nano-doctor
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes and sources local shell scripts to perform diagnostic tasks and manage telemetry:
~/.claude/skills/nanostack/bin/lib/skill-preamble.sh(sourced)~/.claude/skills/nanostack/bin/nano-doctor.sh(executed)~/.claude/skills/nanostack/bin/lib/skill-finalize.sh(sourced)- [EXTERNAL_DOWNLOADS]: Recommends the command
npx create-nanostackfor repair or reinstallation, which downloads and executes code from the npm registry. - [PERSISTENCE]: The
--fixflag allows the skill to modify the agent's persistent configuration in.claude/settings.jsonby addingPreToolUsehooks for tools like Bash and file editing. - [INDIRECT_PROMPT_INJECTION]: The skill processes output from a local diagnostic script to generate summaries for different user profiles, which creates an attack surface for indirect instruction injection.
- Ingestion points: Standard output from the
~/.claude/skills/nanostack/bin/nano-doctor.shscript is read by the agent to determine the health status and generate the report. - Boundary markers: The skill uses markdown comments (e.g.,
<!-- guided-output:start -->) to define response templates but does not employ specific delimiters for the script's raw output to prevent it from being interpreted as instructions. - Capability inventory: The skill has the capability to execute shell scripts and modify the
.claude/settings.jsonconfiguration file. - Sanitization: There is no evidence of sanitization or verification of the script's output before it is used to determine the summary logic for 'Guided' vs 'Professional' profiles.
Audit Metadata