last_updated: 2026-08-28
This repo documents how to access platform tools. It must never become a credential store or a dump of personal or customer-identifying data.
| Category | Examples |
|---|---|
| Public URLs | https://stargate.odieplat.io, https://argocd.odieplat.io |
| Service account names | plat-bootstrap, github-actions (not keys) |
| Placeholder env var / config keys | jira.api_token, access.tsh_proxy in ~/.config/platform-tools/config.yaml |
| Jira project key | PLAT |
| Architecture and MCP tool names | ikg_search, ikg_get_node |
| Generic / obfuscated examples | customer-example, tenant-a, name@example.com |
| Category | Examples |
|---|---|
| API tokens, passwords | Atlassian tokens, GitHub PATs, AWS keys |
| Private keys | -----BEGIN … PRIVATE KEY----- |
| Kubeconfig contents or paths | Cluster certs/tokens, context names tied to individuals |
| Personal machine paths | Home directories, usernames in paths, IDE config locations |
| Repo-local secret files | .env, config.yaml (use ~/.config/<repo_name>/config.yaml instead) |
| Real AWS account IDs | Unless already public in argocd-tele docs |
| Personal email addresses | Use name@example.com in examples |
| Customer-identifying data | Real customer names, tenant IDs, production hostnames, unredacted logs |
See also how-we-work.md#privacy.
- Personal: Verify access locally; document patterns in PRs. Never sync kubeconfig or home-directory paths into this repo or Jira tickets. Use
~/.config/platform-tools/config.yamlfor local secrets (see how-we-work.md#local-configuration). - Customer: Obfuscate and minimize customer names and data in docs, tickets, and agent context. Use placeholders unless information is already public and approved.
Every file under access/**/*.md must include YAML frontmatter with:
last_updatedgaps(what the reader must fill locally — without pointing at personal paths)tagsincludingeng-information
Pull requests run .github/workflows/validate-docs.yml to block common secret patterns and missing frontmatter.
If secrets or customer data were committed, rotate/review immediately and remove from history per org policy.