@skein-code/cli 0.3.16 → 0.3.18
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 +5 -3
- package/dist/cli.js +79 -21
- package/dist/cli.js.map +1 -1
- package/docs/ARCHITECTURE.md +5 -3
- package/docs/NEXT_STEPS.md +32 -12
- package/package.json +9 -1
package/docs/ARCHITECTURE.md
CHANGED
|
@@ -53,7 +53,9 @@
|
|
|
53
53
|
generation. After the write succeeds, compare only newly added or at least
|
|
54
54
|
1.5x-expanded functions before refreshing changed index paths. Exact
|
|
55
55
|
normalized hashes identify Type-1/2 clones; winnowed 10-token shingles and
|
|
56
|
-
Jaccard similarity identify
|
|
56
|
+
Jaccard similarity identify Type-3 candidates. The repository-owned fixture
|
|
57
|
+
matrix calibrates threshold 0.55: unsuppressed Type-1/2 matches block the
|
|
58
|
+
existing completion gate, while Type-3 remains warning-only. Type-4 semantic
|
|
57
59
|
equivalence is explicitly outside this deterministic contract.
|
|
58
60
|
12. Run configured verification commands after changes. The completion gate accepts
|
|
59
61
|
only current successful test, typecheck, lint, build, check, or `git diff
|
|
@@ -99,8 +101,8 @@ build, targeted upsert, and deletion. Audit receipts never retain source,
|
|
|
99
101
|
normalized tokens, literal contents, prompts, or raw retrieval failures.
|
|
100
102
|
|
|
101
103
|
Duplication findings are folded into the existing completion record rather than
|
|
102
|
-
creating a parallel completion state.
|
|
103
|
-
|
|
104
|
+
creating a parallel completion state. Fixture benchmarks establish the required
|
|
105
|
+
precision for calibrated Type-1/2 blocking; Type-3 remains warning-only. Active
|
|
104
106
|
matches receive stable, bounded ids; the optional `duplication_audit` read tool
|
|
105
107
|
is exposed only while unsuppressed findings exist. Suppression is exact-match,
|
|
106
108
|
reason-coded, and persisted as a content-free audit receipt. Repaired/deleted/
|
package/docs/NEXT_STEPS.md
CHANGED
|
@@ -9,7 +9,7 @@ one of the milestones below.
|
|
|
9
9
|
|
|
10
10
|
- Product name: `Skein`; primary executable: `skein`.
|
|
11
11
|
- Compatibility executables: `mosaic` and `mosaic-code`.
|
|
12
|
-
- Current repository version: `0.3.
|
|
12
|
+
- Current repository version: `0.3.18`.
|
|
13
13
|
- Runtime requirement: Node.js `>=22.16.0` (the runtime uses unflagged
|
|
14
14
|
`node:sqlite` with FTS5, and current CLI/build dependencies require this
|
|
15
15
|
Node 22 baseline).
|
|
@@ -41,8 +41,8 @@ npm audit --omit=dev
|
|
|
41
41
|
npm run release:verify -- --output-dir artifacts/package
|
|
42
42
|
```
|
|
43
43
|
|
|
44
|
-
The latest verified package is `skein-code-cli-0.3.
|
|
45
|
-
its SHA-256 to `artifacts/package/skein-code-cli-0.3.
|
|
44
|
+
The latest verified package is `skein-code-cli-0.3.18.tgz`. The verifier writes
|
|
45
|
+
its SHA-256 to `artifacts/package/skein-code-cli-0.3.18.tgz.sha256`, and CI
|
|
46
46
|
retains the checksum beside the package metadata. The checksum is deliberately
|
|
47
47
|
not copied into this packaged document because doing so would change the
|
|
48
48
|
archive it describes.
|
|
@@ -54,7 +54,25 @@ short-height case. The current full-suite count is recorded from the latest
|
|
|
54
54
|
|
|
55
55
|
## Recommended Order
|
|
56
56
|
|
|
57
|
-
### P0-
|
|
57
|
+
### P0-E: Token Economy measurement and bounded schema disclosure
|
|
58
|
+
|
|
59
|
+
Version `0.3.18` adds the repository-owned `token-economy-benchmark-v1`
|
|
60
|
+
deterministic replay. Its seven retrieval, large-output, CJK, and repeated
|
|
61
|
+
zero-hit fixtures compare the legacy ceilings with the adaptive budget and
|
|
62
|
+
firewall policies while requiring complete evidence and recovery coverage.
|
|
63
|
+
The replay reports a 36.7% estimated-input reduction; this is a deterministic
|
|
64
|
+
budget measurement, not a provider billing or task-success claim.
|
|
65
|
+
|
|
66
|
+
MCP tool definitions are now marked for progressive disclosure. Catalogs of
|
|
67
|
+
eight or fewer eligible tools remain fully visible; larger catalogs disclose a
|
|
68
|
+
deterministically ranked maximum of eight schemas. Selected schemas retain
|
|
69
|
+
their registry order and remain loaded for the current run, and Token Ledger
|
|
70
|
+
receipts record the deferred count. MCP tools remain `network` operations and
|
|
71
|
+
hidden tool calls remain rejected by the existing boundary. This release does
|
|
72
|
+
not defer MCP connection or remote `listTools` discovery and does not claim the
|
|
73
|
+
provider-cache or Context Compaction 2.0 follow-up work is complete.
|
|
74
|
+
|
|
75
|
+
### P0-D: Repository reuse and calibrated duplication enforcement
|
|
58
76
|
|
|
59
77
|
Version `0.3.14` added the prompt ladder and pre-write `ReuseReceipt`. Version
|
|
60
78
|
`0.3.15` added a post-write deterministic audit for ordinary TS/TSX/JS/JSX/MJS/
|
|
@@ -66,14 +84,16 @@ renames, moves, deletions, small functions, tests, fixtures, generated/vendor/
|
|
|
66
84
|
dist/minified/declaration files, and failed writes stay quiet. Receipts expose
|
|
67
85
|
at most eight path/symbol/similarity matches through session, TUI, JSON, and
|
|
68
86
|
JSONL without storing source, normalized tokens, prompts, or raw index errors.
|
|
69
|
-
Type-
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
duplication summaries into the single completion
|
|
76
|
-
|
|
87
|
+
Type-3 remains warning-only and unavailable evidence is `unresolved`, never a
|
|
88
|
+
false pass. The repository-owned `duplication-benchmark-v1` fixture matrix
|
|
89
|
+
calibrated threshold `0.55` at 100% recall, 100% precision, and 0% legitimate
|
|
90
|
+
boundary false-positive rate. Unsuppressed Type-1/2 matches now block the
|
|
91
|
+
existing completion record; Type-4 semantic equivalence is not promised.
|
|
92
|
+
|
|
93
|
+
Version `0.3.16` integrated duplication summaries into the single completion
|
|
94
|
+
receipt and every output surface while keeping all matches warning-only. Version
|
|
95
|
+
`0.3.17` keeps Type-3 warning-only but promotes calibrated Type-1/2 matches to
|
|
96
|
+
the completion gate. Active matches receive stable 24-character ids. The read-only
|
|
77
97
|
`duplication_audit` tool is disclosed only when unsuppressed matches exist and
|
|
78
98
|
can suppress one exact match with a reason code plus a bounded explanation;
|
|
79
99
|
wildcard/global suppression, credentials, and code-block reasons are rejected.
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@skein-code/cli",
|
|
3
|
-
"version": "0.3.
|
|
3
|
+
"version": "0.3.18",
|
|
4
4
|
"description": "A context-first, model-agnostic coding agent with an auditable terminal workspace.",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"license": "MIT",
|
|
@@ -16,6 +16,12 @@
|
|
|
16
16
|
},
|
|
17
17
|
"skein": {
|
|
18
18
|
"releaseNotes": [
|
|
19
|
+
"Large MCP catalogs now disclose at most eight request-relevant schemas per run while preserving network permissions",
|
|
20
|
+
"Token Ledger receipts report deferred progressive schemas and selected definitions remain stable across tool turns",
|
|
21
|
+
"A deterministic Token Economy replay benchmark guards evidence, output-firewall, and no-progress invariants",
|
|
22
|
+
"Calibrated Type-1/2 duplication matches now block completion until reuse or exact audited suppression",
|
|
23
|
+
"Type-3 duplication remains warning-only and Type-4 semantic equivalence remains explicitly unsupported",
|
|
24
|
+
"Duplication benchmark fixtures report deterministic recall, precision, false-positive rate, and latency",
|
|
19
25
|
"Duplication warnings now aggregate into the single completion receipt without changing verified status",
|
|
20
26
|
"Exact match ids support bounded, reason-coded suppression with audit and credential/code-block safeguards",
|
|
21
27
|
"Duplication audit tools are disclosed only when active warnings exist and repaired paths clear stale matches",
|
|
@@ -54,6 +60,8 @@
|
|
|
54
60
|
"pretest:pty": "npm run build",
|
|
55
61
|
"test:pty": "sh test/pty/run-visual.sh",
|
|
56
62
|
"benchmark:context": "tsx scripts/benchmark-local-index.ts",
|
|
63
|
+
"benchmark:duplication": "tsx scripts/benchmark-duplication.ts",
|
|
64
|
+
"benchmark:token-economy": "tsx scripts/benchmark-token-economy.ts",
|
|
57
65
|
"verify:package": "node scripts/verify-package.mjs",
|
|
58
66
|
"release:verify": "npm run check && npm run verify:package --",
|
|
59
67
|
"typecheck": "tsc --noEmit",
|