Index of every page on this site: llms.txt
All skills
Run it yourself

Threat-models a change that crosses a trust boundary, turns each abuse case into a failing test, triages a scanner or advisory finding, and writes a design review into the design doc. Use when adding or changing auth, login, sessions, cookies, passwords, tokens, JWT, permissions, roles, user input, file uploads, webhooks, a fetch of a user-supplied URL, secrets, API keys, CORS, CSRF, XSS, SQL injection, a vulnerability, a CVE, a dependency or npm audit finding, a threat model, personal data, PII, GDPR, or an LLM feature that handles untrusted text, or when the user asks whether something is secure. Not for reviewing a whole diff - use `review`; not for Go-specific APIs - use go; not for SQL or schema design - use database.

  • 6 files
  • 26.2 KB
  • Updated 3 weeks ago
  • GitHub
Use this Skill: https://skilld.dev/gh/avik-creator/agent-skills/security

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencesdependencies-and-scanning.md

≈1.1k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Dependencies and scanning

Read when triaging a finding, adding a dependency, preparing a release, or when the repo has no scanners. A new dependency is also reviewed with review's references/dependencies.md (maintainer, age, typosquat, license, lockfile diff). Do not repeat that review here.

Triage

Finding
├── Reachable on a runtime, build, test, or deploy path?
│   ├── yes → a fix is a candidate for this change
│   └── no, confirmed → still record it; it is not a blocker
├── Is a patched version published?
│   ├── yes → propose that upgrade, one finding at a time, after a yes
│   └── no → workaround, replace, or defer
└── Impact × likelihood, from threat-model.md

Apply one agreed upgrade after a yes, with the manager's normal command for that one package, and read the changelog between the two versions. A deferral names the reason and a review date. Do not run a bulk auto-fix (npm audit fix --force, pnpm audit --fix, pip-audit --fix, cargo audit fix) unless asked. Why: it upgrades packages nobody reviewed.

An audit matches published advisories. A new package, and a package whose name is one character off a known one, both come back clean.

What to run

Run the scanners the project already configured (CI steps, and tools it already uses, such as semgrep, gosec, CodeQL, trivy, gitleaks). Then, on a security-relevant change, if the command is already available and you do not have to install it onto PATH:

Project Command
Go govulncheck, as the go skill describes
npm npm audit
pnpm pnpm audit
Python pip-audit
Rust cargo audit

Yarn's audit command was not re-checked. Use the command that version's own docs name.

cargo audit in a project with no Cargo.lock can generate one by running Cargo, and Cargo can run the project's code. On a project you do not trust, skip it and say so.

Also run the staged secrets check from SKILL.md. It prints names, never values.

Do not install a scanner, add one to CI, or add a config unless asked. A suppression names the rule and the reason. For semgrep the comment is nosemgrep: <rule-id> on the matched line or the line above; put the reason in that comment. nostmt is not the comment.

Suggest, do not run

When the repo has no scanners, before a release, or when the user asks for a security audit, suggest this set. Run one only after a yes. Do not install it yourself.

Tool Command Setup, after a yes
gitleaks gitleaks git --redact=100 . The current command is git, not the deprecated detect. --redact=100 keeps the secret out of the output. Install a release binary from the gitleaks README.
semgrep semgrep scan --config auto Install the CLI from the Semgrep docs. --config auto pulls registry rules and the CLI docs say that reports pseudonymous metrics and the project URL. Add --metrics=off when the user does not want that.
osv-scanner osv-scanner scan source -r . Install a release binary from the osv-scanner docs. Do not go install an @latest build.

Each finding they print goes through the triage above.

Lockfile and install scripts

Checked on 2026-11-02. One lockfile per install root, committed, and CI does not rewrite it. If the manifest, the lockfile, and CI name different managers, stop and say so.

Manager Lockfile install Scripts
npm 11.19.0 npm ci requires package-lock.json or npm-shrinkwrap.json, errors if they disagree with package.json, and does not write either file --ignore-scripts skips lifecycle scripts (default is false). --strict-allow-scripts (default false) fails the install when a dependency's install scripts are outside allowScripts, instead of warning
pnpm 11.11.0 --frozen-lockfile fails if the lockfile would change; the help says this is already on in CI --ignore-scripts skips lifecycle scripts
Yarn Berry yarn install --immutable aborts if the lockfile would change (yarnpkg.com/cli/install) The same page says --mode=skip-build skips build scripts for that install. Re-read the pinned version before relying on a default

For a manager not in this table, read its current docs. Do not borrow another manager's flag.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at a0324a3. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 4 days ago.

Activeupdated 3 weeks ago

README badge

README badge for avik-creator/agent-skills/security