diff --git a/.github/workflows/public-repo-guard.yml b/.github/workflows/public-repo-guard.yml index 719718a..977f731 100644 --- a/.github/workflows/public-repo-guard.yml +++ b/.github/workflows/public-repo-guard.yml @@ -13,10 +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 three 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. @@ -25,24 +27,53 @@ 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-${{ github.ref }} - cancel-in-progress: true +# Concurrency is per JOB, not per workflow: the two jobs want opposite behaviour. +# A workflow-level group would force one policy on both, and it showed: rapid body +# edits cancelled the tree job over and over, and every cancelled check-run stays +# attached to the commit, so the PR reported UNSTABLE while the live runs were green. jobs: guard: name: Secrets + content policy + # Skips issue/comment events (the tree scan has nothing to say about a comment, + # and the org should not pay for a gitleaks run every time anyone posts one). + # + # Deliberately does NOT skip `edited`, even though a title/body edit cannot + # change the tree. This workflow subscribes to `edited` for the body gate, and + # a job-level skip still emits a check run under this job's name — the ONE + # required check — with conclusion `skipped` on the PR head SHA. A skipped + # required check satisfies branch protection, so skipping here would let a + # body edit re-report a FAILED tree scan as satisfied on the same commit, + # without a re-scan. The redundant scan is the price of an honest verdict. + # (Issue/comment skips are safe: those check runs attach to the default + # branch tip, never to a PR head, so no merge gate ever reads them.) + if: >- + github.event_name == 'pull_request' + || github.event_name == 'push' + || github.event_name == 'workflow_dispatch' + concurrency: + group: public-repo-guard-tree-${{ 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 @@ -64,10 +95,181 @@ 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) + - name: Install ripgrep (PCRE2 build required) + env: + RIPGREP_VERSION: "14.1.1" + RIPGREP_SHA256: "4cf9f2741e6c465ffdb7c26f38056a59e2a2544b51f7cc128ef28337eeae4d8e" + run: | + # Both policy scripts depend on `rg -P`. A ripgrep built without PCRE2 + # (some distro packages) makes every scan exit 2 — fail closed, but a + # runner-image change shipping such a build would leave the gate + # permanently red with no remediation. Use the preinstalled rg when it + # has PCRE2; otherwise fall back to the upstream release binary (always + # PCRE2-enabled), pinned and checksum-verified like gitleaks above. + if ! command -v rg >/dev/null || ! rg --pcre2-version >/dev/null 2>&1; then + 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 "${RIPGREP_SHA256} ripgrep.tar.gz" | sha256sum -c - + tar -xzf ripgrep.tar.gz --strip-components=1 "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl/rg" + sudo install -m 0755 rg /usr/local/bin/rg + rm -f rg ripgrep.tar.gz + # Drop the cached path of the preinstalled rg (hashed by the probe + # above), or the check below would re-run the old PCRE2-less binary + # and fail the step right after a successful install. + hash -r + fi + rg --pcre2-version - name: content policy (WAVE trade-secret / internal-leak gate) 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: + # The gate must run from TRUSTED history. The default checkout on a + # pull_request event is the PR merge ref, so the PR's own copy of + # body-policy.sh would be the one that runs — a fork PR could edit it to + # exit 0 and wave its own body through. Unlike the tree scan (which must + # look at PR content), the body gate has no reason to execute anything + # PR-controlled: pin to the base branch tip. On issues/issue_comment + # events there is no PR ref and github.ref is already the default branch. + # + # Known limit, considered and accepted: on pull_request the workflow FILE + # itself comes from the PR head, so this trusted-ref design holds only for + # PRs that do not edit this workflow. That edit is loudly visible in the + # PR's own diff, and the same move neuters the REQUIRED tree check too + # (which must run PR content, so pull_request_target cannot protect it) — + # the class is closed org-side, by requiring review on workflow changes, + # not by a per-job trigger swap. Swapping this job to pull_request_target + # would even WEAKEN the gate: its run would attach a skipped + # "Secrets + content policy" check to the head SHA, and a skipped + # required check satisfies branch protection — a workflow-deleting PR, + # unmergeable today ("Expected — waiting"), would become mergeable. + ref: ${{ github.event.pull_request.base.sha || github.ref }} + # Only the gate's own scripts are needed — no reason to pay for the whole + # tree on every comment. The workflow file rides along solely so the + # bootstrap check below can tell "the body gate is not installed on this + # ref yet" apart from "installed, but its script went missing" (which + # must fail) — by looking at whether the trusted-ref copy of this + # workflow references body-policy.sh at all. + sparse-checkout: | + scripts/public-repo-guard + .github/workflows/public-repo-guard.yml + sparse-checkout-cone-mode: false + + - name: Install ripgrep (PCRE2 build required) + env: + RIPGREP_VERSION: "14.1.1" + RIPGREP_SHA256: "4cf9f2741e6c465ffdb7c26f38056a59e2a2544b51f7cc128ef28337eeae4d8e" + run: | + # body-policy.sh depends on `rg -P`. Same fallback as the tree job's + # step: a PCRE2-less preinstalled rg must not leave the body gate + # permanently red — install the pinned, checksum-verified upstream + # release binary (always PCRE2-enabled) instead. + if ! command -v rg >/dev/null || ! rg --pcre2-version >/dev/null 2>&1; then + curl -fsSL --proto '=https' --tlsv1.2 -o ripgrep.tar.gz \ + "https://github.com/BurntSushi/ripgrep/releases/download/${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 --strip-components=1 "ripgrep-${RIPGREP_VERSION}-x86_64-unknown-linux-musl/rg" + sudo install -m 0755 rg /usr/local/bin/rg + rm -f rg ripgrep.tar.gz + # Drop the cached path of the preinstalled rg (hashed by the probe + # above), or the check below would re-run the old PCRE2-less binary + # and fail the step right after a successful install. + hash -r + fi + rg --pcre2-version + + # The body is read straight out of the event payload FILE and written to + # another file. It is never interpolated into a run: block and never placed + # in an environment variable, so shell metacharacters in a hostile PR body + # have nothing to act on. jq is preinstalled on the GitHub-hosted images. + - name: Materialize the untrusted title/body to a file + run: | + set -euo pipefail + mkdir -p "$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 + # Scan ONLY the object this event is about. An issue_comment payload also + # carries the parent issue's title/body, but re-scanning those on every + # comment would let one old violation in the issue body fail every future + # comment on it, with line numbers into a concatenated buffer that maps + # to nothing. The issue body already got its own verdict on its own + # `issues` event; each text gets judged exactly once, when it changes. + jq -r 'if has("comment") then [.comment.body] + elif has("pull_request") then [.pull_request.title, .pull_request.body] + else [.issue.title, .issue.body] end + | 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: ${{ vars.GUARD_PRIVATE_REPOS }} + run: | + # BOOTSTRAP: the gate runs from the TRUSTED ref (see the checkout above), + # so on the very PR that first installs it the base branch has no script + # yet. But "script missing" alone must never mean "pass" — a rename, a + # sparse-checkout drift, or a partial install would turn the gate into a + # permanent green rubber stamp. Disambiguate with the workflow file on + # the SAME trusted ref: if that copy never mentions body-policy.sh (the + # file is absent, or it is an OLDER tree-only version of this gate — + # exactly what a repo upgrading from the 27-repo shape has on its base + # branch), the body gate simply is not installed there yet (bootstrap), + # skip with a visible warning. If the trusted-ref workflow DOES + # reference the script and the script is missing → broken install; + # fail closed like every other error path. + if [ ! -f scripts/public-repo-guard/body-policy.sh ]; then + if ! grep -q 'scripts/public-repo-guard/body-policy\.sh' .github/workflows/public-repo-guard.yml 2>/dev/null; then + echo "::warning title=public-repo-guard (body-guard)::body-policy.sh is not on the trusted ref yet (bootstrap PR installing or upgrading the gate) — body scan skipped; it activates once this merges." + exit 0 + fi + echo "::error title=public-repo-guard (body-guard)::the trusted ref's workflow references scripts/public-repo-guard/body-policy.sh but the script is NOT there — broken install (rename, sparse-checkout drift, or partial copy). Refusing to report a pass on an unscanned body." + exit 1 + fi + # An empty GUARD_PRIVATE_REPOS silently disables the private-repo + # proximity rule (the script deliberately skips it, for local runs). + # In CI that silence is a hazard — fork-triggered runs may not receive + # org variables — so make "off" visible instead of quietly green. + if [ -z "${GUARD_PRIVATE_REPOS:-}" ]; then + echo "::warning title=public-repo-guard (body-guard)::GUARD_PRIVATE_REPOS is empty for this run — the private-repo proximity rule is DISABLED. If this is a fork PR, org variables may not be exposed to it." + fi + bash scripts/public-repo-guard/body-policy.sh "$RUNNER_TEMP/bodyscan/body.txt" diff --git a/scripts/public-repo-guard/body-policy.sh b/scripts/public-repo-guard/body-policy.sh new file mode 100755 index 0000000..f8afe10 --- /dev/null +++ b/scripts/public-repo-guard/body-policy.sh @@ -0,0 +1,172 @@ +#!/usr/bin/env bash +# WAVE public-repo BODY policy — the internal-leak gate for PR/issue/comment text. +# +# Companion to content-policy.sh. That script scans the published working TREE; +# this one scans the other half of a public repo's surface: pull-request titles +# and bodies, issue bodies, and comment bodies. Those are equally world-readable +# and, until this script existed, were scanned by NOTHING server-side. That gap +# was not theoretical — a PR was merged whose wrangler.toml was correctly BLOCKED +# for naming a private repo while the PR body named the same repo, with more +# operational detail attached, and sailed through. +# +# Usage: scripts/public-repo-guard/body-policy.sh +# holds the untrusted text, already materialized to disk. It is passed as +# a PATH and only ever read — the body is never interpolated into a command line +# or an environment variable, so no amount of shell metacharacters in a PR body +# can influence what runs here. +# +# Exit: 0 clean · 1 blocking violation · 2 scanner error (fail closed). +# +# Allowlisting: a line carrying `guard:allow ` is exempt (an accidental +# leak never carries the marker; a deliberate one is visible in a public diff), as +# is any line matching the ABOUT-THE-CONTROL allowlist below — except for the +# credential-format rules, which honour only `guard:allow`: naming the gate must +# never launder a live key past it. +set -uo pipefail + +FILE="${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 below uses `rg -P`. A ripgrep built without PCRE2 (some distro +# packages) exits 2 on the first rule, which fails closed but reads as "scanner +# broke" with no hint why. Name the real cause up front — still exit 2 either way. +printf 'x' | rg -qP 'x' 2>/dev/null || { echo "::error::body-policy: this ripgrep build lacks PCRE2 (-P) support — install a PCRE2-enabled rg (the upstream release binaries are). Failing closed."; 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 [exempt] +# exempt defaults to "about-exempt": lines matching ABOUT_THE_CONTROL are +# skipped. Credential-format rules pass "no-about-exempt" — a line that names +# the gate can still carry a live key, and talking ABOUT the control must never +# launder a credential PAST it. Only `guard:allow ` (explicit, visible +# in the public body) exempts those. +check() { + local sev="$1" name="$2" re="$3" why="$4" exempt="${5:-about-exempt}" + [[ -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:]]' || true)" + if [[ "$exempt" == "about-exempt" ]]; then + matches="$(printf '%s' "$matches" \ + | rg -vNiP -- "$ABOUT_THE_CONTROL" || true)" + fi + [[ -z "$matches" ]] && return 0 + local count; count="$(printf '%s\n' "$matches" | grep -c '')" + # Print the LINE NUMBER only — never the matched text. This annotation is itself + # world-readable, so echoing the hit would re-publish the very thing we caught. + echo "::group::[$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 -------------------------- +# no-about-exempt: a credential-shaped string is a credential no matter what else +# its line says — mentioning "body-policy" next to a live key must not pass it. +check BLOCK stripe-live-key '(sk|rk)_live_[A-Za-z0-9]{16,}' 'Live Stripe secret/restricted key' no-about-exempt +check BLOCK stripe-account 'acct_[A-Za-z0-9]{16,}' 'Live Stripe account ID — financial infra, never publish' no-about-exempt +check BLOCK anthropic-key 'sk-ant-(api|admin)[0-9]{2}-[A-Za-z0-9_-]{20,}' 'Real Anthropic API/admin key' no-about-exempt +check BLOCK github-pat 'github_pat_[A-Za-z0-9_]{30,}' 'GitHub fine-grained PAT' no-about-exempt +check BLOCK supabase-pat 'sbp_[a-f0-9]{40}' 'Supabase personal access token' no-about-exempt +check BLOCK aws-akid 'AKIA[0-9A-Z]{16}' 'AWS access key ID' no-about-exempt +check BLOCK private-key '-----BEGIN [A-Z ]*PRIVATE KEY-----' 'Embedded private key material' no-about-exempt + +# --- Infrastructure identifiers ---------------------------------------------- +# no-about-exempt for the same reason as the credential formats: a REAL internal +# IP or operator path on a line that happens to name the gate is still a leak. +# A discussion of these rules that needs a live example has `guard:allow`. +# 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' no-about-exempt +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' no-about-exempt +# shellcheck disable=SC2016 # $HOME is literal guidance text +check BLOCK abs-user-path '/(Users|home)/(?!runner/)[A-Za-z][A-Za-z0-9._-]+/' 'Operator absolute home path — leaks identity and local layout' no-about-exempt + +# --- Self-identified internal material --------------------------------------- +# USE vs MENTION. A body that SAYS "internal-only" is leaking; a body that QUOTES +# the phrase is describing a policy — including this one. The lookarounds exempt a +# marker wrapped in straight, smart, or backtick quotes. +# +# Not hypothetical: the first run of this job failed on its own pull request, +# because a review bot had edited the PR body to summarize the change and its +# summary quoted the phrase verbatim. The line-level allowlist could not help — +# that line named no gate. Only use-vs-mention separates the two. +# +# A quoted marker is also a trivial bypass, and that is an accepted trade. The +# threat here is the ACCIDENTAL paste; a deliberate evader has easier routes, and +# `guard:allow ` already exists as the honest, visible one. +# +# (?i): confidentiality banners are MORE often shouted than whispered — +# "INTERNAL ONLY", "DO NOT SHARE" — and a case-sensitive rule published exactly +# the loud form while blocking the quiet one. +check BLOCK internal-marker '(?i)(?#260"). A gate that fires on all of +# those gets switched off, and then it protects nothing. +# +# So a bare mention stays silent. What fires is a private repo name within ~140 +# characters of INTERNAL OPERATIONAL DETAIL — a SCREAMING_CASE credential NAME, a +# secret-binding verb, a service binding, or a secret COUNT. That is the topology +# of what is wired to what, and it is the shape that actually leaked. +# +# Names are NOT hardcoded (this file is public); CI injects them via the +# GUARD_PRIVATE_REPOS variable. Unset locally → this check is skipped. +if [[ -n "${GUARD_PRIVATE_REPOS:-}" ]]; then + # The credential-name branch is wrapped in (?-i:…): the composed pattern below + # opens with a global (?i) so REPO NAMES match case-insensitively, but the + # SCREAMING_CASE requirement is the whole point of this branch — without the + # override, everyday lowercase identifiers like `api_key` or `retry_token` + # near a repo name would read as operational detail and block clean PRs. + # The leading segment allows underscores ([A-Z][A-Z0-9_]*): a multi-segment name + # like EXAMPLE_LEASE_SECRET must match from its FIRST character. Without them the + # only viable sub-match (LEASE_SECRET) starts right after an underscore — a word + # character — so the \b the composed pattern requires before OPS_DETAIL can never + # sit there, and the name-then-detail order silently missed every such name. + OPS_DETAIL='(?:(?-i:[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: $VIOLATIONS blocking body-policy violation(s) — see annotations above." + exit 1 +fi +echo "public-repo-guard: body policy OK" diff --git a/scripts/public-repo-guard/tests/body-policy.test.sh b/scripts/public-repo-guard/tests/body-policy.test.sh new file mode 100755 index 0000000..4bbf699 --- /dev/null +++ b/scripts/public-repo-guard/tests/body-policy.test.sh @@ -0,0 +1,144 @@ +#!/usr/bin/env bash +# Fixture tests for body-policy.sh. +# +# Deliberately fixture-only: the gate is NEVER proved by writing a real leak into a +# live public PR body, because doing so would publish the exact thing it guards. +# +# The negatives here are the load-bearing half. A leak gate that blocks everything +# is trivially "correct" and useless — it gets disabled within a week. The bare +# cross-reference case below is the one that keeps this gate deployable. +set -uo pipefail + +SCRIPT="$(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. +# Deliberately fictional (this file is public and exempt from the tree gate): +# any names exercise the same regexes, so no real private repo name belongs here. +export GUARD_PRIVATE_REPOS="example-private-one, example-private-two, example-private-three" + +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: EXAMPLE_LEASE_SECRET is bound on example-private-one now.' +expect 1 'private repo + credential name, reverse order' \ + 'The EXAMPLE_JOIN_SECRET was added; example-private-two picks it up on deploy.' +# Regression: a multi-segment name after the repo name. The credential branch once +# barred underscores from its leading segment, so the only sub-match started right +# after an underscore — where the \b the pattern requires can never sit — and the +# name-then-detail order missed every multi-segment credential name. +expect 1 'private repo then multi-segment credential name' \ + 'example-private-one now reads EXAMPLE_LEASE_SECRET at boot.' +expect 1 'private repo + secret count' \ + 'example-private-one went from 74 secrets to 75 after this change.' +expect 1 'private repo + service binding' \ + 'This adds a service binding from the worker to example-private-three 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.' +# Regression: banners are more often SHOUTED than whispered — a case-sensitive +# rule published "INTERNAL-ONLY" while blocking "internal-only". +expect 1 'internal marker in caps' \ + 'Attaching the INTERNAL-ONLY rollout plan for context.' +expect 1 'do-not-share marker in caps' \ + 'DO NOT SHARE outside the team.' +# Regression: /Users/Alice is as common on macOS as /Users/alice. +expect 1 'operator home path, capitalised username' \ + 'Repro: run it from /Users/SomeOperator/Documents/notes and it fails.' # enforce-ignore (fixture) +# 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." +# Regression: the ABOUT-THE-CONTROL allowlist must not launder credentials — a +# line that names the gate by name can still carry a live key, and must block. +expect 1 'credential on a line naming the gate still blocks' \ + "body-policy flagged ${AKID_FIXTURE} here and public-repo-guard was right to." +expect 1 'internal IP on a line naming the gate still blocks' \ + 'content-policy missed 100.71.4.19 in the fleet notes.' +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 example-private-two#260; merge that one first.' +expect 0 'two private repos, no operational detail' \ + 'Both example-private-one and example-private-two 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.' +# Regression: the global (?i) that lets repo NAMES match case-insensitively must +# not leak into the SCREAMING_CASE credential-name branch — everyday lowercase +# identifiers near a private repo are not operational topology. +expect 0 'lowercase identifier near a private repo is not ops detail' \ + 'We updated example-private-one to read the api_key from config now.' +expect 0 'another lowercase identifier near a private repo' \ + 'Fix example-private-two: the retry_token handling was wrong.' +# Multi-segment lowercase stays clean too: underscores in the leading segment must +# widen only the SCREAMING_CASE branch, never re-admit everyday identifiers. +expect 0 'multi-segment lowercase identifier near a private repo' \ + 'example-private-one now reads the lease_rotation_key from config.' +expect 0 'explicit guard:allow with a reason' \ + 'Example for the docs: example-private-one 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 0 'caps marker MENTIONED in quotes is still a description' \ + 'The rule now also catches "INTERNAL ONLY" and "DO NOT SHARE" banners.' +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"