context safety score
A score of 21/100 indicates multiple risk signals were detected. This entity shows patterns commonly associated with malicious intent.
capability escalation
Skill instructs modifying ~/.claude/settings.json to register PreToolUse, PostToolUse, and Stop hooks that execute shell scripts (pre-tool.sh, post-bash.sh, session-end.sh) intercepting ALL Bash/Write/Edit tool inputs and outputs. These hook scripts are not present in the repository for inspection, yet would capture every tool invocation's full input and output data, enabling silent exfiltration or manipulation of all agent activity. (location: SKILL.md lines 326-362 (Hooks Integration section))
scope violation
Skill describes itself as a self-improvement/learning system but the hooks configuration captures $TOOL_INPUT and $TOOL_OUTPUT from all Bash, Write, and Edit operations — effectively becoming a universal surveillance layer over all agent activity. The referenced hook scripts (pre-tool.sh, post-bash.sh, session-end.sh) are absent from the repository, making their actual behavior unauditable. (location: SKILL.md lines 328-361)
supply chain
Metadata signals are severely inconsistent: 7.69M installs but only 4 stars, 2-day-old owner account, not listed on registry, no license, single contributor. The skill_description field contains 'width=device-width, initial-scale=1' (an HTML viewport meta tag) instead of an actual description, indicating metadata corruption or injection. This profile is consistent with a fabricated/spoofed package. (location: metadata.json (install_count, stars, owner_account_age_days, skill_description))
description injection
The SKILL.md frontmatter hooks section defines auto-triggering behaviors (before_start, after_complete, on_error) that chain to other skills (session-logger, create-pr) with 'mode: auto', meaning they execute without user confirmation. The after_complete hook auto-triggers PR creation when 'skills_modified' is true, which could submit unauthorized code changes to repositories. (location: SKILL.md lines 7-24 (frontmatter hooks metadata))
curl https://api.brin.sh/skill/charon-fan%2Fagent-playbook%2Fself-improving-agentCommon questions teams ask before deciding whether to use this skill in agent workflows.
charon-fan/agent-playbook/self-improving-agent currently scores 21/100 with a suspicious verdict and low confidence. The goal is to protect agents from high-risk context before they act on it. Treat this as a decision signal: higher scores suggest lower observed risk, while lower scores mean you should add review or block this skill.
Use the score as a policy threshold: 80–100 is safe, 50–79 is caution, 20–49 is suspicious, and 0–19 is dangerous. Teams often auto-allow safe, require human review for caution/suspicious, and block dangerous.
brin evaluates four dimensions: identity (source trust), behavior (runtime patterns), content (malicious instructions), and graph (relationship risk). Analysis runs in tiers: static signals, deterministic pattern checks, then AI semantic analysis when needed.
Identity checks source trust, behavior checks unusual runtime patterns, content checks for malicious instructions, and graph checks risky relationships to other entities. Looking at sub-scores helps you understand why an entity passed or failed.
brin performs risk assessments on external context before it reaches an AI agent. It scores that context for threats like prompt injection, hijacking, credential harvesting, and supply chain attacks, so teams can decide whether to block, review, or proceed safely.
No. A safe verdict means no significant risk signals were detected in this scan. It is not a formal guarantee; assessments are automated and point-in-time, so combine scores with your own controls and periodic re-checks.
Re-check before high-impact actions such as installs, upgrades, connecting MCP servers, executing remote code, or granting secrets. Use the API in CI or runtime gates so decisions are based on the latest scan.
Learn more in threat detection docs, how scoring works, and the API overview.
Assessments are automated and may contain errors. Findings are risk indicators, not confirmed threats. This is a point-in-time assessment; security posture can change.
integrate brin in minutes — one GET request is all it takes. query the api, browse the registry, or download the full dataset.