supabase-observability
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
supabaseCLI andnpx supabaseto execute a variety of diagnostic subcommands. These includeinspect db table-sizes,index-usage,cache-hit,seq-scans, andlong-running-queries. These tools are used to extract performance metrics from linked Postgres instances and are restricted by theallowed-toolsconfiguration in the skill's manifest. - [EXTERNAL_DOWNLOADS]: The skill references official Supabase development tools, including the
supabaseCLI and the@supabase/supabase-jslibrary. These are standard packages provided by a well-known service and are used for their intended purpose of project monitoring and management. - [DATA_EXFILTRATION]: The instructions include steps to configure log drains that forward project telemetry (API, Auth, and Postgres logs) to external services like Datadog and Logflare. This is a legitimate feature for centralized monitoring and does not involve accessing or sending sensitive local files such as SSH keys, AWS credentials, or environment files.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources that could contain untrusted content, which represents a vulnerability surface.
- Ingestion points: Data enters the agent context through
pg_stat_statements(query logs) inreferences/query-analytics-and-log-drains.md, log drain events inreferences/query-analytics-and-log-drains.md, and custom metrics collected by Edge Functions inreferences/custom-metrics-and-health-checks.md. - Boundary markers: Absent. The instructions do not specify the use of delimiters or specific instructions to ignore potentially malicious content within the logs or metrics.
- Capability inventory: The skill is capable of executing shell commands via the
supabaseCLI (defined inSKILL.md) and performing database writes (defined inreferences/diagnostic-scripts.md). - Sanitization: Absent. There is no evidence of filtering or escaping logic applied to the data retrieved from query previews or log streams before it is presented to the agent.
Audit Metadata