and last seen. |
| Rules | Every rule with category, active, status, severity, number of validations and remediations, and pass rate. |
| Results | The latest result of every rule on every host, with return code and output. Filter All, Pass, Fail, Info. |
| Drift | Active Drift Events, Affected Frameworks and the Snapshot History of every framework. |
| Audit | Every recorded change (who, what, when). Filter INSERT, UPDATE, DELETE. |
Read the Hosts table
| Column | What it shows |
|---|---|
| Status | Whether the host passes all its rules: COMPLIANT when every assigned rule passes, NON-COMPLIANT when one fails. NO RESULTS means no rule has reported yet; informative results are ignored. |
| Worst Severity | Highest severity among the rules that currently fail on the host. Empty when nothing fails. |
| Distribution | Name of the operating system distribution the agent reports, for example Ubuntu. |
| Version | Version of that operating system distribution, as the agent reports it. |
| Last Seen | When the host last sent a rule result, or when its agent last checked in if it has no results yet. |
Read the Rules table
| Column | What it shows |
|---|---|
| Category | Free grouping word set on the rule, for example access_control or network_security. |
| Active | Whether the rule is switched on: only active rules run on hosts and count in compliance. |
| Status | The rule's verdict across all hosts: FAILING when it fails on any host, PASSING when it passes and fails nowhere. INFO means it has no verdict yet or no expectations; INACTIVE means it is switched off. |
| Severity | How serious a failure of the rule is; it weighs the rule in the compliance score. |
| Validations | Number of commands the rule runs on hosts, listed under Validations in the rule editor. |
| Remediations | Number of switched-on fix commands in the rule. Library rules ship with their fix switched off, so they show —. See Fix hosts automatically. |
| Pass Rate | Hosts where the rule's latest result passes, out of all hosts with a result for the rule. INFO results count as not passing. |
Read the Results table
| Column | What it shows |
|---|---|
| Status | Verdict of the rule's latest result on this host: PASS when all expectations are met, FAIL when one is not, INFO when there is no verdict. |
| Return Code | Exit code of the rule's check command on the host; 0 usually means the command ran without error. |
| Last Execution | When the host last ran the rule and sent its output. |
| Stdout | The raw text the check command printed on the host, which the server compares with the rule's expectations. Hover to see all of it. |
Read the Snapshot History
| Column | What it shows |
|---|---|
| Snapshot ID | Start of the snapshot's unique ID, which every evidence record of the snapshot carries. |
| Hosts | Number of active hosts in the framework's Endpoint Group Scope when the snapshot was taken. |
| Controls | Number of rule-to-framework mappings when the snapshot was taken: one per active rule and mapped requirement, and one for each rule with only the framework tag. |
| Pass Rate | Passing records out of all records in the snapshot, without severity weights. A down arrow means it dropped since the previous snapshot. |
| Triggered By | How the snapshot started: schedule for the automatic interval, manual for the Snapshot button, the API or the CLI. |
The columns of the Audit tab are explained in Read the audit log.
Read the audit log
The Audit page and the Audit tab of Reports show the audit log. Auditors use it to see who changed rules, groups, hosts and settings.
Configuration changes
| Column | What it shows |
|---|---|
| Actor | Sign-in name of the user or token that made the change. Empty or - when LEA Platform made the change itself. |
| Record type | The kind of record that changed, for example Rule, Group or Host. Hover it for the internal table name (cmds, client_groups, clients), which exports keep. |
| Action | What happened to the record: INSERT created it, UPDATE changed it, DELETE removed it. BLOCKED and REJECTED mark a one-off command that was refused. DENIED marks a request refused because the role or scope did not allow it. |
| Changes | Short summary: the name or ID of a created or deleted record, or the fields an update changed. Expand a row on the Audit page for the values before and after. |
| Record Name | Name of the rule, group or host the change was about, when it can be looked up. For a DENIED row: "Denied:", the permission that was missing and the request, for example Denied: Edit rules on POST /api/cmds. |
| Actor kind | What kind of identity acted: Person, Application, Service account, Local user, Server tool, or LEA Platform for its own background jobs. |
| Via | How the request reached LEA Platform, for example Console, Personal token, Slack or Teams, or AI assistant. The other values are Service token, User name and password, Legacy token, Identity provider token, Server tool, and Approval when an approved request was run. |
| On behalf of | The identity the actor acted for. On an approval decision it is the person who asked for the approval. |
| Approval | The approval request behind the change, when it was requested, decided or run through an approval. Click it to open the request. |
| Request ID | Identifier of the request that made the change; entries that share it were made together by one action. |
On the Configuration changes tab:
- Click Denied in the summary, or pick it next to the search, to see only refused requests. Expand a row for the permission, the request and the error.
- Use Filters to pick an Actor kind or Via, for example everything done through Slack or Teams, or by service accounts. The filters stay in the page address, so you can share the view.
- Expand any row to see the actor's ID, the token used and the approval ID, when they apply.
Rows written before LEA Platform 2.1 have no actor kind, via or approval.
Notifications
| Column | What it shows |
|---|---|
| Host name | The host whose compliance or managed state changed and caused the entry. |
| Action | What the entry records: COMPLIANCE for a change in the host's compliance or managed state, otherwise the name of the webhook that was called. A name ending in _ERROR means sending failed. |
| Method | HTTP method used to call the webhook (GET, POST, PUT, PATCH or DELETE). Empty for COMPLIANCE entries. |
| Response | HTTP status code the webhook receiver returned; 200 to 299 means delivered and _ERROR entries show 500. COMPLIANCE entries show the host's new state instead: COMPLIANT, NON-COMPLIANT, NO RESULTS or INACTIVE. |
| Details | Why the entry was written: the error of a failed call, the state change of a COMPLIANCE entry, or the event that triggered the webhook. A state change reads, for example, "Compliant → Non-compliant". |
On the Notifications tab:
- Delivery success counts only webhook calls, not
COMPLIANCEentries. Click it, or pick Failed deliveries next to the search, to see only failed calls. - Expand a row to see the full error in red, Rule results of the host (its Rule results tab, where the failing checks are) and Open in webhook log, which shows each attempt of that webhook with its request and response.
- A
COMPLIANCEentry with NON-COMPLIANT is not a failed delivery: the host now fails at least one check. Open Rule results of the host to see which.
Export tables from other pages
The Compliance, Execution and Audit pages have an export button (download arrow) at the top right of the page header.
- Set the filters on the page. The export uses the same filters and the visible columns.
- Click the export button.
- Choose CSV (Comma-Separated Values) or JSON (JavaScript Object Notation).
- Click Export.
From a terminal with the LEA command line: lea compliance (framework overview), lea compliance evidence --framework CIS --result fail and lea compliance report CIS --format pdf. See the CLI reference.
Scripts can call the API directly (administrator token for the first two): POST /lea/compliance/snapshot/trigger with {"framework": "ISO27001"} takes a snapshot; POST /lea/compliance/report with {"framework": "CIS", "format": "pdf"} downloads a report (format is json, pdf or zip; add "audience" (it_manager, ciso, grc_technical, grc_legal or grc_auditor) for a stakeholder PDF, "snapshot_id" for an older snapshot, and "timezone" (for example Europe/Berlin, default UTC) for the times in PDFs); POST /lea/compliance/snapshots/<id>/verify checks every record MAC of a snapshot; GET /lea/compliance/snapshots?framework=CIS lists snapshots and GET /lea/compliance/evidence?framework=CIS&result=FAIL lists evidence records. See the API reference.
Improve a framework score
- Find what fails: on the Dashboard, click the framework in a frameworks widget to see its controls and how each one does. You can also ask the AI assistant for the framework's failing controls, when it is turned on.
- Fix the failures with the highest severity on the most hosts first. They weigh most in the score. See Read results and fix failures.
- Close coverage gaps. A rule with no result on a host is missing, not passing: make sure every host in scope runs every mapped rule. See Groups.
- When the fixes show PASS, click Snapshot so the evidence shows the new state.
Give an auditor what they need
- Agree with the auditor which framework and which hosts are in scope.
- In Settings → Compliance frameworks, enable that framework and set Endpoint Group Scope to the groups in scope. Click Save.
- Check that the rules are mapped: on Rules, filter by the framework's tag (for example
std:ISO27001). Add tags where rules are missing. - Check coverage: every host in scope should have results for the mapped rules. Hosts with NO RESULTS and rules with expectations that show INFO are gaps, not passes. See Check compliance.
- Fix what you can, then click Snapshot on Reports.
- Download:
- Stakeholder Report → GRC (External Auditor) (evidence records with observation times, provenance and the record MAC check),
- JSON (all evidence records with full-precision times and record MACs),
- optionally Download All (ZIP).
- On the Audit tab, download PDF or CSV to show who changed rules, groups and settings.
- Note the Key Fingerprint from Settings → Compliance frameworks, so the auditor can see which key the record MACs were computed with.
- Tell the auditor what the evidence is: automated check results that support their own testing. Record MACs can only be checked with the key, which stays on your LEA Platform; the auditor cannot check them independently.
Tip: Snapshots are kept, so you can show the trend over time in Drift → Snapshot History.
If something goes wrong
- "No compliance data yet": no framework is enabled, or none has data. Enable a framework and click Snapshot.
- JSON, PDF, Stakeholder Report and Inspect are greyed out: the framework has no snapshot yet. Click Snapshot.
- A framework has no controls: no active rule carries a tag for it (for example
std:CISorstd:CIS:<control>). Add tags. - A framework is missing on Reports: it is disabled. It is listed under "Disabled frameworks".
- A new tag does not show a framework: no active rule carries it yet, or the ID has characters other than letters, digits,
_,.and-. - The Report wording dialog opens when you download a PDF: LEA Platform has no wording for this standard. See Describe a standard.
- A host is missing from the evidence: it is inactive or not in the groups of Endpoint Group Scope.
- "Failed to generate report": try again; if it persists, see Troubleshooting.
Related
- Check compliance
- Write and edit rules
- Groups
- Settings
- Dashboard reference