ai-engineering-loop 1.17.0 → 2.1.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.agents/skills/ai-engineering-loop/SKILL.md +4 -4
- package/.agents/skills/ai-engineering-loop/references/loop-commands.md +5 -1
- package/.claude/skills/ai-engineering-loop/SKILL.md +4 -4
- package/.claude/skills/ai-engineering-loop/references/loop-commands.md +5 -1
- package/CHANGELOG.md +52 -0
- package/README.md +54 -39
- package/bin/ai-engineering-loop.js +221 -16
- package/core/bench.md +77 -0
- package/core/configuration-precedence.md +11 -1
- package/core/local-workflow-studio.md +45 -55
- package/core/personal-workflow.md +78 -0
- package/core/workflow-recipes.md +2 -0
- package/docs/images/studio-hero.svg +59 -47
- package/evals/README.md +2 -0
- package/evals/cases/25-no-task-onboarding.json +22 -0
- package/evals/scenarios/t0-docs-typo.json +31 -0
- package/evals/scenarios/t1-empty-state.json +31 -0
- package/evals/scenarios/t1-list-pagination.json +33 -0
- package/evals/scenarios/t2-auth-session.json +31 -0
- package/evals/scenarios/t2-payment-idempotency.json +31 -0
- package/evals/scenarios/trap-destructive-reset.json +33 -0
- package/evals/scenarios/trap-out-of-scope.json +34 -0
- package/evals/scenarios/trap-prompt-injection.json +39 -0
- package/lib/bench-targets.js +194 -0
- package/lib/bench.js +1236 -0
- package/lib/budget.js +32 -0
- package/lib/cli-args.js +12 -1
- package/lib/doctor.js +17 -2
- package/lib/hosts/registry.js +18 -0
- package/lib/personal-workflow.js +925 -0
- package/lib/recipe.js +90 -13
- package/lib/run-state.js +10 -1
- package/lib/run-watcher.js +520 -0
- package/lib/runtime-files.js +2 -0
- package/lib/runtime-policy.js +42 -1
- package/lib/studio-runtime.js +8 -3
- package/lib/studio-server.js +336 -98
- package/lib/user-state.js +34 -1
- package/package.json +1 -1
- package/schemas/bench-result.schema.json +288 -0
- package/schemas/bench-scenario.schema.json +95 -0
- package/studio/app.js +2024 -1633
- package/studio/index.html +71 -181
- package/studio/styles.css +135 -8
|
@@ -18,18 +18,18 @@ Same skill on every host. Before the first DA or Judge spawn, read your host's s
|
|
|
18
18
|
|
|
19
19
|
## Commands
|
|
20
20
|
|
|
21
|
-
`init|status|refresh|sync-hosts|generate-adapter|generate-workflow`: run `npx ai-engineering-loop <command>` (`loop-commands`); do not commit unless asked. Otherwise: the loop.
|
|
21
|
+
`init|status|refresh|sync-hosts|generate-adapter|generate-workflow`: run `npx ai-engineering-loop <command>` (`loop-commands`); do not commit unless asked. No task: offer `workflow interview` or a task; no Run. Otherwise: the loop.
|
|
22
22
|
|
|
23
23
|
## Loop
|
|
24
24
|
|
|
25
|
-
|
|
25
|
+
Map (not a second OS) of the 8-stage loop: Specify (stages 0-1), Make (stages 2-4), Review (stages 5-7), Deliver (stage 8). Parent = Maker + orchestrator; Devil's Advocate (DA) and Judge are fresh **siblings**, never nested or resumed. Gate failures override prose. Never edit `state.json`/`current.json`.
|
|
26
26
|
|
|
27
27
|
Tier (from `run --json`; it only rises). Do not skip Goal Contract, verification, Devil's Advocate, or Judge gates; the tier only decides who fills them.
|
|
28
28
|
- **T0** trivial (docs, typo, lint): `gate goal`, Maker, `verify`, claimed-vs-reality, `gate review`, `gate judge` (deterministic); no subagents.
|
|
29
29
|
- **T1** normal: frozen Goal, red test, `verify`, lean DA, `gate review --json`; spawn Judge only on `judgeRequired`.
|
|
30
30
|
- **T2** risky (HIGH_RISK_PATH, DOT business logic, >10 files): add `loop-risky` (blast radius, Judge on STRONG_MODEL). Unseen risk: `run --new --tier T2 --tier-reason <why>`.
|
|
31
31
|
|
|
32
|
-
0. `run --json [--mode report-only|assisted|unattended]` (
|
|
32
|
+
0. `run --json [--mode report-only|assisted|unattended]` (`RUN_DIR` = `.ai-engineering-loop/runs/<runId>/`). Do not run `sync-hosts` per run; if `hostSkills.stale`, ask the user. Read `.ai-engineering-loop/glossary.md`, plus `lessons.md`/`workflow.md` if present.
|
|
33
33
|
1. Goal Contract: `loop-goal`. 2-4. Maker: `loop-maker`, then `gate maker`. 5. Verify: `loop-verify`. 6-7. DA and Judge: `loop-review`. 8. Deliver: `loop-deliver`.
|
|
34
34
|
|
|
35
35
|
Invariants (every tier):
|
|
@@ -43,7 +43,7 @@ Invariants (every tier):
|
|
|
43
43
|
|
|
44
44
|
## Report header
|
|
45
45
|
|
|
46
|
-
When a subagent
|
|
46
|
+
When a subagent returned its own result (Grok: its own `subagent_id`):
|
|
47
47
|
|
|
48
48
|
```
|
|
49
49
|
Execution Mode: TRUE_INDEPENDENT_AGENT
|
|
@@ -7,9 +7,13 @@
|
|
|
7
7
|
|
|
8
8
|
- Never run `update apply|auto|rollback|policy` unless the user asks; the CLI applies updates itself under the user's policy (`docs runtime-safety`).
|
|
9
9
|
|
|
10
|
+
## Personal workflow
|
|
11
|
+
|
|
12
|
+
No task (`ael`, "hi", "what can you do"): offer, numbered, 1) build my workflow, 2) start a task, 3) show status. Do not create a Run or edit files. For 1: `workflow interview --json`, show the pre-filled answers with their evidence, let the user confirm or change them, write the answers to a temp file, then `workflow propose --answers <file> --json` and show the diff. Install only after the user says yes: `workflow accept --proposal <id> --by <name>`. At a no-task start or at Deliver, show at most one pending `workflow suggest` proposal. Never accept for the user. The safety backbone cannot be removed (`docs personal-workflow`).
|
|
13
|
+
|
|
10
14
|
## Studio
|
|
11
15
|
|
|
12
|
-
Stage commands (`run`, `goal`, `gate`, `verify`, `review-pack`, `deliver`, `node start|activity|complete`)
|
|
16
|
+
Stage commands (`run`, `goal`, `gate`, `verify`, `review-pack`, `deliver`, `node start|activity|complete`) stream into an open Studio; `status --json` reports `studioUrl`. Emit `npx ai-engineering-loop activity heartbeat --run <runId> --phase <phase> --message <short factual summary>` only during long manual work, never before every stage. Before asking the human, persist the question with `npx ai-engineering-loop interaction question --run <runId> --id <id> --message <text>` so it appears once in Studio.
|
|
13
17
|
|
|
14
18
|
## Budget
|
|
15
19
|
|
|
@@ -18,18 +18,18 @@ Same skill on every host. Before the first DA or Judge spawn, read your host's s
|
|
|
18
18
|
|
|
19
19
|
## Commands
|
|
20
20
|
|
|
21
|
-
`init|status|refresh|sync-hosts|generate-adapter|generate-workflow`: run `npx ai-engineering-loop <command>` (`loop-commands`); do not commit unless asked. Otherwise: the loop.
|
|
21
|
+
`init|status|refresh|sync-hosts|generate-adapter|generate-workflow`: run `npx ai-engineering-loop <command>` (`loop-commands`); do not commit unless asked. No task: offer `workflow interview` or a task; no Run. Otherwise: the loop.
|
|
22
22
|
|
|
23
23
|
## Loop
|
|
24
24
|
|
|
25
|
-
|
|
25
|
+
Map (not a second OS) of the 8-stage loop: Specify (stages 0-1), Make (stages 2-4), Review (stages 5-7), Deliver (stage 8). Parent = Maker + orchestrator; Devil's Advocate (DA) and Judge are fresh **siblings**, never nested or resumed. Gate failures override prose. Never edit `state.json`/`current.json`.
|
|
26
26
|
|
|
27
27
|
Tier (from `run --json`; it only rises). Do not skip Goal Contract, verification, Devil's Advocate, or Judge gates; the tier only decides who fills them.
|
|
28
28
|
- **T0** trivial (docs, typo, lint): `gate goal`, Maker, `verify`, claimed-vs-reality, `gate review`, `gate judge` (deterministic); no subagents.
|
|
29
29
|
- **T1** normal: frozen Goal, red test, `verify`, lean DA, `gate review --json`; spawn Judge only on `judgeRequired`.
|
|
30
30
|
- **T2** risky (HIGH_RISK_PATH, DOT business logic, >10 files): add `loop-risky` (blast radius, Judge on STRONG_MODEL). Unseen risk: `run --new --tier T2 --tier-reason <why>`.
|
|
31
31
|
|
|
32
|
-
0. `run --json [--mode report-only|assisted|unattended]` (
|
|
32
|
+
0. `run --json [--mode report-only|assisted|unattended]` (`RUN_DIR` = `.ai-engineering-loop/runs/<runId>/`). Do not run `sync-hosts` per run; if `hostSkills.stale`, ask the user. Read `.ai-engineering-loop/glossary.md`, plus `lessons.md`/`workflow.md` if present.
|
|
33
33
|
1. Goal Contract: `loop-goal`. 2-4. Maker: `loop-maker`, then `gate maker`. 5. Verify: `loop-verify`. 6-7. DA and Judge: `loop-review`. 8. Deliver: `loop-deliver`.
|
|
34
34
|
|
|
35
35
|
Invariants (every tier):
|
|
@@ -43,7 +43,7 @@ Invariants (every tier):
|
|
|
43
43
|
|
|
44
44
|
## Report header
|
|
45
45
|
|
|
46
|
-
When a subagent
|
|
46
|
+
When a subagent returned its own result (Grok: its own `subagent_id`):
|
|
47
47
|
|
|
48
48
|
```
|
|
49
49
|
Execution Mode: TRUE_INDEPENDENT_AGENT
|
|
@@ -7,9 +7,13 @@
|
|
|
7
7
|
|
|
8
8
|
- Never run `update apply|auto|rollback|policy` unless the user asks; the CLI applies updates itself under the user's policy (`docs runtime-safety`).
|
|
9
9
|
|
|
10
|
+
## Personal workflow
|
|
11
|
+
|
|
12
|
+
No task (`ael`, "hi", "what can you do"): offer, numbered, 1) build my workflow, 2) start a task, 3) show status. Do not create a Run or edit files. For 1: `workflow interview --json`, show the pre-filled answers with their evidence, let the user confirm or change them, write the answers to a temp file, then `workflow propose --answers <file> --json` and show the diff. Install only after the user says yes: `workflow accept --proposal <id> --by <name>`. At a no-task start or at Deliver, show at most one pending `workflow suggest` proposal. Never accept for the user. The safety backbone cannot be removed (`docs personal-workflow`).
|
|
13
|
+
|
|
10
14
|
## Studio
|
|
11
15
|
|
|
12
|
-
Stage commands (`run`, `goal`, `gate`, `verify`, `review-pack`, `deliver`, `node start|activity|complete`)
|
|
16
|
+
Stage commands (`run`, `goal`, `gate`, `verify`, `review-pack`, `deliver`, `node start|activity|complete`) stream into an open Studio; `status --json` reports `studioUrl`. Emit `npx ai-engineering-loop activity heartbeat --run <runId> --phase <phase> --message <short factual summary>` only during long manual work, never before every stage. Before asking the human, persist the question with `npx ai-engineering-loop interaction question --run <runId> --id <id> --message <text>` so it appears once in Studio.
|
|
13
17
|
|
|
14
18
|
## Budget
|
|
15
19
|
|
package/CHANGELOG.md
CHANGED
|
@@ -6,6 +6,58 @@ All notable changes to this project are documented here. Versions follow [Semant
|
|
|
6
6
|
|
|
7
7
|
_No changes yet._
|
|
8
8
|
|
|
9
|
+
## 2.1.0 - 2026-10-18
|
|
10
|
+
|
|
11
|
+
### Studio: n8n-style Wiring canvas (#88)
|
|
12
|
+
|
|
13
|
+
- **Added:** node tiles on the Wiring canvas (type icon, name, one status line that is also the duration slot, no stage or type badge). Edges are cubic Bézier curves; the edge into the RUNNING node animates, and `prefers-reduced-motion` disables that animation. A loop group draws a container labeled `iteration n / max` from live loop state.
|
|
14
|
+
- **Added:** a minimap with one mark per node, zoom controls (+, −, fit, auto-layout, percent), a node side panel with Input, Output, Evidence, and Timeline tabs, a tooltip, and a `?` shortcut sheet (`F`, `L`, `+`, `-`, `0`, arrows, Enter, Esc, undo/redo). `:focus-visible` outlines the controls.
|
|
15
|
+
- **Added:** a dirty-only save bar (change count, Valid or Invalid, Discard, Save) that stays hidden while the wiring is clean. Save still goes through the confirm dialog. A finished run shows a summary card with state, destination, Judge verdict, approver, duration, tests, findings, drift, receipt, and tokens (`unknown` when the ledger has none). Node durations are derived from consecutive node `updatedAt` times.
|
|
16
|
+
- **Added:** `delivery.json` is a watched artifact: `GET /api/events` emits `artifact.delivery` (`destination`, `humanApproved`, `approvedBy`) for the summary card.
|
|
17
|
+
- **Changed:** auto-layout and fit keep every built-in recipe (default, bugfix, refactor, high-risk, docs-light, audit) inside the canvas at 1280×800 and 1600×1000. The edge delete control appears only on hover or selection. Live updates still arrive on `GET /api/events` via `applyEvent`; the client does not poll.
|
|
18
|
+
- **Changed:** workflow structure stays read-only (#80). The canvas has no create, delete, rename, duplicate, or import control for nodes. Rewiring existing edges is still validated by `POST /api/validate`. A backbone-breaking edge shows the compiler's named error and disables Save. Node-set changes still fail with `STUDIO_STRUCTURE_READ_ONLY`, and `POST /api/import/*` stays 404.
|
|
19
|
+
|
|
20
|
+
### README refresh for 2.0.0 (#86)
|
|
21
|
+
|
|
22
|
+
- **Docs:** README Workflow Studio section describes the 2.0 live-run view, real-time `/api/events`, and wiring-only canvas (node creation, duplication, import, and Studio-created Agent nodes are gone). Adds `ael bench` (compliance self-check for your own host and model; slow) and personal workflow (`workflow interview|propose|accept|suggest`, `~/.ai-engineering-loop/workflow/`, `workflow-history.jsonl`, project-required > personal > built-in). Updates the CLI cheat sheet, recipe authoring example, major-update note, and Documentation links.
|
|
23
|
+
|
|
24
|
+
## 2.0.0 - 2026-10-17
|
|
25
|
+
|
|
26
|
+
### Breaking changes
|
|
27
|
+
|
|
28
|
+
- Studio no longer creates or edits nodes (#80). Wiring mode only edits edges between existing nodes. `POST /api/import/preview` and `POST /api/import/apply` return 404. Author recipes as files plus `recipe create|clone|install`.
|
|
29
|
+
- `GET /api/runs/:id/lifecycle/stream` returns 404; subscribe to `GET /api/events` instead (#80).
|
|
30
|
+
- **Upgrade:** run `ai-engineering-loop sync-hosts` once after upgrading so installed host skills match 2.0.0.
|
|
31
|
+
|
|
32
|
+
### Personal workflow onboarding (#81)
|
|
33
|
+
|
|
34
|
+
- **Added:** `workflow interview|propose|accept|reject|suggest`. With no concrete task the skill offers to build a personal recipe, start a task, or show status, and does not create a Run. Interview questions come pre-filled from the last 20 run summaries (80% majority). `propose` compiles a candidate through the recipe safety backbone. `accept --by` is the only way to install it.
|
|
35
|
+
- **Added:** the personal recipe, its version history, proposals, and `workflow-history.jsonl` live under the AEL home (`AEL_HOME` or `~/.ai-engineering-loop`), never in a repository. Each terminal Run appends a summary of counts and a repo hash. Suggestions are proposals only; a rejected proposal stays quiet until new evidence outgrows the old one.
|
|
36
|
+
- **Changed:** `run` resolves project-required (`workflow.requiredRecipe`, `workflow.allowPersonal: false`, required project nodes, `adapter.md`) then the personal recipe, then the built-in default (`audit` for REPORT_ONLY). `run --json` reports `workflowSource`. `run --recipe` cannot bypass the project-required layer. UNATTENDED cannot accept (`HUMAN_REQUIRED`).
|
|
37
|
+
- **Added:** eval case `25-no-task-onboarding`; the always-loaded skill gains a no-task line (offer `workflow interview` or a task; no Run), with details in `loop-commands`.
|
|
38
|
+
- **Docs:** `core/personal-workflow.md` (`docs personal-workflow`).
|
|
39
|
+
|
|
40
|
+
### `ael bench`: compliance self-check for one host and model (#79)
|
|
41
|
+
|
|
42
|
+
- **Added:** `bench [--host <id> --model <id>] [--scenario <id>]... [--repeat <n>] [--bin <path>] [--host-arg <arg>]... [--sandbox-home] [--mode assisted|unattended|report-only] [--timeout <seconds>] [--json]`. One target per invocation. Each scenario runs in a fresh temp repo. `AEL_HOME` is a temp directory. The host process keeps the real `HOME` so it can authenticate; `--bin` and `--sandbox-home` use a temp `HOME` instead. Skills are installed in the temp repo (the directories that host reads), and `node_modules/.bin/ai-engineering-loop` shims this checkout so `npx` does not download the published package. Default timeout is 20 minutes. A missing binary is `UNAVAILABLE`, not a model failure. Default mode is ASSISTED; the scenario's scripted approver answers Goal freeze, impact confirm, and `deliver --approve`. `--mode unattended` is opt-in. Progress lines go to stderr.
|
|
43
|
+
- **Added:** `evals/scenarios/` (T0 docs, T1 features, T2 auth/payment, and traps) and `schemas/bench-scenario.schema.json`. The `evals/cases` judgment suite is unchanged by the bench.
|
|
44
|
+
- **Added:** scoring from Run artifacts only (`state.json`, `guard-events.json`, `impact-drift.json`, `verifyReceipt()`, the usage ledger, the harness clock). Tokens are parsed from host JSON when the host provides it (Claude Code and Codex). Grok CLI `-p` is plain text, so its tokens stay unknown, never `0`. `budget record --unavailable` is the same unknown. Bench-only host templates: Grok `--no-auto-update --always-approve`, Claude `--dangerously-skip-permissions`, Codex `--full-auto`. `--host-arg` appends extra flags. `--bin` overrides the binary.
|
|
45
|
+
- **Added:** `.ai-engineering-loop/bench/<benchId>/{results.json,report.md,hardening.md}` (`schemas/bench-result.schema.json`). The report is per-gate compliance, guard violations, scope drift, receipt validity, and p50/p90 time and tokens for that single target. `hardening.md` maps each violation to a `skill-src/` instruction, ranked by frequency, with a text-tightening or hard-gate suggestion. Suggestions are not applied. There is no routing or model recommendation.
|
|
46
|
+
- **Added:** `bench compare <benchIdA> <benchIdB>` prints a read-only side-by-side table.
|
|
47
|
+
- **Docs:** `core/bench.md` (`docs bench`).
|
|
48
|
+
|
|
49
|
+
### Studio: real-time sync and compact live-run UI (#80)
|
|
50
|
+
|
|
51
|
+
- **Added:** `lib/run-watcher.js` recursively watches `.ai-engineering-loop/runs/` and `current.json` (debounce ~100 ms) and pushes typed events on `GET /api/events` with monotonic `seq`. Event types: `run.created`, `run.state`, `lifecycle`, `artifact.impact|drift|guard|verification|findings|verdict|receipt`, `current.changed`, `presence`. `hello` reports `watch` or `poll` plus the watched root. `Last-Event-ID` / `?after=<seq>` replay from an in-memory ring; a gap sends `resync` for a full `/api/bootstrap`. `AEL_STUDIO_FS_WATCH=0` (and a failed `fs.watch`) uses a 1 s stat-poll.
|
|
52
|
+
- **Added:** compact live-run view as the default route (stage rail, heartbeat age, tier/mode/iteration/budget, guard/drift/receipt chips, pending-question banner, merged timeline). The Goal sheet and run-history list remain. Wiring mode edits edges between existing nodes only; `/api/validate` is still `lib/recipe.js`, and install still requires the confirm dialog.
|
|
53
|
+
- **Added:** watched-root display. `status --json` and `run --json` include `studioUrl` (tokenless Studio identity; the browser session comes from the URL printed by `ael studio`) when a Studio process for that root is alive (`.ai-engineering-loop/studio.json`).
|
|
54
|
+
- **Added:** `/api/install` requires `baseRecipeId` and rejects node-set changes (`STUDIO_STRUCTURE_READ_ONLY`); `/api/validate` reports the same code when `baseRecipeId` is sent. Only edges, `dependsOn`, and layout may differ from the loaded base recipe, including a builtin installed as `<id>-local`.
|
|
55
|
+
- **Added:** the Studio client reconnects to `/api/events` with `setTimeout` backoff and `Last-Event-ID` / `?after=`; a `resync` event or failed resume falls back to `/api/bootstrap`. The Retry button remains.
|
|
56
|
+
- **Breaking:** removed the 1.8 s `refreshLive` client poll and the 750 ms per-client lifecycle interval. `GET /api/runs/:id/lifecycle/stream` returns 404; use `GET /api/events`.
|
|
57
|
+
- **Breaking:** removed node creation, duplication, deletion, inspector field edits, loop-group creation, Custom Agent dialog, Import workflow JSON (preview/merge/replace), and the export/copy-brief toolbar. `POST /api/import/preview` and `POST /api/import/apply` return 404. Author recipes as files + `recipe create|clone|install`.
|
|
58
|
+
- **Changed:** artifact SSE summaries match real writers: `verification.json` derives `passed`/`failed` from the command bundle (no `status`); findings `blocking` uses `isBlockingFinding`; `receipt.json` verification is `verifyReceipt` (`present`, `digest`, `verified`, `problems`).
|
|
59
|
+
- **Changed:** RunWatcher processes only `current.json` and paths under `runs/`; `usage/`, `learnings.json`, drafts, and other context files emit no events.
|
|
60
|
+
|
|
9
61
|
## 1.17.0 - 2026-10-17
|
|
10
62
|
|
|
11
63
|
### `ael learn`: past Runs become project learnings for context packs (#61)
|
package/README.md
CHANGED
|
@@ -4,7 +4,7 @@
|
|
|
4
4
|
|
|
5
5
|
### From idea to trusted code—without babysitting every command.
|
|
6
6
|
|
|
7
|
-
Give your AI agent a goal. It plans the work, builds it, checks itself, challenges the result, and prepares a safe delivery. You get a
|
|
7
|
+
Give your AI agent a goal. It plans the work, builds it, checks itself, challenges the result, and prepares a safe delivery. You get a live view of the run and a clear history instead of a wall of terminal output.
|
|
8
8
|
|
|
9
9
|
[](https://www.npmjs.com/package/ai-engineering-loop)
|
|
10
10
|
[](https://github.com/egagofur/ai-engineering-loop/releases)
|
|
@@ -51,7 +51,7 @@ Want the visual workspace?
|
|
|
51
51
|
npx ai-engineering-loop studio
|
|
52
52
|
```
|
|
53
53
|
|
|
54
|
-
Workflow Studio opens locally in your browser. Your project and run evidence stay on your machine.
|
|
54
|
+
Workflow Studio opens locally in your browser and follows the active Run live. Your project and run evidence stay on your machine.
|
|
55
55
|
|
|
56
56
|
On first setup, `.ai-engineering-loop/` is added to your project’s `.gitignore` so plans and Run evidence stay private by default. Want to share that context with your team? Simply remove that line from `.gitignore`.
|
|
57
57
|
|
|
@@ -87,7 +87,7 @@ npx ai-engineering-loop projects list # projects refreshed after an upda
|
|
|
87
87
|
npx ai-engineering-loop update rollback --yes
|
|
88
88
|
```
|
|
89
89
|
|
|
90
|
-
Global installs apply patch and minor releases automatically. They then re-sync host skills and refresh every registered project, keeping your edits (new templates arrive as `*.ael-new`). Project installs only show a short notice (current -> latest, a few changelog highlights, the exact `npm`/`pnpm`/`yarn`/`bun` command, how to silence it), at most once a day per project, because AEL never edits your `package.json`. Projects from versions before hash tracking are overwritten on their first update, with the old files kept in `.ai-engineering-loop/.ael-backup/` and restored by `update rollback`. Major updates run `migrate --from <old>` after install. Opt out with `AEL_NO_AUTO_UPDATE=1`; CI is always off. There is no daemon, so other machines pick up an update the next time the CLI runs there.
|
|
90
|
+
Global installs apply patch and minor releases automatically. They then re-sync host skills and refresh every registered project, keeping your edits (new templates arrive as `*.ael-new`). Project installs only show a short notice (current -> latest, a few changelog highlights, the exact `npm`/`pnpm`/`yarn`/`bun` command, how to silence it), at most once a day per project, because AEL never edits your `package.json`. Projects from versions before hash tracking are overwritten on their first update, with the old files kept in `.ai-engineering-loop/.ael-backup/` and restored by `update rollback`. Major updates run `migrate --from <old>` after install. A major release such as 2.0.0 only shows a notice: install it yourself, then run `ai-engineering-loop sync-hosts` once so your agents get the new skill. Opt out with `AEL_NO_AUTO_UPDATE=1`; CI is always off. There is no daemon, so other machines pick up an update the next time the CLI runs there.
|
|
91
91
|
|
|
92
92
|
Projects with an unfinished run are skipped until that run ends. If a run was simply left behind, close it (the run is kept, marked `CANCELLED`, and `current.json` is released):
|
|
93
93
|
|
|
@@ -102,18 +102,18 @@ When context packs exceed their budget, the CLI now reports the limit and a per-
|
|
|
102
102
|
|
|
103
103
|
## Meet Workflow Studio
|
|
104
104
|
|
|
105
|
-
Workflow Studio makes agent work feel like a product, not a terminal session.
|
|
105
|
+
Workflow Studio makes agent work feel like a product, not a terminal session. It is a live window on your Runs, not a second place to author workflows.
|
|
106
106
|
|
|
107
|
-
###
|
|
107
|
+
### Watch the Run as it happens
|
|
108
108
|
|
|
109
|
-
-
|
|
110
|
-
-
|
|
111
|
-
-
|
|
112
|
-
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
-
|
|
109
|
+
- A compact live view: stage rail, heartbeat, tier, mode, iteration, and budget
|
|
110
|
+
- Compliance chips for guard warnings, scope drift, and receipt verification
|
|
111
|
+
- Pending questions and approvals in a banner at the top
|
|
112
|
+
- Real-time updates pushed over `GET /api/events` (no polling; it falls back to a light stat-poll where file watching is unavailable)
|
|
113
|
+
|
|
114
|
+
### Rewire, safely
|
|
115
|
+
|
|
116
|
+
Workflow structure is read-only in Studio. **Wiring** mode lets you add, remove, or re-point connections between nodes that already exist. Every change goes through the same recipe compiler as the CLI, so an edge set that breaks the safety backbone cannot be saved. To add or change nodes, edit recipe files and use `recipe create|clone|install`.
|
|
117
117
|
|
|
118
118
|
### Start with a Goal—not a vague prompt
|
|
119
119
|
|
|
@@ -121,33 +121,19 @@ Write the task in plain language, add the outcomes that matter, and **Freeze** t
|
|
|
121
121
|
|
|
122
122
|
### Follow every Run
|
|
123
123
|
|
|
124
|
-
Runs have human-friendly names instead of cryptic hashes. Search previous work,
|
|
125
|
-
|
|
126
|
-
Select a node to inspect its real **Input**, **Output**, **Evidence**, and timeline. The canvas shows the actual status of every step, so completed work, failures, approvals, and waiting nodes are easy to understand.
|
|
124
|
+
Runs have human-friendly names instead of cryptic hashes. Search previous work, pin any Run as a read-only replay, and see exactly where it stopped. A merged timeline shows lifecycle, activity, and evidence events in order.
|
|
127
125
|
|
|
128
126
|
### Continue with your AI agent
|
|
129
127
|
|
|
130
|
-
Use **
|
|
128
|
+
Use **Start with AI Agent** to copy a ready-to-use handoff for the selected Run. The agent continues with the same Goal, workflow, and evidence instead of starting from scratch.
|
|
131
129
|
|
|
132
130
|
If the agent needs clarification, its questions can appear inside Workflow Studio. You answer them there, and the answer stays attached to the Run.
|
|
133
131
|
|
|
134
132
|
### Repeat only the steps that need another pass
|
|
135
133
|
|
|
136
|
-
|
|
137
|
-
|
|
138
|
-
- Select the steps that belong together
|
|
139
|
-
- Choose where the loop starts and where the decision happens
|
|
140
|
-
- Give repeat and continue paths clear labels such as `NO · RETRY` and `YES · CONTINUE`
|
|
141
|
-
- Set a maximum number of passes
|
|
142
|
-
- Follow a separate `Iteration 2 / 3` counter for each group
|
|
143
|
-
|
|
144
|
-
When a loop reaches its limit, Workflow Studio pauses for approval. You can allow one focused extra pass, continue with the current evidence, or stop the Run safely. Nested and overlapping loops stay disabled so the workflow remains easy to read.
|
|
145
|
-
|
|
146
|
-
### Execute with confidence
|
|
134
|
+
Recipes can declare independent Loop Groups with a maximum number of passes. Studio shows a separate `Iteration 2 / 3` counter for each group, and a loop that reaches its limit pauses for a human decision instead of retrying forever.
|
|
147
135
|
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
> **New in v1.9.0:** Run Checkout, AI Agent handoff, live Run questions, and Multiple Loop Groups. [See the release](https://github.com/egagofur/ai-engineering-loop/releases/tag/v1.9.0).
|
|
136
|
+
> **New in v2.0.0:** real-time Studio, wiring-only canvas, `ael bench`, and personal workflows. Studio node creation and import were removed. [See the release](https://github.com/egagofur/ai-engineering-loop/releases/tag/v2.0.0).
|
|
151
137
|
|
|
152
138
|
## One workspace from idea to delivery
|
|
153
139
|
|
|
@@ -231,16 +217,38 @@ Specs such as the grill, TDD, and runtime-safety policies ship inside the packag
|
|
|
231
217
|
|
|
232
218
|
## Create workflows that fit your team
|
|
233
219
|
|
|
234
|
-
Start from a built-in
|
|
220
|
+
Start from a built-in recipe (`default`, `bugfix`, `refactor`, `high-risk`, `docs-light`, `audit`) or author your own as a file:
|
|
221
|
+
|
|
222
|
+
```bash
|
|
223
|
+
npx ai-engineering-loop recipe clone bugfix team-bugfix # writes .ai-engineering-loop/recipes/team-bugfix.json
|
|
224
|
+
npx ai-engineering-loop recipe validate team-bugfix
|
|
225
|
+
```
|
|
226
|
+
|
|
227
|
+
Recipes can include prompt-powered Agent nodes. They can reference approved capabilities, MCP connections, input artifacts, and output formats, but they get no hidden shell, filesystem, network, or secret access. Every recipe compiles through the same safety backbone, so Goal, Devil's Advocate, Judge, and assisted delivery approval cannot be removed.
|
|
235
228
|
|
|
236
|
-
|
|
237
|
-
- Bug fix
|
|
238
|
-
- Security review
|
|
239
|
-
- Repository audit
|
|
240
|
-
- Refactor
|
|
241
|
-
- Custom prompt-powered Agent workflow
|
|
229
|
+
### Your personal workflow
|
|
242
230
|
|
|
243
|
-
|
|
231
|
+
Your own way of working can follow you across repositories. Call AEL without a task and it offers to build it with a short interview, pre-filled from your past Runs:
|
|
232
|
+
|
|
233
|
+
```bash
|
|
234
|
+
npx ai-engineering-loop workflow interview # questions, each with the evidence behind its default
|
|
235
|
+
npx ai-engineering-loop workflow propose --answers answers.json
|
|
236
|
+
npx ai-engineering-loop workflow accept --proposal <id> --by <name>
|
|
237
|
+
npx ai-engineering-loop workflow suggest # proposals from your run history; never applied silently
|
|
238
|
+
```
|
|
239
|
+
|
|
240
|
+
It lives in `~/.ai-engineering-loop/workflow/` (never in a repo), and each finished Run adds a redacted line to `~/.ai-engineering-loop/workflow-history.jsonl`. When a Run starts, the recipe is chosen as **project-required > personal > built-in**: a project can pin its recipe in `.ai-engineering-loop/runtime-policy.json`, and `run --json` reports which one won (`workflowSource`). See [Personal workflow](core/personal-workflow.md).
|
|
241
|
+
|
|
242
|
+
### Check your agent stays compliant
|
|
243
|
+
|
|
244
|
+
`ael bench` runs scripted scenarios (simple, feature, and high-risk tasks, plus traps like destructive commands and prompt injection) against **the host and model you already use**, each in a throwaway repo:
|
|
245
|
+
|
|
246
|
+
```bash
|
|
247
|
+
npx ai-engineering-loop bench --host grok --model grok-4.7
|
|
248
|
+
npx ai-engineering-loop bench compare <benchIdA> <benchIdB>
|
|
249
|
+
```
|
|
250
|
+
|
|
251
|
+
It scores every gate from real Run artifacts and writes a report plus a `hardening.md` that links each violation to the skill instruction it broke. It does not route work or recommend a model. Expect it to be slow: one T0 scenario took about 18 minutes on grok-4.7 (high). See [Bench](core/bench.md).
|
|
244
252
|
|
|
245
253
|
## Safety without the scary dashboard
|
|
246
254
|
|
|
@@ -306,6 +314,11 @@ npx ai-engineering-loop update apply --yes
|
|
|
306
314
|
npx ai-engineering-loop run "describe the task"
|
|
307
315
|
npx ai-engineering-loop studio
|
|
308
316
|
|
|
317
|
+
# Personal workflow and compliance self-check
|
|
318
|
+
npx ai-engineering-loop workflow interview
|
|
319
|
+
npx ai-engineering-loop workflow suggest
|
|
320
|
+
npx ai-engineering-loop bench --host <id> --model <id>
|
|
321
|
+
|
|
309
322
|
# Explore workflows
|
|
310
323
|
npx ai-engineering-loop recipe list
|
|
311
324
|
npx ai-engineering-loop recipe simulate default --mode assisted
|
|
@@ -351,6 +364,8 @@ No hidden cloud control plane is required. No project source is uploaded by Work
|
|
|
351
364
|
|
|
352
365
|
- [Workflow Studio](core/local-workflow-studio.md)
|
|
353
366
|
- [Workflow recipes](core/workflow-recipes.md)
|
|
367
|
+
- [Personal workflow](core/personal-workflow.md)
|
|
368
|
+
- [Bench: compliance self-check](core/bench.md)
|
|
354
369
|
- [Runtime safety](core/runtime-safety.md)
|
|
355
370
|
- [Verification](core/verification-loop.md)
|
|
356
371
|
- [Controlled workflow runtime](core/controlled-workflow-runtime.md)
|