Files
Kevin Codex 530563a0bb feat(npm): publish as @twigpine/zero (#1142)
* feat(npm): publish as @twigpine/zero

The npm package moves from @gitlawb/zero to @twigpine/zero, and everything
that finds Zero's GitHub repository by name now names Twigpine/zero.

Package name
- package.json, the lockfile and release-please use @twigpine/zero.
  `repository` names Twigpine/zero: npm checks provenance against it, so a
  publish from this repo with the old URL fails with E422. The homepage is
  zero.twigpine.com.
- The name lives in package.json only. build-platform-packages.mjs names the
  wrapper, platform versions, alias names and alias targets from it, and
  bin/zero.js reads its own package.json for the alias it resolves and for
  every reinstall hint. The Go constant npmPackageName is held equal to
  package.json by a test.

Repository defaults
- The update check (DefaultRepository), install.sh, install.ps1, the npm
  wrapper's fallback downloader, and the GitHub Action's zero-repo default
  all used Gitlawb/zero. That name now belongs to an unrelated GitHub org, and
  these paths download and run what they find there, so they point at
  Twigpine/zero. Docs, README and issue-template links follow. The Go module
  path (and so `go install github.com/Gitlawb/zero/...`) is unchanged here.

Publishing
- The two inline publish steps become scripts/npm/publish-packages.sh, which
  reads the name from the assembled payloads, refuses to start when the
  wrapper's aliases do not match the platform payloads, skips versions
  already on the registry (so re-running a half-finished job is safe), and
  publishes with --ignore-scripts. The dist-tag rules are unchanged.
  internal/npmwrapper/publish_test.go runs it against a fake registry.

Docs explain the switch for existing installs: uninstall @gitlawb/zero first,
because both packages own the `zero` bin and npm refuses the second (EEXIST).

* feat(npm): publish 0.10.0 once more as @gitlawb/zero to point users at the new name

Deprecating @gitlawb/zero alone would strand every existing npm install.
0.9.0's `zero upgrade` runs `npm install -g @gitlawb/zero@latest`; with
nothing newer under that name npm installs nothing, prints no deprecation
warning (it only warns when it installs something), and Zero then reports
"updated via npm to 0.10.0" while staying on 0.9.0.

So this release goes out once more under the old name, as a bridge:

- The publish job assembles the same verified artifacts a second time with
  the new `build-platform-packages.mjs --package-name @gitlawb/zero` and
  publishes them with publish-packages.sh, after the @twigpine/zero publish.
  A gate runs it only while @gitlawb/zero's latest is still 0.9.0 (or is
  this version, on a re-run), so it happens once.
- The build knows its name from the package.json beside it. The Go side
  reports InstallMethodNpmLegacy: `zero upgrade` refuses with the switch
  steps before any release check (so it works offline too), and
  `zero update --check` adds them whether or not an update is out.
- bin/zero.js prints the same steps after zero exits (the TUI would hide a
  notice printed first), only when stderr is a terminal, and not for
  `update`/`upgrade`, which print it themselves.
- Updating from 0.9.0 now installs a real version, so npm also shows the
  deprecation message.

The steps say to uninstall @gitlawb/zero before installing @twigpine/zero:
both own the `zero` bin, npm refuses the second install with EEXIST, and a
forced install loses the bin when the old package is later removed (checked
against npm 11.16).

docs/NPM_PACKAGING.md has the maintainer runbook: bootstrap and trust the new
package, retarget the old one's trusted publisher for the bridge, then
deprecate @gitlawb/zero, revoke its publisher, and delete the bridge steps.

* fix(npm): only skip a publish when npm says the version is not there

publish-packages.sh treated any failed `npm view <name>@<version>` as "not
published". A transient network, auth or registry error on the re-run that the
skip exists for therefore published a version that was already out, and npm
rejected it with EPUBLISHCONFLICT; with the registry unreachable it published
blind. npm 11 reports every kind of missing (no such package, no such version,
including platform versions) as E404, so only E404 now means "not published".
Any other failure is retried three times and then stops the job before it
publishes anything, with npm's own error shown.

The test fake now answers a missing version with E404 the way npm does (it
returned an empty success before), and can fail lookups with a network error.
The two new tests fail on the old lookup: one with EPUBLISHCONFLICT after
transient failures, the other by publishing while the registry is unreachable.

Also renames the unused retry variable in the legacy-name gate step to `_`
(shellcheck SC2034), matching the script.
2026-10-15 11:05:51 +08:00

297 lines
12 KiB
YAML

name: ZERO
description: >-
Run the ZERO coding agent headlessly in a GitHub workflow. ZERO is
model- and provider-agnostic: you supply the provider and API key as inputs.
author: Twigpine
branding:
icon: terminal
color: green
inputs:
prompt:
description: The instruction for ZERO to execute. Mutually exclusive with prompt-file.
required: false
default: ""
prompt-file:
description: Path (relative to working-directory) to a file whose contents are the prompt.
required: false
default: ""
provider:
description: >-
Provider id to activate (for example openai, anthropic, gemini, ollama, or
any configured OpenAI-/Anthropic-compatible endpoint). Optional when the
repository already carries a .zero/config.json with an active provider.
required: false
default: ""
api-key:
description: >-
The provider API key. Pass it from a repository or organization secret;
it is exported only for the ZERO step and never written to the log.
required: false
default: ""
api-key-env:
description: >-
Name of the environment variable the chosen provider reads its key from
(for example OPENAI_API_KEY or ANTHROPIC_API_KEY). When set together with
api-key, that variable is exported only for the ZERO step. Leave the action
provider-agnostic by passing the env name your provider expects rather
than hardcoding one here.
required: false
default: ""
model:
description: Model id to use for the run. Defaults to the resolved provider's default.
required: false
default: ""
mode:
description: Run mode passed to `zero exec --mode` (for example smart, deep, or fast).
required: false
default: ""
auto:
description: >-
Autonomy ceiling passed to `zero exec --auto` (low, medium, or high).
Defaults to the conservative `low` so an unattended CI run cannot take
high-impact actions implicitly.
required: false
default: low
self-correct:
description: >-
When true, allow ZERO to request a stronger model mid-run if it hits a
wall (maps to `zero exec --allow-escalation`).
required: false
default: "false"
add-dir:
description: >-
Newline- or comma-separated extra write roots granted to the run (each
passed as `zero exec --add-dir`). The checked-out repository is always
writable; everything else stays read-only unless listed here.
required: false
default: ""
worktree:
description: When true, run inside an isolated git worktree (`zero exec --worktree`).
required: false
default: "false"
output-format:
description: >-
Output format for `zero exec` (text, json, or stream-json). Defaults to
stream-json, which is captured to a file and exposed as the output-file
output and uploaded by the caller if desired.
required: false
default: stream-json
post-to:
description: >-
Optional destination for a run summary after ZERO finishes: `pr-comment`
posts to the triggering pull request using github-token, `slack` posts to
slack-webhook-url, or `none` (default) posts nowhere.
required: false
default: none
slack-webhook-url:
description: >-
Slack incoming-webhook (or generic webhook) URL used when post-to is
slack. Pass it from a secret; it is never written to the log.
required: false
default: ""
github-token:
description: >-
Token used to post a PR comment when post-to is pr-comment. Defaults to
the workflow's GITHUB_TOKEN; the workflow must grant pull-requests: write.
required: false
default: ${{ github.token }}
working-directory:
description: Directory to run ZERO in. Defaults to the workspace root.
required: false
default: ${{ github.workspace }}
zero-version:
description: >-
ZERO release version/tag to install (for example v1.2.3 or latest).
Defaults to the ref this action was resolved at, falling back to latest.
required: false
default: ""
zero-repo:
description: Repository to install the ZERO release from (owner/repo).
required: false
default: Twigpine/zero
outputs:
exit-code:
description: The exit code ZERO returned (0 success, 2 usage, 3 provider, non-zero otherwise).
value: ${{ steps.run.outputs.exit-code }}
output-file:
description: Path to the captured ZERO stdout (the raw output-format stream).
value: ${{ steps.run.outputs.output-file }}
summary:
description: A short human-readable summary line parsed from the run, when available.
value: ${{ steps.run.outputs.summary }}
runs:
using: composite
steps:
- name: Install ZERO
shell: bash
env:
ZERO_REPO: ${{ inputs.zero-repo }}
ZERO_VERSION_INPUT: ${{ inputs.zero-version }}
ACTION_REF: ${{ github.action_ref }}
ACTION_PATH: ${{ github.action_path }}
RUNNER_TEMP: ${{ runner.temp }}
run: |
set -euo pipefail
if [ "${RUNNER_OS}" = "Windows" ]; then
echo "::error::The ZERO action currently supports Linux and macOS runners only." >&2
exit 1
fi
# Resolve the version to install: explicit input wins; otherwise the ref
# this action was checked out at (a tag); otherwise the latest release.
version="${ZERO_VERSION_INPUT}"
if [ -z "${version}" ]; then
version="${ACTION_REF}"
fi
# A commit-SHA pin (github.action_ref is the 40-hex SHA) is not a release
# tag, so fall back to latest rather than fetching releases/download/v<sha>.
case "${version}" in
""|refs/*|*/*) version="latest" ;;
esac
if printf '%s' "${version}" | grep -qiE '^[0-9a-f]{40}$'; then
version="latest"
fi
install_dir="${RUNNER_TEMP}/zero-bin"
mkdir -p "${install_dir}"
# Reuse ZERO's own checksum-verifying installer, shipped with the action.
ZERO_REPO="${ZERO_REPO}" ZERO_VERSION="${version}" ZERO_INSTALL_DIR="${install_dir}" \
bash "${ACTION_PATH}/scripts/install.sh"
echo "${install_dir}" >> "${GITHUB_PATH}"
- name: Run ZERO
id: run
shell: bash
working-directory: ${{ inputs.working-directory }}
env:
# Secrets are passed via the environment and never echoed. `set +x` and
# the absence of any `echo` of these values keep them out of the log.
ZERO_INPUT_API_KEY: ${{ inputs.api-key }}
ZERO_INPUT_API_KEY_ENV: ${{ inputs.api-key-env }}
ZERO_INPUT_PROVIDER: ${{ inputs.provider }}
ZERO_INPUT_PROMPT: ${{ inputs.prompt }}
ZERO_INPUT_PROMPT_FILE: ${{ inputs.prompt-file }}
ZERO_INPUT_MODEL: ${{ inputs.model }}
ZERO_INPUT_MODE: ${{ inputs.mode }}
ZERO_INPUT_AUTO: ${{ inputs.auto }}
ZERO_INPUT_SELF_CORRECT: ${{ inputs.self-correct }}
ZERO_INPUT_ADD_DIR: ${{ inputs.add-dir }}
ZERO_INPUT_WORKTREE: ${{ inputs.worktree }}
ZERO_INPUT_OUTPUT_FORMAT: ${{ inputs.output-format }}
ACTION_PATH: ${{ github.action_path }}
RUNNER_TEMP: ${{ runner.temp }}
run: |
set -euo pipefail
# Export the provider key under the env name the provider expects. This
# keeps the action provider-agnostic: no provider->env mapping is baked
# in. The value is never printed.
if [ -n "${ZERO_INPUT_API_KEY}" ] && [ -n "${ZERO_INPUT_API_KEY_ENV}" ]; then
if ! printf '%s' "${ZERO_INPUT_API_KEY_ENV}" | grep -qE '^[A-Za-z_][A-Za-z0-9_]*$'; then
echo "::error::api-key-env must be a valid environment variable name." >&2
exit 2
fi
# Export in-process for this Run ZERO step so zero can read it; do
# not persist provider credentials to later job steps.
export "${ZERO_INPUT_API_KEY_ENV}=${ZERO_INPUT_API_KEY}"
fi
if [ -n "${ZERO_INPUT_PROVIDER}" ]; then
export ZERO_PROVIDER="${ZERO_INPUT_PROVIDER}"
fi
args=(exec)
if [ -n "${ZERO_INPUT_PROMPT_FILE}" ] && [ -n "${ZERO_INPUT_PROMPT}" ]; then
echo "::error::The 'prompt' and 'prompt-file' inputs are mutually exclusive." >&2
exit 2
fi
if [ -n "${ZERO_INPUT_PROMPT_FILE}" ]; then
args+=(--file "${ZERO_INPUT_PROMPT_FILE}")
elif [ -n "${ZERO_INPUT_PROMPT}" ]; then
args+=(--prompt "${ZERO_INPUT_PROMPT}")
else
echo "::error::Provide either the 'prompt' or 'prompt-file' input." >&2
exit 2
fi
[ -n "${ZERO_INPUT_MODEL}" ] && args+=(--model "${ZERO_INPUT_MODEL}")
[ -n "${ZERO_INPUT_MODE}" ] && args+=(--mode "${ZERO_INPUT_MODE}")
[ -n "${ZERO_INPUT_AUTO}" ] && args+=(--auto "${ZERO_INPUT_AUTO}")
[ "${ZERO_INPUT_SELF_CORRECT}" = "true" ] && args+=(--allow-escalation)
[ "${ZERO_INPUT_WORKTREE}" = "true" ] && args+=(--worktree)
args+=(--output-format "${ZERO_INPUT_OUTPUT_FORMAT}")
# No --notify in CI (no terminal); the optional Slack post is handled
# separately below so it works regardless of output format.
args+=(--no-notify)
# Split add-dir on newlines and commas; each non-empty entry is a root.
if [ -n "${ZERO_INPUT_ADD_DIR}" ]; then
while IFS= read -r dir; do
# Trim only surrounding whitespace so paths with internal spaces survive.
dir="$(printf '%s' "${dir}" | sed 's/^[[:space:]]*//; s/[[:space:]]*$//')"
[ -n "${dir}" ] && args+=(--add-dir "${dir}")
done < <(printf '%s' "${ZERO_INPUT_ADD_DIR}" | tr ',' '\n')
fi
output_file="${RUNNER_TEMP}/zero-output.txt"
# Run ZERO. Capture stdout to a file AND echo it to the log. Surface the
# real exit code as the step's status. The sandbox is active by default;
# this action never enables unsafe permission-skipping.
set +e
zero "${args[@]}" | tee "${output_file}"
code=${PIPESTATUS[0]}
set -e
echo "exit-code=${code}" >> "${GITHUB_OUTPUT}"
echo "output-file=${output_file}" >> "${GITHUB_OUTPUT}"
# Best-effort one-line summary. Structured output ends with lifecycle
# metadata, so parse final/error events instead of tailing stdout.
summary="$(node "${ACTION_PATH}/scripts/action-summary.mjs" "${ZERO_INPUT_OUTPUT_FORMAT}" "${output_file}" || true)"
{
echo "summary<<ZERO_EOF"
printf '%s\n' "${summary}"
echo "ZERO_EOF"
} >> "${GITHUB_OUTPUT}"
exit "${code}"
- name: Post run summary to pull request
if: ${{ always() && inputs.post-to == 'pr-comment' && github.event.pull_request.number != '' }}
shell: bash
env:
GH_TOKEN: ${{ inputs.github-token }}
PR_NUMBER: ${{ github.event.pull_request.number }}
ZERO_EXIT: ${{ steps.run.outputs.exit-code }}
ZERO_SUMMARY: ${{ steps.run.outputs.summary }}
run: |
set -euo pipefail
status="succeeded"
[ "${ZERO_EXIT}" != "0" ] && status="failed (exit ${ZERO_EXIT})"
body="$(printf 'ZERO run %s.\n\n%s' "${status}" "${ZERO_SUMMARY}")"
gh api "repos/${GITHUB_REPOSITORY}/issues/${PR_NUMBER}/comments" \
-f body="${body}" >/dev/null
- name: Post run summary to Slack
if: ${{ always() && inputs.post-to == 'slack' && inputs.slack-webhook-url != '' }}
shell: bash
env:
SLACK_WEBHOOK_URL: ${{ inputs.slack-webhook-url }}
ZERO_EXIT: ${{ steps.run.outputs.exit-code }}
ZERO_SUMMARY: ${{ steps.run.outputs.summary }}
RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
run: |
set -euo pipefail
status="succeeded"
[ "${ZERO_EXIT}" != "0" ] && status="failed (exit ${ZERO_EXIT})"
text="ZERO run ${status} for ${GITHUB_REPOSITORY}@${GITHUB_REF_NAME}. ${ZERO_SUMMARY} ${RUN_URL}"
# Slack incoming webhooks accept {"text": "..."}; jq escapes the body and
# the URL is never echoed. A delivery failure is a warning, not a failure.
payload="$(jq -n --arg text "${text}" '{text: $text}')"
curl --fail --silent --show-error -X POST -H 'Content-Type: application/json' \
-d "${payload}" "${SLACK_WEBHOOK_URL}" >/dev/null || \
echo "::warning::Slack webhook delivery failed (run not affected)."