fix(security): CLI secrets via stdin/file; cap HN comment fan-out (M5/M7) - #18
Conversation
…/M7) PR-H of the security-hardening plan. - M5: secret-bearing CLI args (vault token, OAuth client secret, imported cookies) can now be read from stdin or a --*-file flag instead of a positional/flag value that leaks into shell history and process listings. The explicit arg still works but warns; an interactive TTY with no input errors instead of hanging. - M7: hackernews comments.list caps the comment fan-out (limit param, default 50, hard max 200) instead of fetching every kid, so one request can't spawn thousands of upstream fetches. The action declares `limit` and its pagination max matches. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f0db529baf
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| what: &str, | ||
| ) -> anyhow::Result<String> { | ||
| use std::io::{IsTerminal, Read}; | ||
| if let Some(v) = value { |
There was a problem hiding this comment.
When a caller uses the documented stdin form from docs/design/security-hardening.md:252 (-), this branch treats - as the deprecated explicit secret and returns it before reading stdin. In commands like relais vault store github - < token.txt or relais auth custom --client-secret - ..., the stored OAuth/vault secret becomes the single character - and the piped secret is ignored, so the safe stdin path silently produces unusable credentials.
Useful? React with 👍 / 👎.
1 participant
PR-H of the security-hardening plan — the final PR (M5/M7).
M5 — CLI secret intake
Secret-bearing CLI args (
vault storetoken,auth customclient secret,auth import-cookiescookies) can now be supplied via stdin or a--*-fileflag instead of a positional/flag value that leaks into shell history and process listings. A sharedread_secret(value, file, what)resolves arg → file → stdin. The explicit arg still works but warns (and never logs the value); an interactive TTY with no piped input errors instead of hanging.M7 — Hacker News fan-out cap
comments.listpreviously fetched every kid of a story sequentially (one request → potentially thousands of upstream fetches). It now honors alimitparam (default 50, hard max 200), fetchestake(limit), and reportshas_nexton truncation. The action declareslimitand its paginationmax_limitmatches the 200 cap.Tests
CLI parser tests updated for the optional secret args;
read_secretresolution verified e2e (stdin /--token-file/ positional-warns);cargo test --workspacegreen; clippy clean.Codex-reviewed; the flagged MEDIUM (undeclared
limit/ pagination-max mismatch) and the TTY-hang nicety are fixed.This completes the security-hardening plan: 5 HIGH, 7 MEDIUM, 3 LOW from the whole-repo review, across 8 PRs.
🤖 Generated with Claude Code