@hecer/yoke 1.11.0 → 1.12.0
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/.claude-plugin/plugin.json +13 -13
- package/.codex-plugin/plugin.json +7 -7
- package/CHANGELOG.md +416 -398
- package/README.md +931 -915
- package/TODOS.md +5 -5
- package/agents/docs.toml +6 -6
- package/agents/implementer.toml +6 -6
- package/agents/reviewer.toml +6 -6
- package/agents/security.toml +6 -6
- package/bench/README.md +86 -86
- package/bench/RESULTS.md +35 -35
- package/bench/output-compaction.mjs +65 -65
- package/bench/result-schema.mjs +12 -12
- package/bench/results/claude-2026-07-27T18-03-26.json +50 -50
- package/bench/results/codex-unavailable-1785175418318.json +15 -15
- package/bench/results/gemini-2026-07-27T18-03-44.json +46 -46
- package/bench/run-matrix.mjs +26 -26
- package/bench/run.mjs +106 -106
- package/canon/AGENTS.md +30 -30
- package/canon/context/DECISIONS.md +4 -4
- package/canon/context/GLOSSARY.md +11 -11
- package/canon/context/KNOWLEDGE.md +4 -4
- package/canon/context/PROJECT.md +15 -15
- package/canon/loop/loop-spec.md +65 -65
- package/canon/loop/prd.schema.md +41 -41
- package/canon/manifest.yaml +59 -59
- package/canon/policy/gates.md +7 -7
- package/canon/policy/roles.md +9 -9
- package/canon/skills/ATTRIBUTION.md +99 -99
- package/canon/skills/authoring-prd/SKILL.md +56 -56
- package/canon/skills/brainstorming/SKILL.md +164 -164
- package/canon/skills/codebase-design/DEEPENING.md +15 -15
- package/canon/skills/codebase-design/DESIGN-IT-TWICE.md +12 -12
- package/canon/skills/codebase-design/SKILL.md +39 -39
- package/canon/skills/dispatching-parallel-agents/SKILL.md +182 -182
- package/canon/skills/document-release/SKILL.md +302 -302
- package/canon/skills/domain-modeling/ADR-FORMAT.md +19 -19
- package/canon/skills/domain-modeling/CONTEXT-FORMAT.md +39 -39
- package/canon/skills/domain-modeling/SKILL.md +35 -35
- package/canon/skills/executing-plans/SKILL.md +70 -70
- package/canon/skills/finishing-a-development-branch/SKILL.md +200 -200
- package/canon/skills/health/SKILL.md +177 -177
- package/canon/skills/maintaining-context/SKILL.md +34 -34
- package/canon/skills/minimal-code/SKILL.md +21 -21
- package/canon/skills/no-ai-slop/SKILL.md +103 -103
- package/canon/skills/no-ai-slop/eval.md +43 -43
- package/canon/skills/plan-ceo-review/SKILL.md +541 -541
- package/canon/skills/plan-eng-review/SKILL.md +362 -362
- package/canon/skills/receiving-code-review/SKILL.md +213 -213
- package/canon/skills/requesting-code-review/SKILL.md +105 -105
- package/canon/skills/resolving-merge-conflicts/SKILL.md +18 -18
- package/canon/skills/retro/SKILL.md +397 -397
- package/canon/skills/review/SKILL.md +246 -246
- package/canon/skills/ship/SKILL.md +691 -691
- package/canon/skills/subagent-driven-development/SKILL.md +277 -277
- package/canon/skills/systematic-debugging/SKILL.md +296 -296
- package/canon/skills/tdd/SKILL.md +371 -371
- package/canon/skills/unslop-ui/SKILL.md +34 -34
- package/canon/skills/using-git-worktrees/SKILL.md +218 -218
- package/canon/skills/verification-before-completion/SKILL.md +139 -139
- package/canon/skills/visual-verification/SKILL.md +54 -54
- package/canon/skills/workflow/SKILL.md +22 -22
- package/canon/skills/writing-for-agents/SKILL-MECHANICS.md +27 -27
- package/canon/skills/writing-for-agents/SKILL.md +42 -42
- package/canon/skills/writing-plans/SKILL.md +152 -152
- package/canon/skills/writing-skills/SKILL.md +655 -655
- package/canon/skills/yoke-retrofit/SKILL.md +26 -26
- package/canon/skills/yoke-workflow/SKILL.md +20 -20
- package/canon/tools/codex-rtk-hook.mjs +35 -35
- package/canon/tools/gemini-rtk-hook.mjs +25 -25
- package/canon/tools/graphify.md +3 -3
- package/canon/tools/playwright-mcp.md +3 -3
- package/canon/tools/qwen-rtk-hook.mjs +25 -0
- package/canon/tools/rtk.md +7 -7
- package/canon/tools/serena.md +6 -6
- package/dist/agents/host.js +1 -1
- package/dist/agents/providers.js +18 -5
- package/dist/agents/telemetry.js +35 -36
- package/dist/cli.js +18 -10
- package/dist/dashboard/page.js +122 -122
- package/dist/dashboard/panels.js +91 -91
- package/dist/loop/run-command.js +3 -3
- package/dist/prd/command.js +17 -17
- package/dist/retrofit/apply.js +8 -1
- package/dist/retrofit/config.js +1 -1
- package/dist/retrofit/detect.js +2 -0
- package/dist/retrofit/planners/claude.js +14 -14
- package/dist/retrofit/planners/qwen.js +3 -3
- package/dist/retrofit/preserve.js +2 -2
- package/dist/retrofit/qwen-settings.js +17 -0
- package/dist/retrofit/skill-actions.js +1 -1
- package/dist/setup/command.js +22 -8
- package/dist/setup/model-presets.js +48 -0
- package/docs/CAPABILITY-ROUTING.md +51 -51
- package/docs/DASHBOARD-EVOLUTION.md +33 -33
- package/docs/MIGRATING-TO-1.0.md +33 -33
- package/docs/MIGRATING-TO-1.1.md +27 -27
- package/docs/MIGRATING-TO-1.4.md +70 -70
- package/docs/PRODUCT-DIRECTION-2026-09-05.md +210 -210
- package/docs/PUBLISHING.md +114 -114
- package/docs/QWEN-MODEL-SUPPORT.md +142 -0
- package/docs/VERIFIED-PROJECTS-VALIDATION.md +29 -29
- package/docs/VERIFIED-PROJECTS.md +167 -167
- package/docs/superpowers/plans/2026-06-28-baustein-e-context-layer.md +981 -981
- package/docs/superpowers/plans/2026-06-29-baustein-f-routing.md +258 -258
- package/docs/superpowers/plans/2026-06-29-baustein-g-loop-observability.md +1006 -1006
- package/docs/superpowers/plans/2026-06-29-baustein-h-loop-robustness.md +374 -374
- package/docs/superpowers/plans/2026-06-30-baustein-i-visual-design-verification.md +450 -450
- package/docs/superpowers/plans/2026-07-02-baustein-k-zero-to-100-bootstrap.md +1024 -1024
- package/docs/superpowers/plans/2026-07-02-baustein-m-flow-smoke-proofs.md +574 -574
- package/docs/superpowers/plans/2026-08-13-gauntlet-quality-loop.md +537 -537
- package/docs/superpowers/plans/2026-08-16-artifact-backed-output-compaction.md +329 -329
- package/docs/superpowers/plans/2026-09-05-verified-projects.md +83 -83
- package/docs/superpowers/specs/2026-06-28-baustein-e-context-layer-design.md +146 -146
- package/docs/superpowers/specs/2026-06-29-baustein-f-routing-design.md +106 -106
- package/docs/superpowers/specs/2026-06-29-baustein-g-loop-observability-design.md +186 -186
- package/docs/superpowers/specs/2026-06-29-baustein-h-loop-robustness-design.md +113 -113
- package/docs/superpowers/specs/2026-06-30-baustein-i-visual-design-verification-design.md +98 -98
- package/docs/superpowers/specs/2026-07-02-baustein-k-zero-to-100-bootstrap-design.md +200 -200
- package/docs/superpowers/specs/2026-07-02-baustein-m-flow-smoke-proofs-design.md +155 -155
- package/docs/superpowers/specs/2026-08-13-gauntlet-quality-loop-design.md +422 -422
- package/docs/superpowers/specs/2026-08-16-artifact-backed-output-compaction-design.md +166 -166
- package/gemini-extension.json +6 -6
- package/hooks/hooks.json +19 -19
- package/package.json +87 -87
- package/dist/dashboard/discovery.js +0 -73
- package/docs/community-outreach-2026-08-20.md +0 -85
- package/docs/launch-copy-2026-08-21.md +0 -193
package/docs/MIGRATING-TO-1.1.md
CHANGED
|
@@ -1,27 +1,27 @@
|
|
|
1
|
-
# Migrating to Yoke 1.1
|
|
2
|
-
|
|
3
|
-
Yoke 1.1 makes Claude, Codex, and Gemini share the same setup, planning, runner-selection, and decision behavior.
|
|
4
|
-
|
|
5
|
-
Run the setup wizard once in an existing project:
|
|
6
|
-
|
|
7
|
-
```bash
|
|
8
|
-
npx @hecer/yoke@1.1.0 setup .
|
|
9
|
-
```
|
|
10
|
-
|
|
11
|
-
It preserves existing Yoke configuration and asks for target agents, code graph, loop state, default runner, and decision mode. A non-interactive agent can apply explicit choices with `--yes`.
|
|
12
|
-
|
|
13
|
-
The new config fields are optional and backward-compatible:
|
|
14
|
-
|
|
15
|
-
```yaml
|
|
16
|
-
runner:
|
|
17
|
-
agent: codex
|
|
18
|
-
permissions: safe
|
|
19
|
-
loop:
|
|
20
|
-
enabled: true
|
|
21
|
-
timeoutMinutes: 30
|
|
22
|
-
decisionPolicy: critical # auto | critical
|
|
23
|
-
```
|
|
24
|
-
|
|
25
|
-
`loop.onAmbiguity: resolve|abort` and `--on-ambiguity=` still work. Prefer `decisionPolicy` for new projects: `auto` resolves routine choices without asking; `critical` pauses only for high-impact decisions and continues after `yoke loop answer`. Resume restores the original isolation, review, runner, permissions, timeout, and policy flags instead of silently weakening the run. If the restart cannot begin, `yoke loop resume` retries from the request-bound state stored under Git's private state directory.
|
|
26
|
-
|
|
27
|
-
Restart an already-open Codex task after retrofit if it does not discover the newly generated `.agents/skills/yoke-workflow/SKILL.md`.
|
|
1
|
+
# Migrating to Yoke 1.1
|
|
2
|
+
|
|
3
|
+
Yoke 1.1 makes Claude, Codex, and Gemini share the same setup, planning, runner-selection, and decision behavior.
|
|
4
|
+
|
|
5
|
+
Run the setup wizard once in an existing project:
|
|
6
|
+
|
|
7
|
+
```bash
|
|
8
|
+
npx @hecer/yoke@1.1.0 setup .
|
|
9
|
+
```
|
|
10
|
+
|
|
11
|
+
It preserves existing Yoke configuration and asks for target agents, code graph, loop state, default runner, and decision mode. A non-interactive agent can apply explicit choices with `--yes`.
|
|
12
|
+
|
|
13
|
+
The new config fields are optional and backward-compatible:
|
|
14
|
+
|
|
15
|
+
```yaml
|
|
16
|
+
runner:
|
|
17
|
+
agent: codex
|
|
18
|
+
permissions: safe
|
|
19
|
+
loop:
|
|
20
|
+
enabled: true
|
|
21
|
+
timeoutMinutes: 30
|
|
22
|
+
decisionPolicy: critical # auto | critical
|
|
23
|
+
```
|
|
24
|
+
|
|
25
|
+
`loop.onAmbiguity: resolve|abort` and `--on-ambiguity=` still work. Prefer `decisionPolicy` for new projects: `auto` resolves routine choices without asking; `critical` pauses only for high-impact decisions and continues after `yoke loop answer`. Resume restores the original isolation, review, runner, permissions, timeout, and policy flags instead of silently weakening the run. If the restart cannot begin, `yoke loop resume` retries from the request-bound state stored under Git's private state directory.
|
|
26
|
+
|
|
27
|
+
Restart an already-open Codex task after retrofit if it does not discover the newly generated `.agents/skills/yoke-workflow/SKILL.md`.
|
package/docs/MIGRATING-TO-1.4.md
CHANGED
|
@@ -1,70 +1,70 @@
|
|
|
1
|
-
# Migrating to Yoke 1.4
|
|
2
|
-
|
|
3
|
-
Yoke 1.4 adds dependency-aware parallel workers, isolated candidate selection, and a reference-driven quality gauntlet. Existing projects remain serial and skip quality comparison unless you opt in.
|
|
4
|
-
|
|
5
|
-
Upgrade and refresh generated harness files:
|
|
6
|
-
|
|
7
|
-
```bash
|
|
8
|
-
npm install -g @hecer/yoke@1.4.0
|
|
9
|
-
yoke retrofit .
|
|
10
|
-
```
|
|
11
|
-
|
|
12
|
-
## Parallel execution
|
|
13
|
-
|
|
14
|
-
Run dependency-ready stories concurrently with an explicit worker limit:
|
|
15
|
-
|
|
16
|
-
```bash
|
|
17
|
-
yoke loop run . --isolate --parallel=3
|
|
18
|
-
```
|
|
19
|
-
|
|
20
|
-
Parallel runs use isolated worktrees, leased claims, and a FIFO integration queue. Every candidate must pass its worker gates and the merged result must pass the project gates again. Adaptive routing is disabled during parallel and multi-candidate runs so each worker has one auditable provider identity.
|
|
21
|
-
|
|
22
|
-
Use `--candidates=N` to produce and mechanically gate multiple implementations before an identity-blind comparison selects the winner. Candidate mode supports up to five candidates.
|
|
23
|
-
|
|
24
|
-
## Reference-driven quality
|
|
25
|
-
|
|
26
|
-
Quality remains disabled by default. A story must declare a trusted reference, candidate artifacts, and a rubric:
|
|
27
|
-
|
|
28
|
-
```yaml
|
|
29
|
-
quality:
|
|
30
|
-
reference: { name: approved-home, source: design/home.png, kind: file }
|
|
31
|
-
candidate: { kind: screenshots, paths: [.yoke/proof/STORY-1/home.png] }
|
|
32
|
-
rubric: Match the approved layout, hierarchy, spacing, and states.
|
|
33
|
-
policy: blocking
|
|
34
|
-
```
|
|
35
|
-
|
|
36
|
-
Project defaults can bound critic and repair work:
|
|
37
|
-
|
|
38
|
-
```yaml
|
|
39
|
-
quality:
|
|
40
|
-
enabled: false
|
|
41
|
-
policy: blocking
|
|
42
|
-
maxRounds: 3
|
|
43
|
-
maxMinutes: 60
|
|
44
|
-
consistencyChecks: 2
|
|
45
|
-
maxParallelCandidates: 2
|
|
46
|
-
critic: { agent: codex, model: gpt-5.6-sol } # model required for --candidates
|
|
47
|
-
repair: { agent: claude }
|
|
48
|
-
```
|
|
49
|
-
|
|
50
|
-
Enable it for one run with:
|
|
51
|
-
|
|
52
|
-
```bash
|
|
53
|
-
yoke loop run . --quality
|
|
54
|
-
```
|
|
55
|
-
|
|
56
|
-
`consistencyChecks` is fixed at `2`: Yoke runs the swapped-label pair needed to detect identity-sensitive critic output. Advisory mode records both verdicts without blocking; blocking mode permits bounded repairs and reruns all mechanical gates.
|
|
57
|
-
|
|
58
|
-
## Cleanup behavior
|
|
59
|
-
|
|
60
|
-
`yoke loop cleanup` now retains Yoke-created worktrees by default so failed candidates remain inspectable. Remove them explicitly when no longer needed:
|
|
61
|
-
|
|
62
|
-
```bash
|
|
63
|
-
yoke loop cleanup . --remove-worktrees
|
|
64
|
-
```
|
|
65
|
-
|
|
66
|
-
Cleanup still reaps only provider processes recorded for the current project and removes stale loop locks. It never kills providers by machine-wide process name.
|
|
67
|
-
|
|
68
|
-
## Review verdicts
|
|
69
|
-
|
|
70
|
-
Review verdict files now require `schemaVersion: 1` and provenance containing the provider, provider-reported model, review role, prompt version, and permission profile. Custom reviewer integrations must emit the contract printed in the reviewer prompt. Legacy verdict files without this envelope fail closed.
|
|
1
|
+
# Migrating to Yoke 1.4
|
|
2
|
+
|
|
3
|
+
Yoke 1.4 adds dependency-aware parallel workers, isolated candidate selection, and a reference-driven quality gauntlet. Existing projects remain serial and skip quality comparison unless you opt in.
|
|
4
|
+
|
|
5
|
+
Upgrade and refresh generated harness files:
|
|
6
|
+
|
|
7
|
+
```bash
|
|
8
|
+
npm install -g @hecer/yoke@1.4.0
|
|
9
|
+
yoke retrofit .
|
|
10
|
+
```
|
|
11
|
+
|
|
12
|
+
## Parallel execution
|
|
13
|
+
|
|
14
|
+
Run dependency-ready stories concurrently with an explicit worker limit:
|
|
15
|
+
|
|
16
|
+
```bash
|
|
17
|
+
yoke loop run . --isolate --parallel=3
|
|
18
|
+
```
|
|
19
|
+
|
|
20
|
+
Parallel runs use isolated worktrees, leased claims, and a FIFO integration queue. Every candidate must pass its worker gates and the merged result must pass the project gates again. Adaptive routing is disabled during parallel and multi-candidate runs so each worker has one auditable provider identity.
|
|
21
|
+
|
|
22
|
+
Use `--candidates=N` to produce and mechanically gate multiple implementations before an identity-blind comparison selects the winner. Candidate mode supports up to five candidates.
|
|
23
|
+
|
|
24
|
+
## Reference-driven quality
|
|
25
|
+
|
|
26
|
+
Quality remains disabled by default. A story must declare a trusted reference, candidate artifacts, and a rubric:
|
|
27
|
+
|
|
28
|
+
```yaml
|
|
29
|
+
quality:
|
|
30
|
+
reference: { name: approved-home, source: design/home.png, kind: file }
|
|
31
|
+
candidate: { kind: screenshots, paths: [.yoke/proof/STORY-1/home.png] }
|
|
32
|
+
rubric: Match the approved layout, hierarchy, spacing, and states.
|
|
33
|
+
policy: blocking
|
|
34
|
+
```
|
|
35
|
+
|
|
36
|
+
Project defaults can bound critic and repair work:
|
|
37
|
+
|
|
38
|
+
```yaml
|
|
39
|
+
quality:
|
|
40
|
+
enabled: false
|
|
41
|
+
policy: blocking
|
|
42
|
+
maxRounds: 3
|
|
43
|
+
maxMinutes: 60
|
|
44
|
+
consistencyChecks: 2
|
|
45
|
+
maxParallelCandidates: 2
|
|
46
|
+
critic: { agent: codex, model: gpt-5.6-sol } # model required for --candidates
|
|
47
|
+
repair: { agent: claude }
|
|
48
|
+
```
|
|
49
|
+
|
|
50
|
+
Enable it for one run with:
|
|
51
|
+
|
|
52
|
+
```bash
|
|
53
|
+
yoke loop run . --quality
|
|
54
|
+
```
|
|
55
|
+
|
|
56
|
+
`consistencyChecks` is fixed at `2`: Yoke runs the swapped-label pair needed to detect identity-sensitive critic output. Advisory mode records both verdicts without blocking; blocking mode permits bounded repairs and reruns all mechanical gates.
|
|
57
|
+
|
|
58
|
+
## Cleanup behavior
|
|
59
|
+
|
|
60
|
+
`yoke loop cleanup` now retains Yoke-created worktrees by default so failed candidates remain inspectable. Remove them explicitly when no longer needed:
|
|
61
|
+
|
|
62
|
+
```bash
|
|
63
|
+
yoke loop cleanup . --remove-worktrees
|
|
64
|
+
```
|
|
65
|
+
|
|
66
|
+
Cleanup still reaps only provider processes recorded for the current project and removes stale loop locks. It never kills providers by machine-wide process name.
|
|
67
|
+
|
|
68
|
+
## Review verdicts
|
|
69
|
+
|
|
70
|
+
Review verdict files now require `schemaVersion: 1` and provenance containing the provider, provider-reported model, review role, prompt version, and permission profile. Custom reviewer integrations must emit the contract printed in the reviewer prompt. Legacy verdict files without this envelope fail closed.
|