skywalkskills

Know what you’re modernizing

This is ClaimsPortal as Skywalk audits see it: its real modules and the imports between them. The red nodes are where your coding agent found a problem. Move your cursor through it.

web/src/pages/ClaimDetail.jsx40 lines

F3Model output rendered as raw HTML

  • web/src/pages/ClaimDetail.jsx · 40 lines · F3 Model output rendered as raw HTML
  • web/src/pages/Dashboard.jsx · 37 lines · C5 A 4 MB PNG as the hero image
  • web/src/pages/NewClaim.jsx · 62 lines · A2 Maps key bundled into client JavaScript
  • server/auth.js · 20 lines · A4 Token signature never verified; A2 Rejected bearer tokens written to the log
  • server/routes/claims.js · 100 lines · A4 Any signed-in user can read any claim; A6 Full claim body, card number included, logged
  • server/routes/documents.js · 17 lines · A1 ?path= reads any file the server can read
  • server/claims/payout.js · 21 lines · F1 Payout errors swallowed; the claim pays the unadjusted amount
  • server/llm/prompt.js · 14 lines · F3 Claimant text pasted straight into the prompt
  • server/llm/summarise.js · 18 lines · F3 No timeout; an LLM outage blocks submission
  • server/lib/resetToken.js · 7 lines · F1 Reset tokens from Math.random
  • server/schema.js · 73 lines · A6 ID numbers and medical notes in plain text
  • server/services/summaryCache.js · 7 lines · C1 Summary cache grows forever
  • .env · 6 lines · A2 AWS key pair committed: AKIA…MPLE
  • server/config.js · 13 lines · C4 Absolute upload path on local disk

Give your agent a reviewer’s judgement

A skill tells your coding agent what good looks like, what to check, and in what order. It reads your code and hands back evidence, not opinions.

Your agent

Run the A2 secrets audit on this repository, including the full git history.

a2-secrets-credential-exposure / auditgit log -p

Done. Seven findings written to findings.md and findings.json, with a rotation list ordered by blast radius.

Ask your agent…
SkillA2 Secrets scan
ReadsRepo and git history
ChecksSix, every run
EvidenceFile and line
skywalk/skills
Read-onlyNever edits code
RedactsPartial values only
FindingsJSON and Markdown
FixIts own skill
findings.md

Secrets and credential exposure

claimsportal-node, full git history

  • live-credentialsAWS key pair in a committed env file.env
  • git-history-secretsA deleted secrets file is still in git historyserver/config.local.js
  • log-credential-leaksA rejected bearer token is written to the logserver/auth.js
  • scanning-infrastructureNo scanner would stop the next onepre-commit and CI

and 3 more findings

Rotation list readyOrdered by blast radius, with a one-paragraph summary for your CISO

Don’t merge what nobody reviewed

Agents now write code faster than anyone can review it. The skills give your agent a reviewer’s checklist, so every change can be checked the same way.

Run the A2 audit on this repo

Skill loaded

a2-secrets-credential-exposure / audit / SKILL.md

Secrets and Credential Exposure Scan

Read-only. Never modify the repository.

  1. 1
    Working treeKeys, tokens and passwords in code, config and Dockerfiles
  2. 2
    Git historyDeleted files and past “removal” commits
  3. 3
    PreventionPre-commit and CI secret scanning

Watch outNever reproduce a full secret. Redact or truncate it.

Three findings from one claims portal

ClaimsPortal is a test system we built to behave like the legacy apps we meet on client work. These are three of the things the skills found in it.

A key the working tree no longer tells the truth about

ClaimsPortal keeps an AWS key pair in an env file that is committed. A later commit deletes another secrets file. The key in that deleted file is still in git history, and a rejected bearer token is written to the log.

Found. The scan reports the committed key, the history that still holds a deleted secret, the token in the logs, and the absence of a scanner that would stop the next one.

Fix. Rotate the key with the team that uses it, plan the history clean-up, and add a scan that runs before a commit is accepted.

claimsportal-node/.env
AWS_ACCESS_KEY_ID=AKIA************MPLE

Read the full evidence for all three

One catalogue for your whole team

Pick the audits you need, hand them to the agent you already use, and get the same report format from every one.

Skill families

34 skills in the preview

  • ASecurity & Compliance6
  • CPerformance, Scale & Cost5
  • DArchitecture & Code Health4
  • EDelivery & Operations5
  • FAI-Era Code Health3
  • HDelivery Documentation & QA6
  • IDevOps & Platform5

How you get them

From your email to your first audit

  • Sign inA code to your email
  • Pick two skillsFree in the preview
  • DownloadSkill and install guide
  • Run itIn Claude Code

Each skill is a folder with a SKILL.md file. We run and test them in Claude Code; other agents that load skill folders can use them too.

What you get back

The same shape from every audit

  • findings.mdFor people
  • findings.jsonFor tools
  • EvidenceFile and line
  • ReproduceSteps anyone can run
  • Fix tierT1 to T3
Signed inDownload any two skills, with install guides

Built to be safe on day one

The audit runs in your agent, against your code. Skywalk never receives your code or your findings. Here is what each skill tells your agent it may and may not do.

Skill

Secrets scanA2 audit

Runs in

Your agentUnder your accounts

Reads

Your codeRepo and git history

The skill’s rules

  • Read code and history
  • Edit your code
  • Use or test a found key
  • Show a full secret

What the audit did

  • Wrote findings.md and findings.jsonDone
  • Rotate the AWS key in .envLeft to you

When you want help

Run it yourselfRun it with us

A Skywalk engineer can run the same skills with your team and plan the fixes.

See what’s really in your system before you change it

Sign in with your email and download two skills free while we take feedback from the first teams.

Sign in to download

Use the email and password you signed up with.