@mrciphersmith/keryx 0.3.111 → 0.3.113
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/README.md +17 -3
- package/THIRD_PARTY_NOTICES.md +56 -0
- package/dist/cli.js +102938 -97144
- package/dist/core.js +6072 -4676
- package/package.json +3 -1
- package/src/gdgraph/affected-report.ts +14 -4
- package/src/gdgraph/build.ts +12 -9
- package/src/gdgraph/core-sources.ts +1 -0
- package/src/gdgraph/data-dir.ts +14 -0
- package/src/gdgraph/enrich.ts +5 -3
- package/src/gdgraph/query.ts +3 -2
- package/src/gdgraph/staleness.ts +8 -1
- package/src/gdgraph/wiki-layer.ts +9 -6
- package/src/gdskills/bundled/skills/orchestration/flow-orchestrator/SKILL.md +3 -3
- package/src/gdskills/bundled/skills/orchestration/job-orchestrator/SKILL.md +3 -3
- package/src/gdskills/bundled/skills/orchestration/job-orchestrator/orchestrator-prompt.md +2 -2
- package/src/gdskills/bundled/skills/orchestration/task-implementer/SKILL.md +4 -4
- package/src/gdskills/bundled/skills/planning/autodoc-orchestrator/SKILL.md +1 -1
- package/src/gdskills/bundled/skills/planning/interviewer/SKILL.md +3 -3
- package/src/gdskills/bundled/skills/quality/api-truth/SKILL.md +2 -2
- package/src/gdskills/bundled/skills/quality/deprecation-path/SKILL.md +3 -3
- package/src/gdskills/bundled/skills/quality/fresh-eyes/SKILL.md +2 -2
- package/src/gdskills/bundled/skills/quality/perf-check/SKILL.md +1 -1
- package/src/gdskills/bundled/skills/quality/root-cause/SKILL.md +2 -2
- package/src/gdskills/bundled/skills/review/review-architecture/SKILL.md +14 -15
- package/src/gdskills/bundled/skills/review/review-backend/SKILL.md +14 -14
- package/src/gdskills/bundled/skills/review/review-frontend/SKILL.md +13 -16
- package/src/gdskills/bundled/skills/review/review-logic/SKILL.md +14 -14
- package/src/gdskills/bundled/skills/review/review-orchestrator/SKILL.detail.md +42 -3
- package/src/gdskills/bundled/skills/review/review-orchestrator/SKILL.md +21 -19
- package/src/gdskills/bundled/skills/review/review-orchestrator/reviewer-finding.schema.json +16 -0
- package/src/gdskills/bundled/skills/review/review-orchestrator/reviewer-input.schema.json +47 -0
- package/src/gdskills/bundled/skills/review/review-orchestrator/templates/pr-comment-frontend.md +1 -1
- package/src/gdskills/bundled/skills/review/review-performance/SKILL.md +13 -13
- package/src/gdskills/bundled/skills/review/review-regression/SKILL.md +23 -25
- package/src/gdskills/bundled/skills/review/review-security-code/SKILL.md +13 -15
- package/src/gdskills/bundled/skills/review/review-testing-practices/SKILL.md +31 -57
package/README.md
CHANGED
|
@@ -40,7 +40,7 @@ the completion itself signed.
|
|
|
40
40
|
|
|
41
41
|
## Quick start
|
|
42
42
|
|
|
43
|
-
**Requirements:** `git` and `bun` (>= 1.3.14 — older Bun can close the terminal input of `keryx shell`; see [onboarding](docs/docs/onboarding.md#bun-version)).
|
|
43
|
+
**Requirements:** `git` (>= 2.24) and `bun` (>= 1.3.14 — older Bun can close the terminal input of `keryx shell`; see [onboarding](docs/docs/onboarding.md#bun-version)).
|
|
44
44
|
|
|
45
45
|
```bash
|
|
46
46
|
npm install -g @mrciphersmith/keryx
|
|
@@ -491,7 +491,17 @@ What is in it today:
|
|
|
491
491
|
inside a running session, decide whether that approval is asked for at all
|
|
492
492
|
— `ask` (default), `trust` (safe calls run, a destructive one still asks),
|
|
493
493
|
`auto` (nothing asks except a credentials-touching command, which no mode
|
|
494
|
-
ever auto-approves
|
|
494
|
+
ever auto-approves; it also answers the agent's own questions by picking the
|
|
495
|
+
recommended option and saying so, while questions about irreversible or
|
|
496
|
+
untrusted actions still ask you). The mode also sets how many model rounds one turn may
|
|
497
|
+
run: 40 in `ask`, 200 in `trust` (asks at the limit) and 200 in `auto`, where
|
|
498
|
+
reaching the limit extends the turn without asking and the turn stops at 200
|
|
499
|
+
(`KERYX_AGENT_MAX_ROUNDS` overrides all three, up to 200). A turn that
|
|
500
|
+
belongs to an **automatic review** (named in the request; the package is
|
|
501
|
+
created with `keryx review start --automatic`) gets the 200 limit in any
|
|
502
|
+
mode, runs that review's own commands without a prompt and asks only in a
|
|
503
|
+
closed list of emergencies; `review complete` and its gate are unchanged.
|
|
504
|
+
A session-level layer only: `harness run`/`exec`,
|
|
495
505
|
`keryx serve`, and MCP keep the unconditional policy engine above,
|
|
496
506
|
untouched. See the [permission modes guide](docs/docs/guides/permission-modes.md).
|
|
497
507
|
- **Kernel-enforced containment underneath.** The OS sandbox sits *below* the
|
|
@@ -848,7 +858,7 @@ next.
|
|
|
848
858
|
| Requirement | Status |
|
|
849
859
|
|-------------|--------|
|
|
850
860
|
| Bun | >= 1.3.14 |
|
|
851
|
-
| Git | Required for hooks, `--changed` scopes and the managed installer; the core runs without it |
|
|
861
|
+
| Git | >= 2.24 (`--end-of-options`, used wherever a typed revision reaches git). Required for hooks, `--changed` scopes and the managed installer; the core runs without it |
|
|
852
862
|
| ripgrep | Required only for `keryx ctx rg` and the agent's `search_code` tool |
|
|
853
863
|
| Model provider credential | Required only for the optional AI commands below |
|
|
854
864
|
| macOS | Full support, including the complete policy sandbox |
|
|
@@ -1134,6 +1144,10 @@ repository checkout, since the evaluation corpus is not shipped in the npm
|
|
|
1134
1144
|
package. See
|
|
1135
1145
|
[run keryx in CI](docs/docs/guides/run-in-ci.md).
|
|
1136
1146
|
|
|
1147
|
+
### Layers and coverage in a review
|
|
1148
|
+
|
|
1149
|
+
Every managed review start, attach or ingest (and any `dispatch-check` or `blast-radius` given `--package`) refreshes the graph, the wiki and the memory index for the PR head and writes `layers.json`, with **no model call**; `wiki enrich` is the only model step and is never automatic (`keryx review layers refresh --package <dir> [--write-wiki]`). Blast-radius files beyond the 40-file cap are reviewed in chunks (`keryx review chunks estimate|next|record|merge`; the ceiling is 25 passes of 40 files, `reviewChunks.maxPasses` in `routing.config.json`), `keryx review coverage` counts what is covered, omitted by name, and uncovered, and `keryx review complete` refuses while a file is uncovered unless you pass `--waive-coverage "<your words>"`. `/reviews` and `/review layers` in the shell show it.
|
|
1150
|
+
|
|
1137
1151
|
### Review bot on pull requests
|
|
1138
1152
|
|
|
1139
1153
|
A GitHub Action reviews each same-repository pull request, has a second model
|
|
@@ -0,0 +1,56 @@
|
|
|
1
|
+
# Third-party notices
|
|
2
|
+
|
|
3
|
+
Keryx adapts a small number of techniques from third-party works. This file
|
|
4
|
+
carries the notices their licenses require. Everything else in this
|
|
5
|
+
repository is covered by [LICENSE](LICENSE).
|
|
6
|
+
|
|
7
|
+
## agent-skills (MIT)
|
|
8
|
+
|
|
9
|
+
Source: https://github.com/addyosmani/agent-skills
|
|
10
|
+
|
|
11
|
+
Techniques adapted from this skills library and re-expressed in Keryx's own
|
|
12
|
+
words and mechanisms:
|
|
13
|
+
|
|
14
|
+
- `src/gdskills/bundled/skills/quality/root-cause` — carrying a debugging skill.
|
|
15
|
+
- `src/gdskills/bundled/skills/quality/fresh-eyes` — the in-flight doubt pass
|
|
16
|
+
(a reader gets the artifact and the contract, never the author's reasoning).
|
|
17
|
+
- `src/gdskills/bundled/skills/quality/api-truth` — checking a dependency before
|
|
18
|
+
calling it and marking what went out unchecked.
|
|
19
|
+
- `src/gdskills/bundled/skills/quality/deprecation-path` — carrying a
|
|
20
|
+
deprecation skill.
|
|
21
|
+
- `src/gdskills/bundled/skills/quality/perf-check` — "neutral is a revert" and
|
|
22
|
+
the ledger of reverted attempts.
|
|
23
|
+
- `src/gdskills/bundled/skills/planning/interviewer` — a hedged agreement is not
|
|
24
|
+
approval.
|
|
25
|
+
- `src/gdskills/bundled/skills/orchestration/task-implementer` — the
|
|
26
|
+
noticed-but-not-touching / assumptions / not-touched result fields.
|
|
27
|
+
- `src/review/floor.ts` — the catalogue of four test-weakening edits.
|
|
28
|
+
- `src/gdskills/bundled-eval.ts` — the required skill anatomy, the length budget
|
|
29
|
+
and description discipline, made executable.
|
|
30
|
+
- `src/commands/routing-corpus.ts` — the routing corpus shape (positive prompts
|
|
31
|
+
per skill, negatives naming the owner that must outrank them).
|
|
32
|
+
- `docs/skills/rejected-skill-changes.md` — the rejected-change ledger.
|
|
33
|
+
|
|
34
|
+
```
|
|
35
|
+
MIT License
|
|
36
|
+
|
|
37
|
+
Copyright (c) 2025 Addy Osmani
|
|
38
|
+
|
|
39
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
40
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
41
|
+
in the Software without restriction, including without limitation the rights
|
|
42
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
43
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
44
|
+
furnished to do so, subject to the following conditions:
|
|
45
|
+
|
|
46
|
+
The above copyright notice and this permission notice shall be included in all
|
|
47
|
+
copies or substantial portions of the Software.
|
|
48
|
+
|
|
49
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
50
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
51
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
52
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
53
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
54
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
55
|
+
SOFTWARE.
|
|
56
|
+
```
|