hindsight-local
Pass
Audited by Gen Agent Trust Hub on Mar 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to use
uvxto download and execute thehindsight-embedpackage from a public registry. - [COMMAND_EXECUTION]: The skill relies on the execution of multiple shell commands via the
uvxpackage runner to manage the memory daemon and perform data operations. - [CREDENTIALS_UNSAFE]: The configuration step (
uvx hindsight-embed configure) explicitly prompts the user to input LLM provider API keys, which are then handled and stored by the externalhindsight-embedtool. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8).
- Ingestion points: Data is ingested via the
uvx hindsight-embed memory recallcommand (referenced in SKILL.md). - Boundary markers: The instructions do not define delimiters or specific 'ignore embedded instructions' warnings for recalled context.
- Capability inventory: The agent has the capability to execute shell commands and write to memory via the
uvxtool (referenced in SKILL.md). - Sanitization: There is no evidence of sanitization or validation of the content retrieved from the memory store before it is used as context for the agent.
Audit Metadata