.svg)
Ask your agent to run the security check.
Heeler ships eight skills your coding agent discovers on its own. Ask in plain language — scan this repo for secrets, is this dependency safe to add, run a security review before we release — and the check runs where the code is being written, left of the pull request. Each skill wraps a heelercli command, so the same check runs identically in the agent, in a terminal, and in CI.
Security checks live everywhere except where the code is written.
Security checks usually live everywhere except the place the code is actually being written.
The scan runs after they've moved on
By the time a scan runs in CI, the developer has moved on. Fixing it means a context switch back into code they've stopped thinking about.
Nobody wants to learn another CLI
Asking a developer to learn a security tool's flags is asking them to stop what they're doing. Adoption dies there, however good the scanner is.
The agent already has the context
The agent writing the code is the cheapest place to check it — it already has the repository, the diff, and the developer's attention.
Eight skills your agent finds on its own.
Each skill activates automatically when the conversation calls for it — there's no command to remember, though you can invoke one by name. Most need heelercli installed and authenticated. Secret scanning is the exception: it runs with no Heeler account at all.
heeler-secrets-scan
Scans the repo or staged changes for exposed credentials, keys and tokens. Runs fully offline with no Heeler auth, so it works as a pre-commit gate on day one. Reports each finding's validation status and what triggered it.
heeler-vulnerabilities-scan
CVE analysis across every supported ecosystem, with severity gating, baseline diffing and exploitability-aware triage.
heeler-license-check
Discovers dependency licenses and checks them against your allow and deny policy. For any failing package it proposes one to three compliant alternatives in the same ecosystem.
heeler-malicious-package-scan
Typosquatting, dependency hijacking and known-malicious detection, drawing on Heeler's package-risk intelligence — before the dependency lands.
heeler-recommended-version
The safest version to install or upgrade to — the most widely-adopted version with no active vulnerabilities, which is rarely the latest tag.
heeler-scan-all
Secrets, vulnerabilities, licenses and malicious packages in one pass, returning a unified report with a single consolidated verdict.
heeler-security-review
A structured full-repo AppSec audit across seven categories, closing with a prioritized remediation plan and a release decision. Detailed below.
heeler-threat-modeling
Gathers threat-model context from Heeler's APIs, writes a STRIDE/PASTA prompt to disk, then offers to validate the top findings against repository evidence.
What a full security review actually checks.
heeler-security-review runs the other skills as substeps — always secrets, vulnerabilities and license, plus recommended-version and malicious-package checks when it detects dependency changes — then adds analysis that goes beyond automated scanning, across seven categories.
Secrets and credential exposure
Validation status, blast radius and true-positive likelihood — not just a list of strings that look like keys.
Dependency vulnerabilities
Severity, exploitability, CVSS vectors, and reachability from network entry points.
Application-layer SSRF
Whether attacker-controlled input can reach outbound request sinks and internal targets.
License and compliance
Prohibited, unknown and high-review licenses, plus missing metadata.
Supply chain and build integrity
Pinning strategy, lockfile hygiene and CI controls — the machinery around the code, not only the code.
Repository operational security
Least-privilege tokens, logging and redaction, and risky shell patterns.
LLM and agent attack surface
Prompt injection, tool-misuse and exfiltration controls — assessed when the repository has an agent surface to assess.
And a decision at the end
Every Critical and High finding carries an entry point, propagation path, impact, a file-path evidence reference and remediation steps. The review closes with a risk register, a needs-verification list, and a verdict: ready, ready with conditions, or not ready. Theoretical best-practice notes are dropped so what's left is actionable.
Useful on day one. Enforceable once you set policy.
The scanning skills run in advisory mode when no policy is configured, and switch to strict pass/fail once one is — so the same skill helps a developer before your team has agreed a standard, and gates after.
Vulnerabilities
Advisory returns ship, ship-with-monitoring, or block for remediation — critical and actively-exploitable first, with a reachability check on the top findings. With a policy it's strict pass/fail against severity thresholds or named CVE IDs.
Baseline mode
Evaluates only findings absent from a prior baseline, so a pull request is judged on what it adds rather than the debt it inherited.
Licenses
Advisory prefers permissive licenses and flags copyleft, unknown and custom for review. With a policy it's pass/fail against your allow and deny lists.
Malicious packages
Each flag is either high confidence — a release blocker unless the risk is explicitly accepted — or needs verification, which arrives with concrete steps to take before you rip anything out.
One run, one verdict
heeler-scan-all returns four sections then a final verdict: PASS or FAIL for policy-gated checks, or ADVISORY with an explicit risk judgment where no policy is set. A missing toolchain doesn't abort the run — the other scans continue and partial coverage is reported.
One command, across every repo.
Skills install through dotagents and are discovered automatically by the agent — nothing for a developer to configure, and nothing to remember once they're in.
Install with dotagents
Add the Heeler skill pack with dotagents and install — all eight at once, or name individual skills to roll out a subset first.
Six agents, one catalog
Claude Code, Cursor, GitHub Copilot, OpenCode, Codex and VS Code — the same skills, discovered the same way, whichever your developers have chosen.
The same check in three places
Because every skill wraps a CLI command, the identical check runs from the agent, from a terminal or pre-commit hook, and in CI — one result to reconcile instead of three tools disagreeing.
Not to be confused with vetting them
This page is about the skills Heeler ships for your agents to run. Scoring the agent files already sitting in your repositories — including skills someone else wrote — is a separate capability: Agent Skills Security.
Put the check where the code gets written.
Install the skills and your developers get secrets, dependency, license and malicious-package checks just by asking for them — in the agent they already have open.
