code-intelligence-skills
Agent skill for running the code-intel CLI: deterministic security and test-quality evidence on the files you changed.
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
It has changed since its scan on 20 Nov 2026.
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/micio86dev/code-intelligence-skills 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/micio86dev/code-intelligence-skills'
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/micio86dev--code-intelligence-skills)HTML
<a href="/listings/micio86dev--code-intelligence-skills"><img src="/api/listings/micio86dev--code-intelligence-skills/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Install the CLI first ([instructions](https://github.com/micio86dev/code-intelligence#install)), then the skill for the agents you actually use:
npx github:micio86dev/code-intelligence-skills install --agent claude,pi npx github:micio86dev/code-intelligence-skills install --agent codex --project npx github:micio86dev/code-intelligence-skills install --agent all --dry-run
- `npx github:micio86dev/code-intelligence-skills install --agent codex --project`: this repository only - `npx github:micio86dev/code-intelligence-skills install --agent all --dry-run`: preview, change nothing
`--agent` takes `claude`, `codex`, `opencode`, `pi`, `gentle-shell` (comma separated) or `all`. Details, paths and alternatives (Claude Code plugin, `pi install`) are in [docs/installation.md](docs/installation.md).
| Agent | Skill location (user scope) | |---|---| | Claude Code | `~/.claude/skills/code-intelligence` | | Codex, OpenCode, Pi, Gentle Shell | `~/.agents/skills/code-intelligence` |
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.
.claude-plugin/marketplace.json · Manifest · 15 lines
- {
- "name": "code-intelligence-skills",
- "owner": {
- "name": "micio86dev"
- },
- "plugins": [
- {
- "name": "code-intelligence",
- "description": "Skill for running the code-intel CLI: scan changed files for secrets, vulnerabilities and unsafe code, find affected tests, mutation-test changed code.",
- "source": "./",
- "category": "security"
- }
- ],
- "description": "Agent skill for running the code-intel CLI: deterministic security and test-quality evidence on the files you changed."
- }
README.md · README · 104 lines
- # code-intelligence-skills
- The agent-integration layer for [`code-intelligence`](https://github.com/micio86dev/code-intelligence) (the `code-intel` CLI). It contains **one canonical skill** that teaches an AI coding agent *when* to run `code-intel`, how to read its JSON, and what to do (and not do) with the findings, plus a small installer and the manifests that let each agent pick it up.
- The CLI does the deterministic work (secret, SAST, dependency and architecture scans on the files you changed; affected tests; mutation testing). The skill is the agent-facing instruction manual for it. Nothing here runs a scan.
- ## What problem it solves
- Without guidance an agent either never runs the scanners, or runs a full scan after every edit and burns your time. The skill encodes the middle path:
- - run `code-intel scan --format json` once, after a security-relevant change, before saying "done";
- - treat exit code 3 ("incomplete") as **not a pass** and say which checks did not run;
- - never print or commit a secret, never silence findings by editing config, never install tools without asking;
- - use `impact` to choose tests, and `mutate` only when test quality is the question.
- See [`skills/code-intelligence/SKILL.md`](skills/code-intelligence/SKILL.md): it is short on purpose.
- ## What it deliberately does not do
- - **No orchestration, memory, model selection, skill management or code review.** Those belong to your agent framework. This repository does not replace Gentle AI, Gentle Shell, or any agent: it adds one skill to them.
- - **No code of its own that scans anything**, no MCP server, no daemon, no hooks, no background process.
- - **No per-agent copies of the workflow.** One `SKILL.md`; every integration points at it.
- - **No installation unless you name an agent**, and it never modifies an agent's configuration files.
- ## Install
- Install the CLI first ([instructions](https://github.com/micio86dev/code-intelligence#install)), then the skill for the agents you actually use:
- ```bash
- npx github:micio86dev/code-intelligence-skills install --agent claude,pi
- npx github:micio86dev/code-intelligence-skills install --agent codex --project
- npx github:micio86dev/code-intelligence-skills install --agent all --dry-run
- ```
- - `npx github:micio86dev/code-intelligence-skills install --agent codex --project`: this repository only
- - `npx github:micio86dev/code-intelligence-skills install --agent all --dry-run`: preview, change nothing
- `--agent` takes `claude`, `codex`, `opencode`, `pi`, `gentle-shell` (comma separated) or `all`. Details, paths and alternatives (Claude Code plugin, `pi install`) are in [docs/installation.md](docs/installation.md).
- | Agent | Skill location (user scope) |
- |---|---|
- | Claude Code | `~/.claude/skills/code-intelligence` |
- | Codex, OpenCode, Pi, Gentle Shell | `~/.agents/skills/code-intelligence` |
- ## First use (five minutes)
- 1. **Install the CLI and the scanners** the skill relies on (the skill never installs anything by itself):
- ```bash
- npm install -g code-intelligence
- brew install gitleaks semgrep osv-scanner
- code-intel doctor
- ```
- - `brew install gitleaks semgrep osv-scanner`: or follow what `code-intel doctor` prints
- 2. **Install the skill** for your agent, then check it landed:
- ```bash
- npx code-intelligence-skills install --agent claude
- npx code-intelligence-skills status --agent claude
- ```
- - `npx code-intelligence-skills install --agent claude`: or codex, opencode, pi, gentle-shell
- 3. **Ask your agent explicitly**, in a repository with uncommitted changes: _"Use the code-intelligence skill to check my changes."_ It should run `code-intel doctor` and then `code-intel scan --format json`, and report findings with their severity and a fix. Vague requests such as "is this safe?" may be answered with the agent's own built-in tools instead; the per-agent results are below.
- 4. **To see it catch something**, follow the ten-minute [first test in the CLI README](https://github.com/micio86dev/code-intelligence#first-test-10-minutes-nothing-touches-your-real-projects) first (it needs no agent), then repeat step 3 inside that throwaway repository.
- 5. **Remove it** whenever you like: `npx code-intelligence-skills uninstall --agent claude` removes exactly what was installed.
- ## Verification status per agent
- Whether each agent really discovers and *uses* the skill was tested by running it headless on a throwaway repository, and the results, limitations and exact versions are recorded in **[docs/integrations/](docs/integrations/README.md)**. An agent is only listed as working there if a real invocation was observed. Anything not observed is marked **unverified**.
- ## Use with several agents
- All agents read the same skill text and call the same CLI, so the behavior is consistent. Install it for each agent you use; agents that share `~/.agents/skills` need it only once.
- ## Update, disable, remove
- ```bash
- code-intelligence-skills status --agent all
- npx github:micio86dev/code-intelligence-skills install --agent claude
- npx github:micio86dev/code-intelligence-skills uninstall --agent claude
- ```
- - `npx github:micio86dev/code-intelligence-skills install --agent claude`: update
- - `npx github:micio86dev/code-intelligence-skills uninstall --agent claude`: removes exactly what it installed
- The installer records what it installed in `.code-intelligence-skills.json`. It refuses to overwrite a directory it did not create, never writes through a symlink, never replaces files you edited unless you pass `--force`, and keeps any file you added next to the skill when uninstalling. To remove everything: run `uninstall --agent all` (and `--project` for per-repository installs), then uninstall the CLI.
- ## Layout
- ```
- skills/code-intelligence/ the skill (SKILL.md + references/)
- installer/install.mjs zero-dependency installer
- .claude-plugin/ Claude Code plugin + marketplace manifest (points at skills/)
- package.json "pi" Pi / Gentle Shell package manifest (points at skills/)
- snippets/AGENTS.md optional always-on hint
- docs/ installation, architecture, integrations, troubleshooting
- ```
- ## Documentation
- [Installation](docs/installation.md) · [Architecture](docs/architecture.md) · [Integration guides](docs/integrations/README.md) · [Troubleshooting](docs/troubleshooting.md) · [Contributing](CONTRIBUTING.md) · [Security](SECURITY.md) · [Changelog](CHANGELOG.md) · [Roadmap](ROADMAP.md)
- ## License
- MIT
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.