git 작업 관리, 테스트 증거 기록 및 검증 정책 강제를 통해 에이전트 기반 코딩을 위한 워크플로우 가이드를 제공합니다.
다른 이름 BrainerVirus/workit, @workit-dev/core, @workit-dev/cursor
- 스타
- 1
- Tools
- —
- 다운로드, 30일
- —
- 최신 버전
- —
- 마지막 커밋
- 2026년 11월 13일
- 라이선스
- MIT
workit MCP란?
git 작업 관리, 테스트 증거 기록 및 검증 정책 강제를 통해 에이전트 기반 코딩을 위한 워크플로우 가이드를 제공합니다.
프로젝트 README에서 생성된 요약
workit MCP server 정보
에이전트 기반 코딩 도구를 사용하는 개발자는 workit을 사용하여 자동화된 프로그래밍 작업을 위한 구조화된 워크플로우 가이드를 구축할 수 있습니다. 브랜치 생성, 커밋, 푸시, pull request 생성과 같은 git 작업의 기술적 실행을 관리하므로, 사용자는 수동적인 기록 작업 대신 제품 판단에 집중할 수 있습니다. 이 시스템은 Claude Code, Cursor, OpenCode, Codex, Pi와 같은 인기 있는 AI 호스트와 통합되거나 독립형 CLI로 작동합니다.
신뢰성을 보장하기 위해 이 서버는 테스트 결과를 단순한 주장이 아닌 관찰된 증거로 기록하는 검증 시스템을 사용합니다. 또한 SHA 키 기반의 원장을 유지하여 코드 작성자와 코드 검증자 사이의 엄격한 분리를 강제합니다. 사용자는 워크스페이스별 권한 설정을 통해 에이전트가 수동 개입 없이 코드를 푸시, pull request 생성 또는 머지할 수 있는지 결정하여 제한된 자율성을 정의할 수 있습니다. 이 설정을 위해서는 Node.js 24 이상의 버전이 필요하며, workit CLI를 통해 다양한 개발 환경에 걸쳐 네이티브 통합을 구성할 수 있습니다.
활용 사례
- 1git 워크플로우 실행 자동화AI 에이전트가 브랜치 생성, 커밋, 저장소로의 코드 푸시 등 전체 프로세스를 처리할 수 있습니다.
- 2증거 기반 테스트 결과 검증개발자는 에이전트가 주장하는 테스트 통과 여부가 실제 기록된 종료 코드 및 트리 상태에 의해 뒷받침되는지 확인할 수 있습니다.
- 3독립적인 코드 검증 강제다른 세션의 판정을 요구함으로써 AI 에이전트가 자신이 작업한 내용을 스스로 검증하는 것을 방지할 수 있습니다.
- 4에이전트 자율성 제한 설정개발자는 에이전트가 코드 커밋만 수행할 수 있고 허가 없이 머지할 수는 없도록 제한할 수 있습니다.
프로젝트의 README를 바탕으로 작성되었습니다. 모든 활용 사례는 해당 내용의 인용문과 대조하여 확인합니다.
workit MCP 설치 방법
형식 확인: 2026년 11월 10일Cursor에 추가Goose에 추가Claude Code에 workit MCP 설치
$ claude mcp add workit -- npx -y @workit-dev/coreCursor에 workit MCP 설치
~/.cursor/mcp.json
{
"mcpServers": {
"workit": {
"command": "npx",
"args": [
"-y",
"@workit-dev/core"
]
}
}
}VS Code에 workit MCP 설치
.vscode/mcp.json
{
"servers": {
"workit": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@workit-dev/core"
]
}
}
}VS Code CLI에 workit MCP 설치
$ code --add-mcp '{"name":"workit","type":"stdio","command":"npx","args":["-y","@workit-dev/core"]}'Claude Desktop에 workit MCP 설치
claude_desktop_config.json
{
"mcpServers": {
"workit": {
"command": "npx",
"args": [
"-y",
"@workit-dev/core"
]
}
}
}Codex에 workit MCP 설치
$ codex mcp add workit -- npx -y @workit-dev/coreCodex config.toml에 workit MCP 설치
~/.codex/config.toml
[mcp_servers.workit] command = "npx" args = ["-y", "@workit-dev/core"]
Gemini CLI에 workit MCP 설치
$ gemini mcp add -s user workit npx -- -y @workit-dev/coreGemini settings.json에 workit MCP 설치
~/.gemini/settings.json
{
"mcpServers": {
"workit": {
"command": "npx",
"args": [
"-y",
"@workit-dev/core"
]
}
}
}자동 검사
2026년 11월 14일에 실행됨전체 보고서 →자동 검사를 통과했습니다. 이는 보안 감사가 아닙니다. 코드를 직접 읽고 서버에 필요한 최소한의 권한만 부여하십시오. 검사 방식
- 소스 저장소 연결 및 접근 가능brainervirus/workit is public and reachablegithub.com/brainervirus/workit ↗통과
- 패키지가 이 저장소를 소스로 명시@workit-dev/core, @workit-dev/cursor name brainervirus/workit as sourcenpmjs.com/package/@workit-dev/core ↗통과
- 빌드 출처(Sigstore 증명)Sigstore attestation for @workit-dev/core, @workit-dev/cursornpmjs.com/package/@workit-dev/core/v/2.5.0#provenance ↗통과
- 오픈 소스 라이선스 (OSI)MITspdx.org/licenses/MIT.html ↗통과
- 최근 1년 내 커밋 또는 릴리스Last commit or release 2026-11-13github.com/brainervirus/workit ↗통과
- OpenSSF Scorecard검사되지 않음
- 패키지 이름이 인기 패키지와 유사하지 않음No popular package one edit away from @workit-dev/core, @workit-dev/cursor통과
- 패키지 및 해당 종속성에 중요한 보안 권고 없음검사되지 않음
- 모델에 대한 숨겨진 지침 없음0 tool descriptions and the README read; no override phrases, credential paths or hidden characters통과
- 파괴적인 도구에 라벨 표시됨검사되지 않음
- 원격 엔드포인트가 MCP 핸드셰이크에 응답검사되지 않음
README
GitHub에서 보기Workit
Workflow rails for agentic coding. The agent decides what to do; the
workit CLI does the how (branch, commit, push, PR, CI, stack, merge) and
records proof of what it observed. You stay in the loop for product
judgment, not for checking whether the tests really ran or the push really
landed.
One shared core ships as native integrations for Claude Code, OpenCode, Cursor, Codex CLI/desktop and Pi, plus a standalone CLI.
- Observed, not claimed.
workit checkruns your real test command and records the exit code, HEAD and tree. An agent saying "tests pass" is a note, not evidence. - Author ≠ verifier. Verdicts are recorded per session in a SHA-keyed ledger; the session that wrote the code can never verify it.
- Bounded autonomy. Per-workspace grants decide how far an agent may go (push, PR, merge) without asking. Your host's permission prompts still win.
- No bookkeeping. Every branch is its own implicit task. Agents never manage ids.
Contents
Quickstart
Requires Node.js 24+. Claude Code, Codex and Cursor hooks start the node
on your PATH; in a repo pinned to an older Node they find a Node 24+ on their
own (WORKIT_NODE, the Node workit setup recorded, then fnm, nvm, volta, mise
and asdf installs). workit doctor (row hook_node) shows which Node they use;
with none, install Node 24 or run workit setup from a Node 24 shell. The setup wizard detects your hosts, installs the
ones you pick with their native commands, and writes your global config:
npx @workit-dev/cli init
workit doctor # after the CLI install below
workit setup # optional: let agents run Workit's delivery verbs without a prompt
workit setup shows the exact diff, asks once (--yes for a script,
--dry-run to only look) and writes a Workit-owned block into each host's own
permission config: enumerated workit git|pr|ci|stack|release|task|ledger|…
verbs only, never workit check -- …, workit lock run -- … or workit grant. Run it in your own terminal, not from an agent session (it refuses
there). workit uninstall removes exactly what it added, and workit doctor
reports drift. Per-host details are in
the hosts guide.
Or install a single host by hand. Whatever the install path, every host
updates with workit upgrade and uninstalls with workit uninstall (pick the
host); the native commands below are the fallback.
To install or repair Workit on some hosts without the wizard, name them:
workit install --hosts=cursor --preview prints the exact files, config keys,
host commands and notes and changes nothing; --yes installs them. A healthy
host plans nothing, so a second run changes nothing; a host whose CLI is not on
PATH shows a blocked row. It never touches another host or ~/.config/workit.
To remove Workit from some hosts without the picker, name them:
workit uninstall --hosts=cursor --preview prints the exact files, config keys
and host commands it would remove and changes nothing; --yes removes them.
Other hosts, your own rules and ~/.config/workit stay as they are.
workit setup, workit install and workit uninstall are yours to run, in
your own terminal: the CLI refuses --yes from an agent session and hands you
the command, and the host asks about or denies an agent's call.
Coming from @brainervirus/workit-* (14.x)? The packages moved to
@workit-dev/* and restarted at 1.0.0: run
npx -y @brainervirus/workit-cli@latest upgrade once, and see
the migration guide.
A CLI older than 8.38.0 has no --yes (it fails with unknown option --yes).
Update the CLI once with npx -y @workit-dev/cli@latest upgrade --cli --hosts=none --yes
(or, on the old CLI, workit upgrade --cli --hosts=none --apply --confirm), then run
workit upgrade.
claude plugin marketplace add BrainerVirus/workit
claude plugin install workit@workit
Update: workit upgrade (fallback: claude plugin marketplace update workit && claude plugin update workit@workit; Claude Code does not auto-update plugins by
default). Restart Claude Code afterwards: it loads plugins once per process.
Uninstall: workit uninstall, or claude plugin uninstall workit@workit.
Pin to a local checkout instead (sources run with Bun, edits apply without a
rebuild; disable the marketplace copy with claude plugin disable workit@workit):
bun install
bun packages/workit-claude-code/scripts/build.ts --skills-only
claude --plugin-dir "$PWD/packages/workit-claude-code"
Skills appear as /workit:<name>, plus verifier, reviewer and
implementer agents. See the Claude Code guide.
{
"$schema": "https://opencode.ai/config.json",
"plugins": ["@workit-dev/opencode"]
}
Update: workit upgrade (fallback:
opencode plugin update @workit-dev/opencode; other plugins stay as
they are). OpenCode 2.x loads the new version live, with no restart. Use the
plugins key; workit setup --hosts=opencode moves an entry out of the
deprecated plugin key. Uninstall: workit uninstall, or remove the entry
from plugins.
Local pin: build once with bun packages/workit-opencode/scripts/build.ts, then
use "plugins": ["file:///path/to/workit/packages/workit-opencode"].
Run npx @workit-dev/cli init and select Cursor: it installs the
plugin (MCP server, session hook, contract rule, skills). Update:
workit upgrade (fallback: workit upgrade --hosts=cursor), which also
reinstalls missing plugin files. Uninstall: workit uninstall. Manual MCP
setup is in the hosts guide.
codex plugin marketplace add https://github.com/BrainerVirus/workit.git
codex plugin add workit@workit
Update: workit upgrade, which updates the plugin and refreshes the Codex agent
files; the app-server daemon keeps running and its new threads load the update (fallback:
codex plugin marketplace upgrade workit && codex plugin add workit@workit, then a new Codex session). Uninstall:
workit uninstall.
Skills are invoked as $workit-<name> or from the /skills picker.
pi install npm:@workit-dev/pi
pi install ./packages/workit-pi -l --approve # local checkout
Update: workit upgrade (fallback: pi update --extension npm:@workit-dev/pi). Uninstall: workit uninstall.
npx -y @workit-dev/cli@latest upgrade --cli --hosts=none --yes
workit help
This installs the CLI into ~/.local/share/workit/versions/<v> with a
~/.local/bin/workit launcher (Windows: %LOCALAPPDATA%\workit\install),
independent of which Node version manager a shell uses. workit upgrade
moves an older npm i -g copy over and workit doctor lists every copy on
your machine (install guide).
Every verb works without a host; hosts just put workit and the skills in
front of the agent. Update with workit upgrade --hosts=none --cli; remove it with
workit uninstall (pick the CLI). It removes the ~/.local/bin/workit link
only when it points into the install; in bin/, only the launcher Workit wrote
and its temp files; it unlinks current; and from versions/, only entries
holding the Workit CLI package. Anything else (a file of your own, a link
elsewhere, an install root set to your home) is left untouched and named.
Core concepts
Implicit task per branch. The first note, check, ledger record or commit
on a branch creates its task; every task command without an id applies to it.
State lives in the git common directory (.git/workit/, shared by all
worktrees) as an append-only event log. workit task status shows it;
workit handoff prints a resume brief.
Observed evidence (workit check). workit check test runs the configured
command (from .workit/checks.json, else detected defaults such as
<pm> run test, go test ./..., cargo test, pytest) and records exit
code, HEAD and tree. Gates accept only a fresh passing run of a configured
check; a check is stale once the worktree changes. Ad-hoc
workit check -- <cmd> runs are recorded but never satisfy a gate.
Ledger verdicts (author ≠ verifier). workit ledger verdict records a
result (verified, tests-verified, type-check-only, blocked, failed)
for a branch head under the acting session. A verdict from the session that
authored the commits is never accepted. Verdicts carry across rebases when
the patch-id and diff are unchanged, and across docs-only follow-up pushes;
one verdict at the top of a stack covers its unchanged lower PRs
(carry-over).
workit ledger check answers "is this head independently verified?".
Autonomy grants. Per-workspace grants in ~/.config/workit/workspaces.json
set the ceiling: push, pr and rerun are allowed by default; merge needs
true or verified, and both merge only with the verdict the PR's review
lane asks for (an accepted independent one outside verification: proportional; true also allows a ledger-recorded --unverified --reason bypass);
defaultEndpoint is where an unnamed delivery request
stops: commit (default), pr (push and open the PR), green (open the PR,
then babysit CI, review threads and required rebases until it is merge-ready,
never merging) or merged (green, then workit pr merge; it acts as green
until the workspace has the merge grant). Without the push or pr grant any
of them acts as commit; workit grant show reports the effective endpoint. Raising a grant or the endpoint needs
you at a terminal; agents can only lower them. The authority fields are sealed
(workspaces.json.seal): workit workspace seal, in your own terminal, shows
what changed and accepts it; while the seal is missing or does not match,
nothing above the last seal (or the defaults) applies, a changed forge identity
blocks forge verbs, and workit doctor warns (grant_seal); workit upgrade
and workit setup at your terminal offer the first seal. It stops an
honest agent, not one that also forges the seal (D18). A config redirect
(WORKIT_CONFIG and the like) never raises a grant. See
grants. workit workspace list|add|set adds or
adjusts one entry without the wizard and never raises a grant
(configuration).
Policy. Judge tracked work in four calls (workit policy assess --judge risk=trivial|normal|high behavior=yes|no product-choice=yes|no plan=yes|no,
or the flat workit_policy tool fields) and Workit derives what it needs:
nothing for trivial work, an observed workit check test and the verdict the
change's review lane asks for, an independent live verdict and a plan for high
risk. Under the default verification: "proportional" the lane comes from the
diff: none (docs, comments, formatting, generated copies) needs no verdict,
self (a small change off risk paths) the author's own --self verdict,
focused and full (risk paths such as hooks, auth, release, migrations or
the repo's own .workit/risk-paths, a lockfile, or a large diff) a non-author
verdict, at task close, the Stop hook and the merge gate alike; when unsure, a
file counts as a behavior change. workit pr status and workit policy assess show it. An open product choice or a needed plan blocks code
edits until recorded, on every host with a pre-write hook (advisory on Codex).
See verification. Host permissions
(allow/ask/deny, sandbox) always stay authoritative; Workit adds no consent
prompts of its own.
Release tracks. A repository with several release lines (say
nun-develop -> nun-master and develop -> master) lists them under the
workspace's releaseTracks; branch bases, PR targets, merge-back hints and
protected branches then follow the line each branch belongs to. See
release tracks.
Stacks. workit stack manages plain base-branch PR chains on GitHub and
GitLab (no Graphite or gh stack needed): plan, restack after a merge, and
land the stack all or nothing: every PR is checked before any merges
(--max N lands just the first N). See
delivery and stacks.
Daily workflow
shape → implement → review → ship
│ │ │ └─ push, PR, CI green, land when granted
│ │ └─ non-author verdict in the ledger
│ └─ small steps, `workit check`, prove it on the running app
└─ brainstorm, grill open choices, slice into PRs
- Shape (
/wk-shape,/workit:shape): agree what to build and how to slice it. A spec or ADR is proposed only when it pays off. - Implement: build in small steps; run
workit check <name>; commit withworkit git commit. - Review: a different session (the
verifier/revieweragent, or a reviewer started with its ownWORKIT_SESSION_ID) recordsworkit ledger verdict. - Ship:
workit git push,workit pr create --issue <#n|ID>(links the issue on GitHub, GitLab or YouTrack),workit ci wait,workit pr status;workit pr mergeonly when the PR is READY, verified and themergegrant allows it, and then deletes the merged head branch (workspacedeleteBranch,--keep-branchto opt out; never a protected, fork, stack-parent or foreign-author branch).workit verify-deliveryconfirms it landed. A workspace can opt in to a changelog step (workit workspace set <ws> changelog=update|create docs=when-user-facing, default off): ship writes the Unreleased entry withworkit changelog applybefore the PR, andpr createrefuses a user-facing PR without one unless given--no-changelog "<reason>"(changelog gate). The PR body fills the repo's GitHub or GitLab template, else Workit's Summary, Evidence and Merge danger:workit pr templateprints it. When a template is chosen (the repo's, or Workit's withprTemplate: workit),pr createrefuses a body that leaves out its headings or keeps its<!-- -->placeholders (workspaceprTemplate: auto|repo|workit|<path>, PR template).
For independent slices, fanout runs one worker per isolated worktree with
a fixed brief and file-scope manifest, and a non-author verifier per slice.
workit fanout plan refuses incomplete briefs and overlapping scopes before
any spawn, workit fanout brief renders each worker's brief with the lead's
standing orders (workit ledger standing; workers already running get each
change on their next tool call, and standing add --stop tells them to stop
and report), workit fanout status flags
stuck workers and suggests a landing
order, workit fanout worktree create|release gives each worker its own
worktree on every host (with node_modules linked so its tests import its own
code; Workit's hooks keep an implementer inside it), and workit fanout check
gates fan-in. See
parallel slices.
Skills
Skills load automatically when the task fits, or explicitly: /wk-<name>
(OpenCode, Cursor, Pi), /workit:<name> (Claude Code), $workit-<name>
(Codex).
| Skill | Use it to |
|---|---|
shape |
Brainstorm, grill open choices, challenge premises, slice into PRs, propose a spec/ADR when it pays |
implement |
Build a change in small verified steps and hand verification to a non-author |
review |
Review a diff/branch/PR (intent, standards, test quality, blast radius) as a non-author verdict |
debug |
Find the root cause from a deterministic repro before patching |
ship |
Open or stack PRs, fix red CI, answer threads, land verified PRs when granted |
continue |
Checkpoint, hand off, and resume across sessions by re-verifying inherited claims |
bdd |
Turn requirements into Given/When/Then and work test-first in RED/GREEN slices |
test-audit |
Find tautological or low-value tests and replace them with ones that catch breaks |
deslop |
Remove dead code, restating comments and filler before a PR, behavior unchanged; depth follows the review lane |
fanout |
Run independent slices in parallel worktrees with per-slice verifiers |
verify-app |
Generate the project's own verify-<app> skill that drives the real app |
retro |
User-invoked: find repeated friction in recent sessions and propose cited fixes ranked by enforcer strength; never applies them |
architecture |
User-invoked: rank cited deepening opportunities from churn and ledger friction, grill the chosen one into slices; also restructures AGENTS.md in reviewable passes and sweeps tests; never refactors without approval |
CLI reference
workit help <command> [<subcommand>], or -h/--help anywhere after a
command (before a bare --), prints its usage, flags and examples and runs
nothing. A bare -- ends any command's options: after it -h is an ordinary
value, and a command that takes no positional arguments refuses anything there
(exit 2) before it runs. A value that starts with -
goes as --flag=value (workit git commit --message=-h); -h right after a
flag that takes a value is refused as ambiguous. A --flag a command does not
take fails with exit 2 and an unblock naming the closest valid flag (or the
right form: workit status points at workit task status, pr create --head
at switching branches). Alternate spellings work: --fanout and --name both
select a fanout plan, ledger verdict --sha <sha> is --head <sha> (the sha
must be the head being judged), and task list --status all lists every task.
Global flags:
--json (envelope {"ok","code","data","error"?,"unblock"?}), --cwd <dir>,
--version. Exit codes: 0 ok, 1 failed, 2 usage, 3 blocked,
4 busy/pending, 5 unavailable.
In a color terminal, help prints bold headings, commands and flags in cyan
and dim summaries; NO_COLOR, --json, a pipe or TERM=dumb give the plain
text.
| Command | Purpose |
|---|---|
workit init |
Interactive setup wizard (hosts, config, optional project files) |
workit setup [--hosts=…] [--dry-run | --yes] |
Write Workit's host permission block after showing the diff |
workit doctor [--json] [--fix] [--fix-lock] |
Offline installation health; repair what it can; clear a stale store lock |
workit upgrade [--preview | --yes] [--hosts=…] [--cli] |
Upgrade the CLI and every host after one confirmation, then verify with doctor |
workit upgrade --check / --set-updates=prompt|auto|off |
Compare installed and latest versions; choose how a launch handles a new release |
workit launch <host> --auto-upgrade |
Upgrade, then start a host |
workit completion zsh|bash|fish|powershell |
Print the shell completion script (setup and upgrade install it for $SHELL) |
workit gc [--dry-run] |
Compact task logs, drop unreferenced blobs and old check logs |
workit install --hosts=<h,…> [--preview|--yes] |
Install or repair Workit on the named hosts only; a healthy host changes nothing |
workit uninstall [--hosts=<h,…> [--preview|--yes]] |
Pick hosts (and the owned CLI install), or name them, and remove what Workit put there (keeps ~/.config/workit) |
workit grant show|set|unset |
Workspace autonomy grants and defaultEndpoint |
workit workspace list|add|set|rm|seal |
List, add, edit or remove workspace entries without the wizard (never raises a grant); seal accepts the authority fields at your terminal |
workit models show |
The model and effort each role (decider, implementer tiers, verifier, reviewer) runs on for a host |
workit task status|start|note|close|adopt |
The current branch's task |
workit handoff |
Resume brief: branch, HEAD, dirty state, verdict, next command |
workit check <name> |
Run a configured check and record observed evidence |
workit lock status|wait|run|heavy |
The machine-wide heavy-check lease: holders, waiting, running a command under it, heavy patterns |
workit test-audit [--diff] [--mutate] |
Flag tautological tests; optional diff-scoped mutation |
workit deslop scan|result |
Class the comments, suppressions and debug leftovers a branch adds, by review lane, with no agent; record one result line per PR (review depth) |
workit knowledge lint |
Lint AGENTS.md budget, local links, scaffold-only and duplicated rules; register as workit check knowledge |
workit ledger decision|ruling|verdict|standing|list|check|verify-integrity |
Decisions, rulings, SHA-keyed verdicts, a fanout's standing orders and the hash chain |
workit git branch|commit|push|trailers |
Policy-checked branch/commit; leased push with verified remote tip |
workit pr status|create|ready|edit|threads|reply|merge |
PR state and next action; SHA-bound create; review threads; gated merge |
workit ci wait|rerun|logs |
Wait for CI on the PR head; rerun failed jobs once per head; print the failing jobs' log tails |
workit stack plan|status|sync|land |
Base-branch PR stacks |
workit fanout plan|brief|check|status|worktree |
Register parallel slices (complete briefs, disjoint scopes); render each worker's brief with the standing orders; watch them (STUCK, PR, CI, verdict, landed); create and release slice worktrees; gate fan-in on scope, merge conflicts and landing order |
workit verify-delivery [push|pr|merge|release] |
Confirm on the remote that it landed |
workit release start|finish|publish|notes |
git-flow releases without the git-flow binary: cut, finish (merge, tag, merge back, forge release with user-facing notes) and publish per release track; notes draft|show|approve drafts the release notes, and under releaseNotes: review only the user approves them (release notes) |
workit youtrack note|time|meeting|update |
YouTrack comments, attachments, work time and state, one yes per update (YouTrack guide) |
workit changelog apply |
Add entries under the changelog's Unreleased section |
workit <family> <action> |
Low-level task families: task, policy, evidence, finding, decision, worker, state |
workit MCP 질문
workit MCP를 Claude Code에 추가하려면 어떻게 하나요?
터미널에서 claude mcp add workit -- npx -y @workit-dev/core를 실행합니다. 모든 프로젝트에서 사용할 수 있게 하려면 --scope user를 추가합니다.
workit MCP를 Cursor에 추가하려면 어떻게 하나요?
~/.cursor/mcp.json를 열고 mcpServers 아래에 서버를 추가합니다. 위 설치 섹션에 표시된 Cursor 구성을 사용합니다.
workit이 에이전트 기반 코딩에서 수행하는 역할은 무엇인가요?
에이전트가 수행할 작업을 결정하면, workit은 브랜치 생성, 커밋, 푸시, PR, CI 및 머지 등 실행 세부 사항을 처리하면서 관찰된 증거를 기록합니다.
workit은 테스트가 실제로 통과했는지 어떻게 보장하나요?
workit check를 사용하여 실제 테스트 명령을 실행하고 종료 코드, HEAD, 트리 상태를 기록함으로써 단순한 주장이 아닌 관찰된 증거를 제공합니다.
에이전트가 직접 작성한 코드를 스스로 검증할 수 있나요?
아니요, 코드를 작성한 세션이 해당 코드를 검증할 수 없도록 시스템에서 규칙을 강제합니다.
에이전트의 권한을 어떻게 제어할 수 있나요?
워크스페이스별 권한 설정을 통해 에이전트가 승인 없이 푸시, PR 생성 또는 머지를 수행할 수 있는지 등 작업 범위를 결정할 수 있습니다.