From 209e63cbff98ec485f155f32f10f5832e82a92f7 Mon Sep 17 00:00:00 2001 From: yakimoto <66892052+yakimoto@users.noreply.github.com> Date: Thu, 6 Aug 2026 13:20:32 -0400 Subject: [PATCH 01/11] ci: this repo's public-repo-guard never scanned a single issue or comment body MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Measured across all 28 public wave-av repos (claude-workstation#1747, #1794): TWO coverage shapes satisfy the one required check name `Secrets + content policy`. 27 repos triggers: pull_request, push, workflow_dispatch jobs: guard 1 repo triggers: + issues, issue_comment jobs: + body-guard This repo is in the 27. All 28 report the same green check. The outlier is wave-moq-edge, and its own comment says why it matters: "`edited` matters as much as `opened`: a body can be made to leak long after the PR is first raised, and until this workflow covered it, nothing ever re-scanned." A PR/issue/comment BODY is exactly as world-readable as the tree, and until now it was scanned by nothing server-side. That gap was not theoretical on wave-moq-edge: a PR was blocked for naming a private repo in wrangler.toml while the very same name, with more operational detail attached, sat unchallenged in its body. WHAT LANDS HERE — the bundle the workflow's own header names, minus what this repo already has (.gitleaks.toml and content-policy.sh are already vendored): .github/workflows/public-repo-guard.yml replaced (73 -> 163 lines) scripts/public-repo-guard/body-policy.sh new, mode 100755 scripts/public-repo-guard/tests/body-policy.test.sh new, mode 100755 Copied from wave-moq-edge, which has run this shape in production. Modes preserved via the git trees API — the contents API would have created both scripts 100644. HONEST ABOUT WHAT IT CAN DO. On a PR this PREVENTS the merge. On an issue or comment the text is already public the moment it posts, so this is DETECTION: it says go redact, fast. Only a client-side pre-write hook stops that class before publication. Also inherited from the reference: concurrency moves from workflow-level to PER JOB, because the two jobs want opposite behaviour. A workflow-level group forced one policy on both, and rapid body edits cancelled the tree job repeatedly — every cancelled check-run stays attached to the commit, so the PR reported UNSTABLE while the live runs were green. The body gate ships with its own fixtures and runs them in CI. Its NEGATIVE cases are the load-bearing half: a leak gate that blocks legitimate cross-repo references gets switched off, and then it protects nothing. Refs wave-av/claude-workstation#1747. Co-Authored-By: Claude Opus 5 (1M context) --- .github/workflows/public-repo-guard.yml | 100 ++++++++++++- scripts/public-repo-guard/body-policy.sh | 139 ++++++++++++++++++ .../tests/body-policy.test.sh | 108 ++++++++++++++ 3 files changed, 342 insertions(+), 5 deletions(-) create mode 100755 scripts/public-repo-guard/body-policy.sh create mode 100755 scripts/public-repo-guard/tests/body-policy.test.sh diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index 719718a..bba2f67 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -13,10 +13,11 @@ name: public-repo-guard # wave-av/.github must not be able to alter another repo's secret scanner). The # gitleaks binary is version-pinned AND SHA-256-verified before it runs. # -# To install on a new repo, copy all three files together: +# To install on a new repo, copy all four files together: # .github/workflows/public-repo-guard.yml # .gitleaks.toml # scripts/public-repo-guard/content-policy.sh +# scripts/public-repo-guard/body-policy.sh # # Scan scope: the published working TREE (gitleaks --no-git), NOT git history. The # goal is "what is public right now is clean", so a shallow checkout is sufficient. @@ -25,24 +26,44 @@ name: public-repo-guard # path glob to a repo-root `.guardignore`, or extend the repo-local `.gitleaks.toml`. on: + # `edited` matters as much as `opened`: a body can be made to leak long after the + # PR is first raised, and until this workflow covered it, nothing ever re-scanned. pull_request: + types: [opened, edited, reopened, synchronize] + issues: + types: [opened, edited] + issue_comment: + types: [created, edited] push: branches: [main, master] workflow_dispatch: +# `pull_request`, deliberately NOT `pull_request_target`: a fork PR must never get +# a write token or repo secrets just because a gate wanted to read its body. permissions: contents: read -concurrency: - group: public-repo-guard-${{ github.ref }} - cancel-in-progress: true +# Concurrency is per JOB, not per workflow: the two jobs want opposite behaviour. +# A workflow-level group would force one policy on both, and it showed: rapid body +# edits cancelled the tree job over and over, and every cancelled check-run stays +# attached to the commit, so the PR reported UNSTABLE while the live runs were green. jobs: guard: name: Secrets + content policy + # Skips issue/comment events (the tree scan has nothing to say about a comment, + # and the org should not pay for a gitleaks run every time anyone posts one) and + # skips `edited` (a title or body edit does not change the tree). + if: >- + (github.event_name == 'pull_request' && github.event.action != 'edited') + || github.event_name == 'push' + || github.event_name == 'workflow_dispatch' + concurrency: + group: public-repo-guard-tree-${{ github.event.pull_request.number || github.ref }} + cancel-in-progress: true runs-on: ubuntu-latest steps: - - uses: actions/checkout@93cb6efe18208431cddfb8368fd83d5badbf9bfd # v5.0.1 + - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 # gitleaks' GitHub Action requires a paid license for organizations; the CLI # itself is MIT-licensed and free. Pin the version AND verify the release @@ -71,3 +92,72 @@ jobs: env: GUARD_PRIVATE_REPOS: ${{ vars.GUARD_PRIVATE_REPOS }} run: bash scripts/public-repo-guard/content-policy.sh . + + # The body gate's own fixtures. Its negatives are the load-bearing half — a + # leak gate that blocks legitimate cross-repo references gets switched off, + # and then it protects nothing. Runs here so a regression is caught by CI + # rather than by a leak. + - name: body policy self-test (fixtures) + run: bash scripts/public-repo-guard/tests/body-policy.test.sh + + # The other half of a public repo's surface. `guard` above scans the published + # TREE; a PR/issue/comment BODY is just as world-readable and, until this job, + # was scanned by nothing server-side. That gap was real, not theoretical: a PR + # was blocked for naming a private repo in wrangler.toml while the very same + # name, with more operational detail attached, sat unchallenged in its body. + # + # Honest about what it can and cannot do. On a PR this PREVENTS the merge. On an + # issue or comment the text is already public the moment it posts, so this is + # detection — it tells us to go redact, fast. Only the client-side pre-write hook + # can stop that class before publication. + body-guard: + name: Body content policy + if: github.event_name == 'pull_request' || github.event_name == 'issues' || github.event_name == 'issue_comment' + concurrency: + # Keyed on the specific PR / comment / issue rather than github.ref, because + # issue events all report the default branch and a ref-keyed group would let + # two comments cancel each other, leaving one unscanned. + # + # cancel-in-progress is deliberately FALSE. Every version of a body deserves a + # verdict, the job is seconds long, and a cancelled check-run lingers on the + # commit and makes an otherwise-green PR look broken. + group: public-repo-guard-body-${{ github.event.pull_request.number || github.event.comment.id || github.event.issue.number || github.ref }} + cancel-in-progress: false + runs-on: ubuntu-latest + steps: + - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 + with: + # Only the gate's own scripts are needed — no reason to pay for the whole + # tree on every comment. + sparse-checkout: scripts/public-repo-guard + sparse-checkout-cone-mode: false + + - name: Install ripgrep + run: command -v rg >/dev/null || (sudo apt-get update -qq && sudo apt-get install -y -qq ripgrep) + + # The body is read straight out of the event payload FILE and written to + # another file. It is never interpolated into a run: block and never placed + # in an environment variable, so shell metacharacters in a hostile PR body + # have nothing to act on. jq is preinstalled on the GitHub-hosted images. + - name: Materialize the untrusted title/body to a file + run: | + set -euo pipefail + mkdir -p "$RUNNER_TEMP/bodyscan" + # An UNRECOGNIZED payload shape must fail, never quietly scan nothing and + # report a pass. If the event schema ever moves, this job must go red + # rather than become a green rubber stamp over an unscanned body. + if [ "$(jq -r 'has("pull_request") or has("issue") or has("comment")' "$GITHUB_EVENT_PATH")" != "true" ]; then + echo "::error title=public-repo-guard (body-guard)::Event payload contains no pull_request/issue/comment object — refusing to report a pass on an unscanned body." + exit 1 + fi + jq -r '[.pull_request.title, .pull_request.body, + .issue.title, .issue.body, + .comment.body] + | map(select(. != null)) | join("\n")' \ + "$GITHUB_EVENT_PATH" > "$RUNNER_TEMP/bodyscan/body.txt" + echo "scanning $(wc -l < "$RUNNER_TEMP/bodyscan/body.txt") line(s) of body text" + + - name: body policy (PR / issue / comment text) + env: + GUARD_PRIVATE_REPOS: ${{ vars.GUARD_PRIVATE_REPOS }} + run: bash scripts/public-repo-guard/body-policy.sh "$RUNNER_TEMP/bodyscan/body.txt" diff --git a/scripts/public-repo-guard/body-policy.sh b/scripts/public-repo-guard/body-policy.sh new file mode 100755 index 0000000..a0b421f --- /dev/null +++ b/scripts/public-repo-guard/body-policy.sh @@ -0,0 +1,139 @@ +#!/usr/bin/env bash +# WAVE public-repo BODY policy — the internal-leak gate for PR/issue/comment text. +# +# Companion to content-policy.sh. That script scans the published working TREE; +# this one scans the other half of a public repo's surface: pull-request titles +# and bodies, issue bodies, and comment bodies. Those are equally world-readable +# and, until this script existed, were scanned by NOTHING server-side. That gap +# was not theoretical — a PR was merged whose wrangler.toml was correctly BLOCKED +# for naming a private repo while the PR body named the same repo, with more +# operational detail attached, and sailed through. +# +# Usage: scripts/public-repo-guard/body-policy.sh +# holds the untrusted text, already materialized to disk. It is passed as +# a PATH and only ever read — the body is never interpolated into a command line +# or an environment variable, so no amount of shell metacharacters in a PR body +# can influence what runs here. +# +# Exit: 0 clean · 1 blocking violation · 2 scanner error (fail closed). +# +# Allowlisting: a line carrying `guard:allow ` is exempt (an accidental +# leak never carries the marker; a deliberate one is visible in a public diff), as +# is any line matching the ABOUT-THE-CONTROL allowlist below. +set -uo pipefail + +FILE="${1:-}" +[[ -n "$FILE" && -f "$FILE" ]] || { echo "::error::body-policy: usage: body-policy.sh "; exit 2; } +command -v rg >/dev/null 2>&1 || { echo "::error::body-policy: ripgrep (rg) required"; exit 2; } + +VIOLATIONS=0 + +# Lines that TALK ABOUT the control rather than leaking through it. Without this, +# the gate blocks its own pull requests and every security discussion — the +# self-referential trap that gets a gate switched off. Ported verbatim in intent +# from the client-side gate's allowlist, which was built for exactly this. +ABOUT_THE_CONTROL='(public-repo-guard|body-policy|content-policy|public-github-write-gate|\bNDA\s+(gate|guard|policy|denylist|sweep|scan|hook)\b|\bno\s+NDA\b|responsib\w*\s+disclos|SECURITY\.md)' + +# check +check() { + local sev="$1" name="$2" re="$3" why="$4" + [[ -z "$re" ]] && { echo "::error::body-policy: internal bug — empty regex for rule '$name'"; exit 2; } + # rg exit: 0=match, 1=no match, >=2=real error → FAIL CLOSED. A gate that passes + # because its scanner broke is worse than no gate: it reports success. + local raw rc + raw="$(rg -nP --no-filename -- "$re" "$FILE" 2>/dev/null)"; rc=$? + if (( rc >= 2 )); then + echo "::error title=public-repo-guard ($name)::ripgrep failed (exit $rc) scanning rule '$name' — failing closed." + exit 2 + fi + # Filter with rg, not grep: BSD/macOS grep has no -P, so a `grep -P` allowlist + # silently errors out locally while working on GNU/CI — the gate would then + # disagree with itself depending on where it ran. rg is already required above. + local matches + matches="$(printf '%s' "$raw" \ + | rg -vN -- 'guard:allow[[:space:]]+[^[:space:]]' \ + | rg -vNiP -- "$ABOUT_THE_CONTROL" || true)" + [[ -z "$matches" ]] && return 0 + local count; count="$(printf '%s\n' "$matches" | grep -c '')" + # Print the LINE NUMBER only — never the matched text. This annotation is itself + # world-readable, so echoing the hit would re-publish the very thing we caught. + echo "::group::[$sev] $name — $why" + printf '%s\n' "$matches" | sed -E 's/^([0-9]+):.*/ line \1: «match redacted — view the body to see it»/' + echo "::endgroup::" + if [[ "$sev" == "BLOCK" ]]; then + echo "::error title=public-repo-guard ($name)::$why — $count occurrence(s) in the title/body. Edit the body to remove it, then re-run." + VIOLATIONS=$((VIOLATIONS+1)) + else + echo "::warning title=public-repo-guard ($name)::$why — $count occurrence(s) (non-blocking; review)." + fi +} + +# --- Credential formats — never legitimate in prose -------------------------- +check BLOCK stripe-live-key '(sk|rk)_live_[A-Za-z0-9]{16,}' 'Live Stripe secret/restricted key' +check BLOCK stripe-account 'acct_[A-Za-z0-9]{16,}' 'Live Stripe account ID — financial infra, never publish' +check BLOCK anthropic-key 'sk-ant-(api|admin)[0-9]{2}-[A-Za-z0-9_-]{20,}' 'Real Anthropic API/admin key' +check BLOCK github-pat 'github_pat_[A-Za-z0-9_]{30,}' 'GitHub fine-grained PAT' +check BLOCK supabase-pat 'sbp_[a-f0-9]{40}' 'Supabase personal access token' +check BLOCK aws-akid 'AKIA[0-9A-Z]{16}' 'AWS access key ID' +check BLOCK private-key '-----BEGIN [A-Z ]*PRIVATE KEY-----' 'Embedded private key material' + +# --- Infrastructure identifiers ---------------------------------------------- +# shellcheck disable=SC2016 # $CLOUDFLARE_ACCOUNT_ID is literal guidance text +check BLOCK cf-account-id 'account_id\s*[:=]\s*["'"'"']?[0-9a-f]{32}' 'Hardcoded Cloudflare account_id — reference the env var instead' +check BLOCK internal-ip '100\.(6[4-9]|[7-9][0-9]|1[01][0-9]|12[0-7])\.[0-9]{1,3}\.[0-9]{1,3}' 'Internal Tailscale-CGNAT IP (100.64.0.0/10) — internal fleet address' +# shellcheck disable=SC2016 # $HOME is literal guidance text +check BLOCK abs-user-path '/(Users|home)/(?!runner/)[a-z][a-z0-9._-]+/' 'Operator absolute home path — leaks identity and local layout' + +# --- Self-identified internal material --------------------------------------- +# USE vs MENTION. A body that SAYS "internal-only" is leaking; a body that QUOTES +# the phrase is describing a policy — including this one. The lookarounds exempt a +# marker wrapped in straight, smart, or backtick quotes. +# +# Not hypothetical: the first run of this job failed on its own pull request, +# because a review bot had edited the PR body to summarize the change and its +# summary quoted the phrase verbatim. The line-level allowlist could not help — +# that line named no gate. Only use-vs-mention separates the two. +# +# A quoted marker is also a trivial bypass, and that is an accepted trade. The +# threat here is the ACCIDENTAL paste; a deliberate evader has easier routes, and +# `guard:allow ` already exists as the honest, visible one. +check BLOCK internal-marker '(?#260"). A gate that fires on all of +# those gets switched off, and then it protects nothing. +# +# So a bare mention stays silent. What fires is a private repo name within ~140 +# characters of INTERNAL OPERATIONAL DETAIL — a SCREAMING_CASE credential NAME, a +# secret-binding verb, a service binding, or a secret COUNT. That is the topology +# of what is wired to what, and it is the shape that actually leaked. +# +# Names are NOT hardcoded (this file is public); CI injects them via the +# GUARD_PRIVATE_REPOS variable. Unset locally → this check is skipped. +if [[ -n "${GUARD_PRIVATE_REPOS:-}" ]]; then + OPS_DETAIL='(?:[A-Z][A-Z0-9]*_(?:SECRET|TOKEN|KEY|PASSWORD)|wrangler\s+secret|secret\s+(?:is\s+)?(?:bound|binding|list)|(?:is\s+)?bound\s+on|service\s+binding|\d{2,}\s+secrets)' + _ALT='' + IFS=', ' read -r -a _PRIV <<< "$GUARD_PRIVATE_REPOS" + for _name in "${_PRIV[@]}"; do + [[ -z "$_name" ]] && continue + # Regex-escape so metacharacters in a name match literally. + _esc="$(printf '%s' "$_name" | sed -E 's/[][(){}.^$*+?|\\]/\\&/g')" + _ALT="${_ALT:+$_ALT|}${_esc}" + done + if [[ -n "$_ALT" ]]; then + # Both orders: name-then-detail and detail-then-name. + check BLOCK private-repo-ops \ + "(?i)\\b(?:${_ALT})\\b[^\\n]{0,140}?\\b${OPS_DETAIL}|${OPS_DETAIL}[^\\n]{0,140}?\\b(?:${_ALT})\\b" \ + 'A private WAVE repo named alongside internal operational detail (credential name, secret binding, or secret count) — the wiring topology is not public' + fi +fi + +if (( VIOLATIONS > 0 )); then + echo "::error::public-repo-guard: $VIOLATIONS blocking body-policy violation(s) — see annotations above." + exit 1 +fi +echo "public-repo-guard: body policy OK" diff --git a/scripts/public-repo-guard/tests/body-policy.test.sh b/scripts/public-repo-guard/tests/body-policy.test.sh new file mode 100755 index 0000000..13cc9bf --- /dev/null +++ b/scripts/public-repo-guard/tests/body-policy.test.sh @@ -0,0 +1,108 @@ +#!/usr/bin/env bash +# Fixture tests for body-policy.sh. +# +# Deliberately fixture-only: the gate is NEVER proved by writing a real leak into a +# live public PR body, because doing so would publish the exact thing it guards. +# +# The negatives here are the load-bearing half. A leak gate that blocks everything +# is trivially "correct" and useless — it gets disabled within a week. The bare +# cross-reference case below is the one that keeps this gate deployable. +set -uo pipefail + +SCRIPT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)/body-policy.sh" +TMP="$(mktemp -d)" +trap 'rm -rf "$TMP"' EXIT + +# The names the real gate is configured with come from an org variable; the tests +# pin their own so they are hermetic and do not depend on CI configuration. +export GUARD_PRIVATE_REPOS="wave-gateway, wave-transports, agent-money" + +PASS=0; FAIL=0 + +# expect +expect() { + local want="$1" name="$2" body="$3" out rc + printf '%s\n' "$body" > "$TMP/body.txt" + out="$(bash "$SCRIPT" "$TMP/body.txt" 2>&1)"; rc=$? + if [[ "$rc" == "$want" ]]; then + PASS=$((PASS+1)); printf ' ok %s\n' "$name" + else + FAIL=$((FAIL+1)); printf ' FAIL %s — want exit %s, got %s\n%s\n' "$name" "$want" "$rc" "$out" + fi + # The annotation is world-readable; a hit must never echo the matched text. + if [[ "$rc" == 1 ]] && printf '%s' "$out" | grep -qF "$body"; then + FAIL=$((FAIL+1)); printf ' FAIL %s — LEAKED the matched text into the annotation\n' "$name" + fi +} + +echo "body-policy fixtures" + +# --- must BLOCK --------------------------------------------------------------- +expect 1 'private repo + credential name' \ + 'Flip is live: WAVE_VIEWPORT_LEASE_SECRET is bound on wave-gateway now.' +expect 1 'private repo + credential name, reverse order' \ + 'The MOQ_JOIN_SECRET was added; wave-transports picks it up on deploy.' +expect 1 'private repo + secret count' \ + 'wave-gateway went from 74 secrets to 75 after this change.' +expect 1 'private repo + service binding' \ + 'This adds a service binding from the worker to agent-money for settlement.' +expect 1 'operator home path' \ + 'Repro: run it from /Users/someoperator/Documents/notes and it fails.' # enforce-ignore (fixture) +expect 1 'internal-only marker' \ + 'Attaching the internal-only rollout plan for context.' +# Assembled at run time rather than written as a literal: a fixture that LOOKS like +# a live AWS key trips this repo's own pre-commit secret scanners (it did, on the +# first draft). Splitting the prefix keeps the fixture exercising the real regex +# without parking a credential-shaped string in source. +AKID_FIXTURE="AKI""A1234567890ABCDEF" +expect 1 'AWS access key id' \ + "The failing job had ${AKID_FIXTURE} configured." +expect 1 'internal tailscale IP' \ + 'It resolves to 100.71.4.19 from inside the fleet.' + +# --- must PASS (precision — these keep the gate deployable) ------------------- +expect 0 'bare private-repo cross-reference' \ + 'This is the companion change to wave-transports#260; merge that one first.' +expect 0 'two private repos, no operational detail' \ + 'Both wave-gateway and wave-transports will need a follow-up for this.' +expect 0 'credential NAME with no private repo nearby' \ + 'The handler now reads SOME_API_TOKEN from the environment instead of a literal.' +expect 0 'public runner path is not an operator path' \ + 'CI checks out to /home/runner/work/repo/repo before the scan runs.' # enforce-ignore (fixture) +expect 0 'talking about the control' \ + 'body-policy blocks a private repo named next to a SECRET_TOKEN; that is intended.' +expect 0 'explicit guard:allow with a reason' \ + 'Example for the docs: wave-gateway holds EXAMPLE_SECRET — guard:allow documented-example' +expect 0 'ordinary clean body' \ + 'Bumps the draft revision and regenerates the fixtures. No behaviour change.' +# Regression: the first CI run of this job failed on its own PR, because a review +# bot edited the body to summarize the change and quoted the marker verbatim. +expect 0 'marker MENTIONED in straight quotes is a description' \ + 'Blocks infra identifiers and markers (account_id, home paths, "internal-only" text).' +expect 0 'marker MENTIONED in a code span' \ + 'The rule matches `internal-only` and `for internal use` in body text.' +expect 0 'marker MENTIONED in smart quotes' \ + 'Blocks operator home paths and “internal-only” text.' +expect 1 'marker USED unquoted still blocks' \ + 'Attaching the internal-only rollout plan; do not share outside the team.' + +# --- fail closed -------------------------------------------------------------- +# Invoked directly, not through expect(): expect() always materializes a file, so +# it cannot reach these paths. A gate that returns "OK" when it was handed nothing +# to scan is the failure mode this whole file exists to prevent. +for case in "no argument at all::" "nonexistent path::$TMP/does-not-exist.txt"; do + name="${case%%::*}"; arg="${case##*::}" + if [[ -n "$arg" ]]; then bash "$SCRIPT" "$arg" >/dev/null 2>&1; else bash "$SCRIPT" >/dev/null 2>&1; fi + rc=$? + if [[ "$rc" == 2 ]]; then + PASS=$((PASS+1)); printf ' ok %s → exit 2 (fails closed)\n' "$name" + else + FAIL=$((FAIL+1)); printf ' FAIL %s — want exit 2, got %s\n' "$name" "$rc" + fi +done + +echo " ---" +if (( FAIL > 0 )); then + echo " $PASS passed, $FAIL FAILED"; exit 1 +fi +echo " $PASS passed, 0 failed" From 823498379f765d0352e4cb1c4b7e72704fe73bbe Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 17:32:37 +0000 Subject: [PATCH 02/11] fix(ci): scope body-policy allowlists and guarantee a PCRE2 ripgrep Three review findings on the body gate: - ABOUT_THE_CONTROL exempted every rule, so a real credential on a line that mentioned the policy scanned clean. The allowlist is now opt-in per rule (prose flag) and only the heuristic rules (internal-marker, private-repo-ops) honour it; credential/infra formats always block. - A global (?i) leaked onto the SCREAMING_CASE credential-name branch of private-repo-ops, so lowercase code talk (session_token) near a repo name blocked ordinary prose. Case-insensitivity is now scoped to the repo names and the English phrase alternatives only. - Both policy scripts require rg -P, but Ubuntu's apt ripgrep is built without PCRE2, which would turn the required check permanently red. Both jobs now install the upstream binary pinned + SHA-256-verified (same pattern as gitleaks), and body-policy.sh refuses a PCRE2-less rg up front with a legible error. Fixtures added for all three regressions; suite passes 26/26. Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 36 ++++++++++++++-- scripts/public-repo-guard/body-policy.sh | 43 ++++++++++++++----- .../tests/body-policy.test.sh | 17 ++++++++ 3 files changed, 81 insertions(+), 15 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index bba2f67..e6be4db 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -85,8 +85,24 @@ jobs: - name: gitleaks (secret scan — published tree) run: gitleaks detect --no-git --source . --config .gitleaks.toml --redact --no-banner --exit-code 1 - - name: Install ripgrep - run: command -v rg >/dev/null || (sudo apt-get update -qq && sudo apt-get install -y -qq ripgrep) + # Both policy scripts are rg -P (PCRE2), and Ubuntu's apt ripgrep is built + # WITHOUT PCRE2 — with it, every rule exits 2 and the required check goes + # permanently red. Install the upstream binary the same way as gitleaks: + # version-pinned and SHA-256-verified. Skipped when a PCRE2-capable rg is + # already on the image. + - name: Install ripgrep (PCRE2-capable, pinned + checksum-verified) + env: + RIPGREP_VERSION: "14.1.1" + RIPGREP_SHA256: "4cf9f2741e6c465ffdb7c26f38056a59e2a2544b51f7cc128ef28337eeae4d8e" + run: | + if command -v rg >/dev/null && rg --pcre2-version >/dev/null 2>&1; then exit 0; fi + curl -fsSL --proto '=https' --tlsv1.2 -o ripgrep.tar.gz \ + "https://github.com/BurntSushi/ripgrep/releases/download/${RIPGREP_VERSION}/ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl.tar.gz" + echo "${RIPGREP_SHA256} ripgrep.tar.gz" | sha256sum -c - + tar -xzf ripgrep.tar.gz "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl/rg" + sudo install -m 0755 "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl/rg" /usr/local/bin/rg + rm -rf ripgrep.tar.gz "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl" + rg --pcre2-version - name: content policy (WAVE trade-secret / internal-leak gate) env: @@ -132,8 +148,20 @@ jobs: sparse-checkout: scripts/public-repo-guard sparse-checkout-cone-mode: false - - name: Install ripgrep - run: command -v rg >/dev/null || (sudo apt-get update -qq && sudo apt-get install -y -qq ripgrep) + # Same PCRE2 requirement as the tree job above — apt's ripgrep won't do. + - name: Install ripgrep (PCRE2-capable, pinned + checksum-verified) + env: + RIPGREP_VERSION: "14.1.1" + RIPGREP_SHA256: "4cf9f2741e6c465ffdb7c26f38056a59e2a2544b51f7cc128ef28337eeae4d8e" + run: | + if command -v rg >/dev/null && rg --pcre2-version >/dev/null 2>&1; then exit 0; fi + curl -fsSL --proto '=https' --tlsv1.2 -o ripgrep.tar.gz \ + "https://github.com/BurntSushi/ripgrep/releases/download/${RIPGREP_VERSION}/ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl.tar.gz" + echo "${RIPGREP_SHA256} ripgrep.tar.gz" | sha256sum -c - + tar -xzf ripgrep.tar.gz "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl/rg" + sudo install -m 0755 "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl/rg" /usr/local/bin/rg + rm -rf ripgrep.tar.gz "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl" + rg --pcre2-version # The body is read straight out of the event payload FILE and written to # another file. It is never interpolated into a run: block and never placed diff --git a/scripts/public-repo-guard/body-policy.sh b/scripts/public-repo-guard/body-policy.sh index a0b421f..aeacbd5 100755 --- a/scripts/public-repo-guard/body-policy.sh +++ b/scripts/public-repo-guard/body-policy.sh @@ -18,13 +18,18 @@ # Exit: 0 clean · 1 blocking violation · 2 scanner error (fail closed). # # Allowlisting: a line carrying `guard:allow ` is exempt (an accidental -# leak never carries the marker; a deliberate one is visible in a public diff), as -# is any line matching the ABOUT-THE-CONTROL allowlist below. +# leak never carries the marker; a deliberate one is visible in a public diff). +# The ABOUT-THE-CONTROL allowlist below additionally exempts the HEURISTIC prose +# rules only — a credential format is a leak no matter what else the line says. set -uo pipefail FILE="${1:-}" [[ -n "$FILE" && -f "$FILE" ]] || { echo "::error::body-policy: usage: body-policy.sh "; exit 2; } command -v rg >/dev/null 2>&1 || { echo "::error::body-policy: ripgrep (rg) required"; exit 2; } +# Every rule here is -P (PCRE2). Distro rg packages are sometimes built without +# it; that build fails every scan with exit 2, so refuse it up front with a +# message that says WHY instead of 21 opaque "ripgrep failed" annotations. +rg --pcre2-version >/dev/null 2>&1 || { echo "::error::body-policy: this ripgrep build lacks PCRE2 (-P) — install a PCRE2-capable rg"; exit 2; } VIOLATIONS=0 @@ -32,11 +37,19 @@ VIOLATIONS=0 # the gate blocks its own pull requests and every security discussion — the # self-referential trap that gets a gate switched off. Ported verbatim in intent # from the client-side gate's allowlist, which was built for exactly this. +# +# SCOPE: only rules that opt in via the `prose` flag honour this allowlist — the +# HEURISTIC rules, whose matches are ordinary words that genuinely occur when +# discussing the gate. The credential-format and infrastructure rules never +# honour it: a real key is a leak even on a line that names SECURITY.md, and +# exempting it there would make "mention the policy" a one-line bypass. ABOUT_THE_CONTROL='(public-repo-guard|body-policy|content-policy|public-github-write-gate|\bNDA\s+(gate|guard|policy|denylist|sweep|scan|hook)\b|\bno\s+NDA\b|responsib\w*\s+disclos|SECURITY\.md)' -# check +# check [prose] +# The optional 5th arg `prose` marks a HEURISTIC rule whose hits may also be +# legitimate discussion of the gate itself; only those honour ABOUT_THE_CONTROL. check() { - local sev="$1" name="$2" re="$3" why="$4" + local sev="$1" name="$2" re="$3" why="$4" scope="${5:-}" [[ -z "$re" ]] && { echo "::error::body-policy: internal bug — empty regex for rule '$name'"; exit 2; } # rg exit: 0=match, 1=no match, >=2=real error → FAIL CLOSED. A gate that passes # because its scanner broke is worse than no gate: it reports success. @@ -51,8 +64,10 @@ check() { # disagree with itself depending on where it ran. rg is already required above. local matches matches="$(printf '%s' "$raw" \ - | rg -vN -- 'guard:allow[[:space:]]+[^[:space:]]' \ - | rg -vNiP -- "$ABOUT_THE_CONTROL" || true)" + | rg -vN -- 'guard:allow[[:space:]]+[^[:space:]]' || true)" + if [[ "$scope" == "prose" ]]; then + matches="$(printf '%s' "$matches" | rg -vNiP -- "$ABOUT_THE_CONTROL" || true)" + fi [[ -z "$matches" ]] && return 0 local count; count="$(printf '%s\n' "$matches" | grep -c '')" # Print the LINE NUMBER only — never the matched text. This annotation is itself @@ -97,7 +112,7 @@ check BLOCK abs-user-path '/(Users|home)/(?!runner/)[a-z][a-z0-9._-]+/' 'O # A quoted marker is also a trivial bypass, and that is an accepted trade. The # threat here is the ACCIDENTAL paste; a deliberate evader has easier routes, and # `guard:allow ` already exists as the honest, visible one. -check BLOCK internal-marker '(? Date: Thu, 6 Aug 2026 17:41:46 +0000 Subject: [PATCH 03/11] ci: scan review bodies, fix install list, use synthetic fixture repo names Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 26 ++++++++++++++----- .../tests/body-policy.test.sh | 26 ++++++++++--------- 2 files changed, 34 insertions(+), 18 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index e6be4db..3de3628 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -13,11 +13,12 @@ name: public-repo-guard # wave-av/.github must not be able to alter another repo's secret scanner). The # gitleaks binary is version-pinned AND SHA-256-verified before it runs. # -# To install on a new repo, copy all four files together: +# To install on a new repo, copy all five files together: # .github/workflows/public-repo-guard.yml # .gitleaks.toml # scripts/public-repo-guard/content-policy.sh # scripts/public-repo-guard/body-policy.sh +# scripts/public-repo-guard/tests/body-policy.test.sh # # Scan scope: the published working TREE (gitleaks --no-git), NOT git history. The # goal is "what is public right now is clean", so a shallow checkout is sufficient. @@ -34,6 +35,13 @@ on: types: [opened, edited] issue_comment: types: [created, edited] + # Review summaries and review-thread comments are just as world-readable as the + # PR body, and `issue_comment` does NOT fire for them — only for top-level PR + # comments. Without these two triggers, review text went unscanned. + pull_request_review: + types: [submitted, edited] + pull_request_review_comment: + types: [created, edited] push: branches: [main, master] workflow_dispatch: @@ -117,8 +125,8 @@ jobs: run: bash scripts/public-repo-guard/tests/body-policy.test.sh # The other half of a public repo's surface. `guard` above scans the published - # TREE; a PR/issue/comment BODY is just as world-readable and, until this job, - # was scanned by nothing server-side. That gap was real, not theoretical: a PR + # TREE; a PR/issue/comment/review BODY is just as world-readable and, until + # this job, was scanned by nothing server-side. That gap was real, not theoretical: a PR # was blocked for naming a private repo in wrangler.toml while the very same # name, with more operational detail attached, sat unchallenged in its body. # @@ -128,7 +136,12 @@ jobs: # can stop that class before publication. body-guard: name: Body content policy - if: github.event_name == 'pull_request' || github.event_name == 'issues' || github.event_name == 'issue_comment' + if: >- + github.event_name == 'pull_request' + || github.event_name == 'issues' + || github.event_name == 'issue_comment' + || github.event_name == 'pull_request_review' + || github.event_name == 'pull_request_review_comment' concurrency: # Keyed on the specific PR / comment / issue rather than github.ref, because # issue events all report the default branch and a ref-keyed group would let @@ -174,12 +187,13 @@ jobs: # An UNRECOGNIZED payload shape must fail, never quietly scan nothing and # report a pass. If the event schema ever moves, this job must go red # rather than become a green rubber stamp over an unscanned body. - if [ "$(jq -r 'has("pull_request") or has("issue") or has("comment")' "$GITHUB_EVENT_PATH")" != "true" ]; then - echo "::error title=public-repo-guard (body-guard)::Event payload contains no pull_request/issue/comment object — refusing to report a pass on an unscanned body." + if [ "$(jq -r 'has("pull_request") or has("issue") or has("comment") or has("review")' "$GITHUB_EVENT_PATH")" != "true" ]; then + echo "::error title=public-repo-guard (body-guard)::Event payload contains no pull_request/issue/comment/review object — refusing to report a pass on an unscanned body." exit 1 fi jq -r '[.pull_request.title, .pull_request.body, .issue.title, .issue.body, + .review.body, .comment.body] | map(select(. != null)) | join("\n")' \ "$GITHUB_EVENT_PATH" > "$RUNNER_TEMP/bodyscan/body.txt" diff --git a/scripts/public-repo-guard/tests/body-policy.test.sh b/scripts/public-repo-guard/tests/body-policy.test.sh index dff36ad..f2d06c2 100755 --- a/scripts/public-repo-guard/tests/body-policy.test.sh +++ b/scripts/public-repo-guard/tests/body-policy.test.sh @@ -14,8 +14,10 @@ TMP="$(mktemp -d)" trap 'rm -rf "$TMP"' EXIT # The names the real gate is configured with come from an org variable; the tests -# pin their own so they are hermetic and do not depend on CI configuration. -export GUARD_PRIVATE_REPOS="wave-gateway, wave-transports, agent-money" +# pin their own so they are hermetic and do not depend on CI configuration. The +# names are deliberately SYNTHETIC: this file is public and exempt from both tree +# scanners, so a real private repo name written here would itself be a leak. +export GUARD_PRIVATE_REPOS="example-private-alpha, example-private-bravo, example-private-charlie" PASS=0; FAIL=0 @@ -39,13 +41,13 @@ echo "body-policy fixtures" # --- must BLOCK --------------------------------------------------------------- expect 1 'private repo + credential name' \ - 'Flip is live: WAVE_VIEWPORT_LEASE_SECRET is bound on wave-gateway now.' + 'Flip is live: WAVE_VIEWPORT_LEASE_SECRET is bound on example-private-alpha now.' expect 1 'private repo + credential name, reverse order' \ - 'The MOQ_JOIN_SECRET was added; wave-transports picks it up on deploy.' + 'The MOQ_JOIN_SECRET was added; example-private-bravo picks it up on deploy.' expect 1 'private repo + secret count' \ - 'wave-gateway went from 74 secrets to 75 after this change.' + 'example-private-alpha went from 74 secrets to 75 after this change.' expect 1 'private repo + service binding' \ - 'This adds a service binding from the worker to agent-money for settlement.' + 'This adds a service binding from the worker to example-private-charlie for settlement.' expect 1 'operator home path' \ 'Repro: run it from /Users/someoperator/Documents/notes and it fails.' # enforce-ignore (fixture) expect 1 'internal-only marker' \ @@ -69,9 +71,9 @@ expect 1 'private key on a line citing SECURITY.md' \ # --- must PASS (precision — these keep the gate deployable) ------------------- expect 0 'bare private-repo cross-reference' \ - 'This is the companion change to wave-transports#260; merge that one first.' + 'This is the companion change to example-private-bravo#260; merge that one first.' expect 0 'two private repos, no operational detail' \ - 'Both wave-gateway and wave-transports will need a follow-up for this.' + 'Both example-private-alpha and example-private-bravo will need a follow-up for this.' expect 0 'credential NAME with no private repo nearby' \ 'The handler now reads SOME_API_TOKEN from the environment instead of a literal.' # Regression: the SCREAMING_CASE credential-name branch is case-SENSITIVE. A @@ -79,17 +81,17 @@ expect 0 'credential NAME with no private repo nearby' \ # (`session_token`) blocked ordinary prose — exactly the false-positive class # that gets a gate switched off. expect 0 'lowercase code identifier near a private repo' \ - 'The wave-gateway worker reads session_token from the request header.' + 'The example-private-alpha worker reads session_token from the request header.' expect 1 'repo name cased differently still pairs with a credential NAME' \ - 'Wave-Gateway now requires LEASE_SECRET at deploy time.' + 'Example-Private-Alpha now requires LEASE_SECRET at deploy time.' expect 0 'talking about the gate with a repo name and credential NAME' \ - 'body-policy blocks wave-gateway next to a SECRET_TOKEN; that is intended.' + 'body-policy blocks example-private-alpha next to a SECRET_TOKEN; that is intended.' expect 0 'public runner path is not an operator path' \ 'CI checks out to /home/runner/work/repo/repo before the scan runs.' # enforce-ignore (fixture) expect 0 'talking about the control' \ 'body-policy blocks a private repo named next to a SECRET_TOKEN; that is intended.' expect 0 'explicit guard:allow with a reason' \ - 'Example for the docs: wave-gateway holds EXAMPLE_SECRET — guard:allow documented-example' + 'Example for the docs: example-private-alpha holds EXAMPLE_SECRET — guard:allow documented-example' expect 0 'ordinary clean body' \ 'Bumps the draft revision and regenerates the fixtures. No behaviour change.' # Regression: the first CI run of this job failed on its own PR, because a review From 45fa1b4486237dd1f4fcf4bf1af678cbe66349f7 Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 17:50:01 +0000 Subject: [PATCH 04/11] fix(ci): run the tree scan on PR edits so a skipped check can't supersede a failing one Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index 3de3628..04ca22f 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -60,10 +60,14 @@ jobs: guard: name: Secrets + content policy # Skips issue/comment events (the tree scan has nothing to say about a comment, - # and the org should not pay for a gitleaks run every time anyone posts one) and - # skips `edited` (a title or body edit does not change the tree). + # and the org should not pay for a gitleaks run every time anyone posts one). + # Deliberately does NOT skip `edited`: a title/body edit doesn't change the + # tree, but a skipped job still publishes a check run for the same head SHA, + # and GitHub treats skipped as passing and evaluates the MOST RECENT check run + # per name — so skipping here would let a description edit supersede a failing + # tree scan with a green rubber stamp. Re-running just reproduces the verdict. if: >- - (github.event_name == 'pull_request' && github.event.action != 'edited') + github.event_name == 'pull_request' || github.event_name == 'push' || github.event_name == 'workflow_dispatch' concurrency: From 9bac94c4af1bb3d0925b30ce492c896de147df46 Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 17:56:35 +0000 Subject: [PATCH 05/11] test: split private-key fixture literal so the foundation-gate secret scan cannot match it Co-authored-by: Codesmith --- scripts/public-repo-guard/tests/body-policy.test.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/public-repo-guard/tests/body-policy.test.sh b/scripts/public-repo-guard/tests/body-policy.test.sh index f2d06c2..cf3ea79 100755 --- a/scripts/public-repo-guard/tests/body-policy.test.sh +++ b/scripts/public-repo-guard/tests/body-policy.test.sh @@ -67,7 +67,7 @@ expect 1 'internal tailscale IP' \ expect 1 'credential on a line that mentions the policy' \ "Discussing public-repo-guard: the key ${AKID_FIXTURE} was rotated." expect 1 'private key on a line citing SECURITY.md' \ - 'Per SECURITY.md: -----BEGIN RSA PRIVATE KEY-----' + "Per SECURITY.md: -----BEGIN RSA ""PRIVATE KEY-----" # pragma: allowlist secret (fixture, split like AKID above) # --- must PASS (precision — these keep the gate deployable) ------------------- expect 0 'bare private-repo cross-reference' \ From ee35c9e4eed21c5294fe1eef8528dae9927549ad Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 18:03:06 +0000 Subject: [PATCH 06/11] fix: drop the stray word boundary so compound credential names match after the repo name Co-authored-by: Codesmith --- scripts/public-repo-guard/body-policy.sh | 7 ++++++- scripts/public-repo-guard/tests/body-policy.test.sh | 5 +++++ 2 files changed, 11 insertions(+), 1 deletion(-) diff --git a/scripts/public-repo-guard/body-policy.sh b/scripts/public-repo-guard/body-policy.sh index aeacbd5..645d8ef 100755 --- a/scripts/public-repo-guard/body-policy.sh +++ b/scripts/public-repo-guard/body-policy.sh @@ -147,8 +147,13 @@ if [[ -n "${GUARD_PRIVATE_REPOS:-}" ]]; then if [[ -n "$_ALT" ]]; then # Both orders: name-then-detail and detail-then-name. Case-insensitivity is # scoped to the repo NAMES only — see the OPS_DETAIL comment above. + # No \b in front of OPS_DETAIL in either branch: for a multi-segment name + # like WAVE_VIEWPORT_LEASE_SECRET the credential alternative can only match + # the trailing LEASE_SECRET, and that position is NOT a word boundary + # (underscore is a word character) — a \b there silently drops every + # compound credential name from the name-first order. check BLOCK private-repo-ops \ - "\\b(?i:${_ALT})\\b[^\\n]{0,140}?\\b${OPS_DETAIL}|${OPS_DETAIL}[^\\n]{0,140}?\\b(?i:${_ALT})\\b" \ + "\\b(?i:${_ALT})\\b[^\\n]{0,140}?${OPS_DETAIL}|${OPS_DETAIL}[^\\n]{0,140}?\\b(?i:${_ALT})\\b" \ 'A private WAVE repo named alongside internal operational detail (credential name, secret binding, or secret count) — the wiring topology is not public' prose fi fi diff --git a/scripts/public-repo-guard/tests/body-policy.test.sh b/scripts/public-repo-guard/tests/body-policy.test.sh index cf3ea79..ccd116e 100755 --- a/scripts/public-repo-guard/tests/body-policy.test.sh +++ b/scripts/public-repo-guard/tests/body-policy.test.sh @@ -44,6 +44,11 @@ expect 1 'private repo + credential name' \ 'Flip is live: WAVE_VIEWPORT_LEASE_SECRET is bound on example-private-alpha now.' expect 1 'private repo + credential name, reverse order' \ 'The MOQ_JOIN_SECRET was added; example-private-bravo picks it up on deploy.' +# Regression: a MULTI-segment credential name after the repo name. The rule can +# only match the trailing LEASE_SECRET, which sits mid-word (after an +# underscore) — a \b in front of OPS_DETAIL silently dropped this whole order. +expect 1 'private repo, then compound credential name' \ + 'example-private-alpha now uses WAVE_VIEWPORT_LEASE_SECRET for renewals.' expect 1 'private repo + secret count' \ 'example-private-alpha went from 74 secrets to 75 after this change.' expect 1 'private repo + service binding' \ From e3a4265af7a36593dc77a1060e014663e99291c8 Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 18:09:24 +0000 Subject: [PATCH 07/11] docs: state precisely what per-job concurrency does and does not fix for the tree job Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index 04ca22f..1b2b5af 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -52,9 +52,14 @@ permissions: contents: read # Concurrency is per JOB, not per workflow: the two jobs want opposite behaviour. -# A workflow-level group would force one policy on both, and it showed: rapid body -# edits cancelled the tree job over and over, and every cancelled check-run stays -# attached to the commit, so the PR reported UNSTABLE while the live runs were green. +# A workflow-level group would force one policy on both, and it showed: comment and +# issue events (which the tree job skips entirely) cancelled in-flight tree scans, +# so a chatty review thread could leave a head SHA with NO completed tree verdict. +# Per-job groups end that: only same-PR pull_request events enter the tree group. +# Within that group, a body edit or new push still supersedes an in-flight tree run +# — deliberately. The replacement scans the same (or newer) head SHA and always +# lands a completed verdict, and GitHub evaluates the most recent check run per +# name, so the superseded run's cancelled record is cosmetic, not load-bearing. jobs: guard: From 5d9db11cc9a0acf11076901e3e01978debc38433 Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 18:15:41 +0000 Subject: [PATCH 08/11] fix: key the body-guard group on the comment/review id first so batched review comments are all scanned Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index 1b2b5af..94f62de 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -152,14 +152,21 @@ jobs: || github.event_name == 'pull_request_review' || github.event_name == 'pull_request_review_comment' concurrency: - # Keyed on the specific PR / comment / issue rather than github.ref, because - # issue events all report the default branch and a ref-keyed group would let - # two comments cancel each other, leaving one unscanned. + # Keyed on the MOST SPECIFIC object in the payload, and the comment/review id + # must come FIRST: `pull_request_review` and `pull_request_review_comment` + # payloads carry a top-level `pull_request` object, so a PR-number-first chain + # would short-circuit and file every review comment on a PR under ONE group. + # GitHub keeps at most one PENDING run per group, so a review submitted with + # several inline comments would silently drop the middle ones — distinct + # comments are distinct world-readable texts, and each needs its own verdict. + # A ref-keyed group has the same flaw (issue events all report the default + # branch). Only same-object versions may share a group: the newest scan of an + # edited body covers what is public now. # # cancel-in-progress is deliberately FALSE. Every version of a body deserves a # verdict, the job is seconds long, and a cancelled check-run lingers on the # commit and makes an otherwise-green PR look broken. - group: public-repo-guard-body-${{ github.event.pull_request.number || github.event.comment.id || github.event.issue.number || github.ref }} + group: public-repo-guard-body-${{ github.event.comment.id || github.event.review.id || github.event.pull_request.number || github.event.issue.number || github.ref }} cancel-in-progress: false runs-on: ubuntu-latest steps: From 7e086ee900f51235e2fcebfbadaf087b1966b030 Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 18:27:21 +0000 Subject: [PATCH 09/11] fix(ci): close two body-guard pass-by-default paths Review events (pull_request_review, pull_request_review_comment) run in the PR's context and publish a check run on the PR head SHA; with the guard job skipping them, that run lands as 'skipped', which GitHub treats as passing while evaluating the most recent check run per name, so any review comment could supersede a failing tree scan with a green rubber stamp. The guard job now re-runs the tree scan on those events, same as pull_request: edited. body-policy.sh now fails closed (exit 2) in CI when GUARD_PRIVATE_REPOS is empty or contains no names: a missing or renamed org variable must go red, not silently skip the private-repo proximity rule and report a pass over an unscanned leak class. Local runs still skip the rule, and both behaviours are pinned by new fixtures. Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 18 ++++++---- scripts/public-repo-guard/body-policy.sh | 34 +++++++++++-------- .../tests/body-policy.test.sh | 19 +++++++++++ 3 files changed, 51 insertions(+), 20 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index 94f62de..a4aeb05 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -55,7 +55,8 @@ permissions: # A workflow-level group would force one policy on both, and it showed: comment and # issue events (which the tree job skips entirely) cancelled in-flight tree scans, # so a chatty review thread could leave a head SHA with NO completed tree verdict. -# Per-job groups end that: only same-PR pull_request events enter the tree group. +# Per-job groups end that: only same-PR pull_request and review events enter the +# tree group. # Within that group, a body edit or new push still supersedes an in-flight tree run # — deliberately. The replacement scans the same (or newer) head SHA and always # lands a completed verdict, and GitHub evaluates the most recent check run per @@ -66,13 +67,18 @@ jobs: name: Secrets + content policy # Skips issue/comment events (the tree scan has nothing to say about a comment, # and the org should not pay for a gitleaks run every time anyone posts one). - # Deliberately does NOT skip `edited`: a title/body edit doesn't change the - # tree, but a skipped job still publishes a check run for the same head SHA, - # and GitHub treats skipped as passing and evaluates the MOST RECENT check run - # per name — so skipping here would let a description edit supersede a failing - # tree scan with a green rubber stamp. Re-running just reproduces the verdict. + # That is safe ONLY because those runs report the DEFAULT branch, not the PR + # head. Review events are different: they run in the PR's context and publish + # a check run on the PR HEAD SHA, and a job skipped by `if` still publishes + # one with conclusion `skipped`, which GitHub treats as passing while + # evaluating the MOST RECENT check run per name. Skipping them would let any + # review comment supersede a failing tree scan with a green rubber stamp, so + # they re-run the tree scan instead (same reason `edited` is not skipped: + # re-running just reproduces the verdict). if: >- github.event_name == 'pull_request' + || github.event_name == 'pull_request_review' + || github.event_name == 'pull_request_review_comment' || github.event_name == 'push' || github.event_name == 'workflow_dispatch' concurrency: diff --git a/scripts/public-repo-guard/body-policy.sh b/scripts/public-repo-guard/body-policy.sh index 645d8ef..282c535 100755 --- a/scripts/public-repo-guard/body-policy.sh +++ b/scripts/public-repo-guard/body-policy.sh @@ -128,7 +128,11 @@ check BLOCK internal-marker '(? 0 )); then diff --git a/scripts/public-repo-guard/tests/body-policy.test.sh b/scripts/public-repo-guard/tests/body-policy.test.sh index ccd116e..f76e8f7 100755 --- a/scripts/public-repo-guard/tests/body-policy.test.sh +++ b/scripts/public-repo-guard/tests/body-policy.test.sh @@ -125,6 +125,25 @@ for case in "no argument at all::" "nonexistent path::$TMP/does-not-exist.txt"; fi done +# An empty GUARD_PRIVATE_REPOS is a documented local convenience but FAILS CLOSED +# in CI (GITHUB_ACTIONS set): a missing or renamed org variable must go red, never +# silently skip the private-repo rule and report a pass over an unscanned class. +printf '%s\n' 'Ordinary clean body with nothing to find.' > "$TMP/body.txt" +env -u GUARD_PRIVATE_REPOS GITHUB_ACTIONS=true bash "$SCRIPT" "$TMP/body.txt" >/dev/null 2>&1 +rc=$? +if [[ "$rc" == 2 ]]; then + PASS=$((PASS+1)); printf ' ok %s → exit 2 (fails closed)\n' 'GUARD_PRIVATE_REPOS unset in CI' +else + FAIL=$((FAIL+1)); printf ' FAIL %s: want exit 2, got %s\n' 'GUARD_PRIVATE_REPOS unset in CI' "$rc" +fi +env -u GUARD_PRIVATE_REPOS -u GITHUB_ACTIONS bash "$SCRIPT" "$TMP/body.txt" >/dev/null 2>&1 +rc=$? +if [[ "$rc" == 0 ]]; then + PASS=$((PASS+1)); printf ' ok %s → exit 0 (rule skipped)\n' 'GUARD_PRIVATE_REPOS unset locally' +else + FAIL=$((FAIL+1)); printf ' FAIL %s: want exit 0, got %s\n' 'GUARD_PRIVATE_REPOS unset locally' "$rc" +fi + echo " ---" if (( FAIL > 0 )); then echo " $PASS passed, $FAIL FAILED"; exit 1 From f2f50d7e566270aa8fd3a83fe915fe11c17e6c63 Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 18:36:27 +0000 Subject: [PATCH 10/11] ci: skip tree scan for review events on closed PRs Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index a4aeb05..e7d29af 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -75,10 +75,15 @@ jobs: # review comment supersede a failing tree scan with a green rubber stamp, so # they re-run the tree scan instead (same reason `edited` is not skipped: # re-running just reproduces the verdict). + # Review events on a CLOSED PR are the one exception: nothing gates a closed + # PR, and refs/pull/N/merge goes stale after close, so checking it out can + # fail and paint a spurious red on the required check. A skipped run is + # harmless exactly there: a rubber stamp with nothing left to stamp. if: >- github.event_name == 'pull_request' - || github.event_name == 'pull_request_review' - || github.event_name == 'pull_request_review_comment' + || ((github.event_name == 'pull_request_review' + || github.event_name == 'pull_request_review_comment') + && github.event.pull_request.state == 'open') || github.event_name == 'push' || github.event_name == 'workflow_dispatch' concurrency: From c284c89bc145caec2b206696d696411926f9f0fc Mon Sep 17 00:00:00 2001 From: yakimoto Date: Thu, 6 Aug 2026 18:42:45 +0000 Subject: [PATCH 11/11] fix(ci): skip body scan for review events on closed PRs Co-authored-by: Codesmith --- .github/workflows/public-repo-guard.yml | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index e7d29af..5d5ab59 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -156,12 +156,21 @@ jobs: # can stop that class before publication. body-guard: name: Body content policy + # Review events on a CLOSED PR are excluded for the same reason as in `guard` + # above: refs/pull/N/merge goes stale after close, so the checkout can fail + # and paint a spurious red on the PR. The cost is a detection gap — a review + # comment on a closed PR goes unscanned server-side — accepted because a + # guaranteed-flaky red is worse than no verdict, and the pre-write hook is + # the control that stops that class before publication anyway. (Comments on + # closed PRs still arrive as `issue_comment` and ARE scanned; only review + # summaries and review-thread comments are affected.) if: >- github.event_name == 'pull_request' || github.event_name == 'issues' || github.event_name == 'issue_comment' - || github.event_name == 'pull_request_review' - || github.event_name == 'pull_request_review_comment' + || ((github.event_name == 'pull_request_review' + || github.event_name == 'pull_request_review_comment') + && github.event.pull_request.state == 'open') concurrency: # Keyed on the MOST SPECIFIC object in the payload, and the comment/review id # must come FIRST: `pull_request_review` and `pull_request_review_comment`