project-memory

Pass

Audited by Gen Agent Trust Hub on Jun 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill employs several shell scripts (start_session.sh, append_capture.sh, bootstrap_memory.sh, etc.) to automate project knowledge management. These scripts use standard local utilities such as awk, sed, mkdir, and printf to create and update markdown files within the project-memory/ directory. No patterns of arbitrary command injection or execution of untrusted remote scripts were found.
  • [EXTERNAL_DOWNLOADS]: The documentation (references/qmd-integration.md) recommends the installation of the qmd search tool (@tobilu/qmd) via a package manager. The skill's scripts do not perform automated downloads or installations; they only check for the presence of the tool on the local system path to provide indexing features.
  • [PROMPT_INJECTION]: The skill instructions facilitate the organization of project logs and the retrieval of past lessons. While the agent is instructed to read and summarize project-local markdown files—which could theoretically serve as a surface for indirect prompt injection if malicious content is committed to the repository—the workflows are designed for internal project use and emphasize user review for final memory saves.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 15, 2026, 01:40 AM
Security Audit — agent-trust-hub — project-memory