kmp-project-docs-maintainer
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONPERSISTENCEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Multiple scripts, including
heal_docs.py,heal_project.py,generate_skills_lock.py, andnew_task.py, use thesubprocess.runmodule to execute external commands. These calls are primarily used for git operations (e.g.,git ls-files,git diff,git grep) and to run internal project maintenance scripts. - [PRIVILEGE_ESCALATION]: The script
scripts/heal_project.pycontains logic to automatically grant executable permissions (chmod +x) to files found inscripts/,tools/, andhooks/directories. This behavior ensures that maintenance tools can be executed within the development environment. - [PERSISTENCE]: The
scripts/heal_project.pyscript manages the installation of a gitpre-commithook. It copies a script from the project'shooks/directory to.git/hooks/pre-commit, causing it to run automatically whenever a commit is attempted. - [EXTERNAL_DOWNLOADS]: The
scripts/check_skills_lock.pyscript initiates network connections to the GitHub API (api.github.com) to fetch release data. This is used to verify if the project's skill lockfile is up to date with the latest upstream versions. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted project content such as README files and source code to synchronize documentation. It possesses the necessary capabilities to write files and execute commands, creating a potential surface for indirect prompt injection.
- Ingestion points: Processes all markdown files in the project's documentation folder (
heal_docs.py) and scans Kotlin source files for TODO comments (heal_project.py). - Boundary markers: No explicit delimiters or instructions are used to isolate ingested data from the agent's instructions.
- Capability inventory: File system write access (
Path.write_text), file deletion (Path.unlink), and command execution viasubprocess.run(git commands and maintenance scripts). - Sanitization: Implements basic regex-based sanitization for task slugs in
new_task.py.
Audit Metadata