jbvc/continuous-learning-v2
Instinct-based learning system that observes sessions via hooks, creates atomic instincts with confidence scoring, and evolves them into skills/commands/agents.
32
40%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
Low
Low-risk findings.
1 low severity finding. Worth noting, but not necessarily harmful.
W011: Third-party content exposure detected (indirect prompt injection risk).
What this means
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
Why it was flagged
The skill includes an explicit import command (scripts/instinct-cli.py cmd_import) that fetches arbitrary HTTP(S) URLs and writes parsed "instinct" files into the inherited instincts directory (also exposed via /instinct-import in SKILL.md), and those imported instincts are then read and used by the system to influence evolution/behavior (e.g., auto-approval thresholds and evolve/generate flows), so untrusted third-party content can materially change agent behavior.
Where we found it
arbitrary HTTP(S) URL fetch via urllib.request.urlopen in cmd_import
url · 14 sites
The cmd_import function fetches arbitrary HTTP(S) URLs using urllib.request.urlopen, parses the response as instinct files, and writes them to the inherited directory. These imported instincts are then loaded by load_all_instincts() and used by status, evolve, and export commands to influence agent behavior (auto-approval thresholds, skill/command/agent generation).
- Audited
- Security analysis