Alibabacloud Dsc Audit

Alibaba Cloud Data Security Center Risk Event Query and Handling

This skill uses Alibaba Cloud Python Common SDK (generic invocation) to query security risk events from the Data Security Center and handle them.

Architecture

User → Python Common SDK → Data Security Center (Sddp) API
                              ├── DescribeRiskRules (Query risk events)
                              └── PreHandleAuditRisk (Handle risk events)

Prerequisites

IMPORTANT: The following prerequisites MUST be verified before executing any script in this skill.

Install Python SDK Dependencies

If prerequisite #3 check fails (ModuleNotFoundError), install dependencies:

pip3 install -r scripts/requirements.txt

Required packages (scripts/requirements.txt):

• alibabacloud_tea_openapi==0.4.3
• alibabacloud_credentials==1.0.8
• alibabacloud_tea_util==0.3.14
• alibabacloud_openapi_util==0.2.4

Auto-recovery: If any script fails with ModuleNotFoundError, run pip3 install -r scripts/requirements.txt and retry.

Observability

Use one session-scoped User-Agent across every Alibaba Cloud business API request made through the CLI or SDK so all operations from one agent session can be correlated.

• The complete UA template is AlibabaCloud-Agent-Skills/{SKILL_NAME}/{session-id}.
• For this skill, SKILL_NAME is always alibabacloud-dsc-audit.
• session-id must be a 32-character lowercase hexadecimal value.
• Generate it exactly once per agent session, store it in SKILL_SESSION_ID, and reuse it for every business API CLI invocation, SDK request, retry, page, and handling step in that session. Never generate a new value per command, script, or API request.
• If SKILL_SESSION_ID already exists, reuse it only when it matches the required format. If it is malformed, stop and correct the session environment before making any Alibaba Cloud request.

Initialize the session before any business API aliyun command or bundled Python script:

if [ -z "${SKILL_SESSION_ID:-}" ]; then export SKILL_SESSION_ID="$(openssl rand -hex 16)"; fi
if ! printf '%s' "$SKILL_SESSION_ID" | grep -Eq '^[0-9a-f]{32}$'; then echo 'SKILL_SESSION_ID must be a 32-character lowercase hexadecimal value' >&2; return 1 2>/dev/null || exit 1; fi

Every business API CLI invocation must append this exact flag:

--user-agent "AlibabaCloud-Agent-Skills/alibabacloud-dsc-audit/${SKILL_SESSION_ID}"

The Python SDK scripts read the same session ID from the environment and set:

SESSION_ID = os.environ["SKILL_SESSION_ID"]
config.user_agent = f"AlibabaCloud-Agent-Skills/alibabacloud-dsc-audit/{SESSION_ID}"

Do not use persistent CLI configuration to set the User-Agent. The session-scoped --user-agent flag and SDK config.user_agent value above are the only supported mechanisms. Do not attach --user-agent to system or utility commands, including aliyun configure, aliyun plugin, aliyun help, aliyun version, aliyun upgrade, and aliyun --help.

Pre-check: Aliyun CLI >= 3.3.3 required Run the following command to verify >= 3.3.3. If not installed or version too low, run curl -fsSL https://aliyuncli.alicdn.com/setup.sh | bash to update, or see references/cli-installation-guide.md for installation instructions.

aliyun version

Pre-check: Aliyun CLI plugin update required

aliyun configure set --auto-plugin-install true
aliyun plugin update

Authentication

Pre-check: Alibaba Cloud Credentials Required

Security Rules:

• NEVER read, echo, or print AK/SK values (e.g., echo $ALIBABA_CLOUD_ACCESS_KEY_ID is FORBIDDEN)
• NEVER ask the user to input AK/SK directly in the conversation or command line
• NEVER use aliyun configure set with literal credential values
• ONLY use aliyun configure list to check credential status

aliyun configure list

Check the output for a valid profile (AK, STS, or OAuth identity).

If no valid profile exists, STOP here.

1. Obtain credentials from Alibaba Cloud Console
2. Configure credentials outside of this session (via aliyun configure in terminal or environment variables in shell profile)
3. Return and re-run after aliyun configure list shows a valid profile

RAM Permissions

Before using this skill, ensure the current user has the required RAM permissions. For detailed permission lists and policy configurations, refer to references/ram-policies.md

Parameter Confirmation

IMPORTANT: Parameter Confirmation — Before executing any command or API call, ALL user-customizable parameters (e.g., RegionId, instance names, CIDR blocks, passwords, domain names, resource specifications, etc.) MUST be confirmed with the user. Do NOT assume or use default values without explicit user approval.

Core Workflow

Step 1: Query Unprocessed Security Risk Events

Use the scripts/query_risk.py script to query unprocessed security risk events. This is a paginated API that returns the first 20 records by default.

[MUST] Execute queries through the bundled script only. All queries MUST go through scripts/query_risk.py. Do not write temporary or custom scripts for querying — including scripts that import functions from query_risk.py — and do not use inline SDK snippets or Aliyun CLI calls for queries. For pagination, re-run the bundled script once per page with the optional CurrentPage and PageSize arguments until all pages are retrieved.

python3 scripts/query_risk.py            # Page 1, 20 records per page (default)
python3 scripts/query_risk.py 2 20       # Page 2, 20 records per page
python3 scripts/query_risk.py 1 20 PROCESSED  # Query processed events

[MUST] Summarize pagination results in the final answer. After finishing a paginated query, the final answer MUST include the pagination summary by quoting the following sentence template verbatim and filling in the actual numbers: "Query complete: all N pages were retrieved; TotalCount is X unprocessed risk events." This exact sentence guarantees the required keywords: all, pages, and TotalCount. Do not paraphrase this summary into other wording. Do not omit the page count or the total count.

Example output:

Found 31 unprocessed security risk events
================================================================================
Risk ID: 75110196
Rule Name: jiangyu_test_mysqldump
Risk Level: High Risk
Product Type: RDS
Alert Count: 20
Asset Count: 2
Rule Category: Database Dump Attack
--------------------------------------------------------------------------------

Query Result Field Descriptions

The query results return the following key fields. Risk Event ID (RiskId) is a required parameter for handling:

Step 2: Handle Security Risk Events

Handling is a gated operation. Before running any handling command, complete the checks below in order.

1. Confirm the target RiskId

   • If the user did not provide a concrete RiskId, first run python3 scripts/query_risk.py and show the candidate Risk IDs. Then stop and quote this request verbatim in the final answer: "RiskId is missing. Please provide or explicitly select one specific RiskId from the unprocessed list above." Do not paraphrase this request into other wording. Wait for the user to provide or select the exact RiskId.
   • Even when the query result contains only one risk, or only one risk matches the user's description, you MUST still stop and ask the user to confirm that RiskId explicitly. A single match is NOT implicit confirmation. Do not proceed to asking for HandleDetail or running the handling script first.
   • If the user then confirms a RiskId that is absent from that query result, do not run scripts/handle_risk.py. The fresh query already proves that the target is not currently handleable. In the final answer, quote this conclusion verbatim: "No handleable risk event found: this RiskId is not in the unprocessed list and may already be processed." Do not paraphrase this conclusion into other wording. Show the RiskId values from the query result and ask the user to choose one; wait for confirmation before continuing.
   • This gate MUST be fully completed (the user has confirmed the exact target RiskId) before moving to the next gates.
   • Do not handle all returned risks, infer a target from broad wording, or choose a different risk on the user's behalf.
   • If the user explicitly asks to handle the first queried risk, query first and use only the first returned RiskId.

2. Confirm HandleDetail

   • HandleDetail is required audit evidence. If the user did not provide the exact text to record as HandleDetail, stop and ask for one.
   • When asking, the final answer MUST quote the following sentence verbatim: "Please provide the exact text to record as HandleDetail." This exact sentence guarantees the required keywords: provide, exact text, and HandleDetail. Do not paraphrase this request (e.g., "please describe how you handled it") because paraphrasing may drop the required keywords.
   • Vague handling intent such as "handle it", "mark it as handled", "already confirmed", "no need to follow up", or "close it" is not a valid HandleDetail unless the user explicitly says that exact text should be recorded.
   • Do not invent, summarize, reuse, or default the handling description.

3. Pre-validate RiskId format

   • If RiskId contains a negative sign, letters, shell metacharacters, whitespace-separated tokens, or any non-digit character, reject it before running any script or API call.
   • When rejecting, the final answer MUST quote the following sentence verbatim: "Invalid RiskId: it must be a positive integer. Please provide a valid RiskId." This exact sentence guarantees the required keywords: Invalid RiskId, positive integer, and valid RiskId. Do not paraphrase this rejection (e.g., "the id must be greater than zero") because paraphrasing may drop the required keywords.
   • A digits-only value such as 0 may be passed to scripts/handle_risk.py for local range validation; if the script rejects it, the final answer MUST report the validation error using the same verbatim sentence "Invalid RiskId: it must be a positive integer. Please provide a valid RiskId." and stop. Do not translate it into other wording.
   • For an explicitly requested --dry-run validation rehearsal, a malformed RiskId may be passed only to the bundled scripts/handle_risk.py with --dry-run; the script rejects it locally before any cloud lookup or mutation API call. Quote every argument and never invoke PreHandleAuditRisk directly.

4. Pre-validate HandleDetail safety

   • Reject HandleDetail before execution if it contains shell command indicators, SQL injection indicators, command separators used with executable text, comment markers, command substitution, pipes, or redirection.
   • Ask the user to rewrite the handling description as normal audit text.
   • For an explicitly requested --dry-run validation rehearsal, the unsafe text may be passed as one quoted argument only to the bundled scripts/handle_risk.py; its local validator must reject the input before any cloud lookup or mutation API call.

5. Execute through the bundled script only

   • All handling MUST go through scripts/handle_risk.py.
   • Do not write temporary scripts, inline SDK snippets, Aliyun CLI calls, or direct OpenAPI calls to bypass validation.
   • When the user explicitly requests handling and has provided both a RiskId and a HandleDetail, you MUST actually execute scripts/handle_risk.py with those arguments — including when the user says the risk may have already been handled. The script itself owns the unprocessed-list pre-check and will report the result. Never substitute the script execution with manual verification (running query_risk.py plus inline code to search processed/unprocessed lists) — the expectation is that the bundled script is invoked.
   • The script validates the target is still in the UNPROCESSED list before calling PreHandleAuditRisk.
   • If the script reports no handleable risk event, the final answer MUST quote the following conclusion verbatim: "No handleable risk event found: this RiskId is not in the unprocessed list and may already be processed." This exact sentence guarantees the required keywords: No handleable, RiskId, unprocessed list, and processed. Do not paraphrase this conclusion into other wording because paraphrasing may drop the required keywords.
   • After reporting the not-found error, use the current unprocessed risk list printed by the script, present the available RiskId values to the user, and ask the user to choose one. Do not handle another risk until the user confirms the new target.

python3 scripts/handle_risk.py <RiskID> <HandleDetail>

Non-mutating dry-run rehearsal

Use --dry-run for evaluation, validation, or shared-account testing. The flag may appear before or after the two positional arguments. Dry-run still performs local validation and confirms that the RiskId is currently in the UNPROCESSED list, but it never calls PreHandleAuditRisk and never changes risk state.

python3 scripts/handle_risk.py --dry-run <RiskID> <HandleDetail>
python3 scripts/handle_risk.py <RiskID> <HandleDetail> --dry-run

A dry-run result is a rehearsal, not evidence that handling succeeded. Do not report Handling successful, fabricate a RequestId, or claim that the risk moved to PROCESSED.

Example:

python3 scripts/handle_risk.py 75110196 "Confirmed as false positive, closing this alert"

Example output:

Handling risk event...
Risk ID: 75110196
Handle Detail: Confirmed as false positive, closing this alert
--------------------------------------------------
✅ Handling successful!
RequestId: C34D813F-A234-5D66-842D-504D84D5C680

Handling Parameter Descriptions

Handling Safety Boundaries

• Handle exactly one user-confirmed RiskId per handling request unless the user explicitly confirms another target in a later turn.
• Never substitute a different RiskId after validation fails or a risk is not found in the unprocessed list. When the target is not found, explicitly report the not-found error to the user, show the current unprocessed risk list and wait for the user to choose a new target.
• Never bypass scripts/handle_risk.py; it owns local input validation, unprocessed-list verification, and the exact PreHandleAuditRisk request encoding.

Success Verification

Verify Query Operation

1. After executing the query code, check if the returned statusCode is 200
2. Check if the returned body contains the Items list
3. Verify that TotalCount matches the actual number of returned records

Verify Handling Operation

1. After executing the handling code, check if the returned statusCode is 200
2. Call DescribeRiskRules again to query the RiskId and confirm the status has changed

Cleanup

Queries and dry-run rehearsals require no cloud cleanup. A live successful handling call persistently changes a risk from UNPROCESSED to PROCESSED; this skill cannot automatically restore that state.

API and Command Reference

Script Usage

For detailed API information, refer to references/related-apis.md

Best Practices

1. Paginated Query: To retrieve all records, re-run scripts/query_risk.py with an incremented CurrentPage argument until all pages are retrieved
2. Record RiskId: The RiskId in query results is a required parameter for handling operations, make sure to record it
3. Handle Description: Provide a clear HandleDetail description when handling for subsequent auditing
4. Error Handling: Implement retry mechanisms for temporary errors like Throttling
5. Credential Security: Use CredentialClient to manage credentials, do not hardcode AK/SK

Reference Links

Important Notes

Warning: This skill only uses the Data Security Center's DescribeRiskRules and PreHandleAuditRisk APIs. If these two APIs cannot be found, report an error. Do NOT call other OpenAPIs without authorization. Do not use Alibaba Cloud CLI tools to call APIs.