agentic-scan
Read-only review skills for AI coding agents: twelve scans (security, privacy, architecture, performance, data integrity, logic, bugs, tests, UX, prompts, infra, pruning) that map an app and deliver findings as a dashboard and report.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
Its scripts were only partly scanned.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/simota/agentic-scan with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/simota/agentic-scan'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/simota--agentic-scan)HTML
<a href="/listings/simota--agentic-scan"><img src="/api/listings/simota--agentic-scan/badge.svg" alt="findsafeskills safety rating" /></a>Install
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 99 lines
- # agentic-scan
- Twelve read-only review skills for agents, kept in one repository. Each skill
- maps one aspect of an application, checks it, and delivers the findings as a
- dashboard and a report. None of them edits the code under review.
- ## The skills
- | Skill | Looks at |
- |---|---|
- | [`architecture-scan`](architecture-scan/) | Modules, dependencies and data ownership; layering, coupling, failure handling, operability |
- | [`bug-scan`](bug-scan/) | How bugs entered, from git history; the same entry points still present in code; a defect register |
- | [`data-integrity-scan`](data-integrity-scan/) | Write paths, schema and migrations; missing constraints, transaction gaps, lost updates |
- | [`infra-security-scan`](infra-security-scan/) | Terraform/OpenTofu, CloudFormation and Kubernetes config; exposure, identity, attack paths |
- | [`logic-scan`](logic-scan/) | Business rules; wrong results on realistic input, each with a counterexample |
- | [`performance-scan`](performance-scan/) | Hot paths; N+1 and unbounded queries, indexes, blocking I/O, caching, bundles |
- | [`privacy-scan`](privacy-scan/) | Personal data and where it flows; minimisation, consent, retention, third-party sharing |
- | [`prompt-scan`](prompt-scan/) | LLM prompts, templates and agent definitions; broken variables, contradictions, injection boundaries |
- | [`prune-scan`](prune-scan/) | What does not earn its place: dead code, unused features, flags, dependencies, config |
- | [`security-scan`](security-scan/) | Application source; authn/authz, cross-tenant access, injection, secrets, dependencies |
- | [`test-gap-scan`](test-gap-scan/) | The test suite; where a defect would slip through, and what to assert |
- | [`ux-scan`](ux-scan/) | Screens and flows; layout, states, forms, accessibility, microcopy |
- Each skill's own README describes its workflow and outputs.
- ## Diff mode
- Every skill can review a change instead of the whole repository: a branch
- against its base, a pull request, a commit range, or uncommitted work. Ask for
- it in the request:
- ```text
- /security-scan review only this branch against main
- /logic-scan the diff of PR 123
- /test-gap-scan my uncommitted changes
- ```
- Each skill's `scripts/diffscope.py` takes the diff from the merge base, as a
- pull request shows it, and the review covers the changed files and the code
- they reach. A finding is kept only when the change causes, exposes or worsens
- it; defects that were already there are mentioned in the chat summary, not
- recorded. The report names the range and says it is not a full assessment, so
- do not compare it with a full run. The procedure is in each skill's
- `reference/diff-mode.md`.
- ## Language
- The documentation is in English. Reports follow the user's language, chosen
- in this order:
- 1. A language named when invoking the skill, as an argument or in the
- request: `lang=ja`, `lang=en`, "in English", "日本語で".
- 2. Otherwise, the language the user has set for the agent: a host setting
- such as Claude Code's `language`, or a standing instruction such as
- "always respond in Japanese".
- 3. Otherwise, the language of the request.
- ```text
- /security-scan lang=en # English report, whatever the agent's language
- /ux-scan lang=ja # Japanese report
- ```
- The dashboards and reports are rendered in Japanese (`ja`) or English (`en`).
- For any other language the report is rendered in English and the agent's
- summary in chat uses the user's language. Each renderer also takes the
- language directly: `scripts/render.py ... --lang ja|en`.
- ## Layout
- ```
- agentic-scan/
- ├── Makefile fans a target out to every skill
- └── <name>/
- ├── Makefile link, unlink, status, test, check, demo (+ skill-specific targets)
- ├── README.md
- ├── skills/<name>/ the installable skill: SKILL.md, reference/, scripts/
- ├── examples/ sample findings used by check and demo
- └── tests/ offline tests
- ```
- Each `<name>/` directory was imported from its former standalone repository
- (`simota/<name>`) with its full git history.
- ## Usage
- From the repository root:
- ```sh
- make link # symlink every skill into claude / codex / agy skills dirs
- make status # what is linked, per skill and host
- make check # tests + self-containment checks for every skill
- make test ONLY="bug-scan ux-scan"
- make link AGENT=claude # AGENT, SKILLS_DIR, PYTHON pass through to each skill
- ```
- A failing skill does not stop the others; the run exits non-zero at the end
- and names the skills that failed.
- To work on one skill, run its own Makefile: `make -C ux-scan demo`.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.