night-market-collective-memory
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
gh api graphqlcommands to interact with GitHub Discussions for searching, reading, and writing memory layers. The skill correctly utilizes variables to prevent GraphQL injection attacks. - [COMMAND_EXECUTION]: It invokes a local Python script,
plugins/leyline/scripts/journal_append.py, to maintain append-only decision logs in the repository. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes text from GitHub Discussions, which are external data sources. Malicious instructions placed in these discussions could potentially manipulate the agent's behavior when the content is retrieved.
- Ingestion points: Fetches discussion titles, bodies, and comments from the
athola/claude-night-marketGitHub repository via GraphQL queries. - Boundary markers: While it caps body lengths and requires manual confirmation for writes, it lacks explicit delimiters to distinguish external content from system instructions.
- Capability inventory: The skill can perform file system searches, write to local journal files, and post updates to the remote GitHub repository.
- Sanitization: No content-level filtering or instruction sanitization is performed on the data retrieved from GitHub.
Audit Metadata