Skip to content

Commit 3567cbd

Browse files
release: 1.653.0 (stamped at land time) (#2273)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
1 parent 8f6d7ba commit 3567cbd

19 files changed

Lines changed: 48 additions & 38 deletions

‎.claude-plugin/plugin.json‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@
22
"name": "perun",
33
"description": "Perun: a suite of skills for building with AI agents. Deep code review, token and cost tools (spend reports, auto-compact with handoff, model-tier guidance), gated agentic delivery (merge train, lanes, destructive-command guard, auto-update), an idea critic, and product, growth, positioning, business-ops and output-safety specialists.",
44
"skills": "./.claude/skills",
5-
"version": "1.652.0",
5+
"version": "1.653.0",
66
"author": {
77
"name": "deep-code-review contributors"
88
},

‎.claude/skills/agentic-delivery/SKILL.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@ description: >-
88
license: MIT
99
metadata:
1010
author: deep-code-review contributors
11-
version: "1.652.0"
11+
version: "1.653.0"
1212
---
1313

1414
# Agentic delivery
Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
1.652.0
1+
1.653.0

‎.claude/skills/deep-code-review/SKILL.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@ description: >-
88
license: MIT
99
metadata:
1010
author: deep-code-review contributors
11-
version: "1.652.0"
11+
version: "1.653.0"
1212
---
1313

1414
# Deep Code Review
Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
1.652.0
1+
1.653.0

‎.claude/skills/idea-critic/SKILL.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@ description: >-
88
license: MIT
99
metadata:
1010
author: deep-code-review contributors
11-
version: "1.652.0"
11+
version: "1.653.0"
1212
---
1313

1414
# Veles: the evil twin (idea-critic)

‎.claude/skills/idea-critic/VERSION‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
1.652.0
1+
1.653.0

‎.codex-plugin/plugin.json‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
{
22
"name": "perun",
3-
"version": "1.652.0",
3+
"version": "1.653.0",
44
"description": "Perun: a skill suite for AI-agent teams: deep code review, token and cost tools, gated delivery, an idea critic and product specialists.",
55
"skills": "./.claude/skills",
66
"interface": {

‎CHANGELOG.md‎

Lines changed: 31 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -3,6 +3,37 @@
33
All notable changes to this repository are documented here. Format loosely
44
follows Keep a Changelog; versioning follows Semantic Versioning.
55

6+
## [1.653.0] — 2026-11-08
7+
8+
### Changed
9+
- `review_checks.sh` collapses bandit B101 (assert) hits per file the same way as gitleaks generic-api-key hits: a file with more than 3 B101 rows becomes one `B101-summary` row (first hit as line, message "N B101 hits in this file", added-line `new` and `added: N` naming apply). Other bandit rules stay per row; 3 or fewer stay; `REVIEW_FIXTURE_ROWS=all` keeps every row. Test-dir B101 hits collapse per file, not into a fixture bucket.
10+
11+
### Changed
12+
- `committed_personal_data.py` (committed-data check) prints one `not-checked` row for a compressed or columnar container kind that more than 3 files share, instead of one row per file: under the first path in sorted order, message `N <kind> containers not scanned (first 3: a, b, c)`. A repository with thousands of PDFs no longer prints thousands of rows. Groups of 3 or fewer, partial-scan rows, address findings and the repository total are unchanged; the container rows now print after the per-file rows.
13+
14+
### Changed
15+
- `dependency_currency.py` TSV output (default mode, `review_checks.sh` dep-currency step) prints one `not-checked` row instead of one per package when more than 3 not-checked rows share a file, ecosystem and note. The row sits at the smallest line and names the first 3 packages; `--json` still lists every row. Groups of 3 or fewer and all other rules print unchanged.
16+
17+
### Changed
18+
- `route_auth_map.py` treats a mutating handler that forwards the caller's credential upstream as verified by that upstream, so those rows no longer report as HIGH. On the real-repo corpus HIGH rows drop 43 to 35, the 2 confirmed unauthenticated routes stay HIGH, and the fixtures stay at 223.
19+
- `invisible_chars.py` goes quiet in a prose or data file for an isolated U+200B, U+200D, U+FE0F or U+FEFF. A file with 4 or more of them gets one summary row, and locale files are exempt. Runs, tag and variation-selector characters, bidi controls and everything else still report, and the agent-instruction file list is widened. Agent-instruction files, `.github`/`.claude`/`.cursor` paths and code keep full reporting. Hidden-unicode corpus rows drop 262 to 55.
20+
21+
### Changed
22+
- `review_checks.sh` collapses gitleaks and gitleaks-history `generic-api-key` hits per file when one non-fixture file has more than 3 of them: one `generic-api-key-summary` row per file, at the first hit's line, with the count, the first three line numbers and a `REVIEW_FIXTURE_ROWS=all` hint. A file with 3 or fewer hits, any other rule in the same file, a fixture-directory bucket (which keeps its own summary) and `REVIEW_FIXTURE_ROWS=all` (every row) are unchanged. No secret value is printed. On a 3,937-row audit corpus the generic-api-key rows fell to 65.
23+
24+
### Added
25+
- The Perun gitleaks template gains a `capability-url` rule: a URL whose query carries a long invite, share or access token, a magic link or a signature is reported, since whoever holds the link gets access. Runs of `x` and `example` values are skipped.
26+
- `route_auth_map.py` recognises more house guards: `require*` now takes Member, Account, Viewer and Principal, and `req.member` / `request.member` reads count as identity evidence (an assignment does not).
27+
- `route_auth_map.py` labels a handler that answers 401 itself (`status: 401`, `.status(401)`, `sendStatus(401)`, `abort(401)`, `HTTPException(status_code=401)`) as `auth=presence-check`. Such a mutating row drops to INFO only when the handler forwards the caller's credential in an outbound `Authorization` header or a Bearer value, so the upstream verifies it; a presence check that then trusts the header stays HIGH.
28+
- size-budget-raise: .claude/skills/deep-code-review/references/method-situational.md 44699→44807 capability-url rule named in the gitleaks template line
29+
30+
### Added
31+
- `route_auth_map.py` (default mode, and `review_checks.sh --full` route-auth) no longer skips a repo whose handlers are in a language it does not judge. A Go, Spring, PHP or Django urlconf file with routes that accept a state-changing method (anything but GET, HEAD or OPTIONS; ANY counts) gets one `UNJUDGED` not-checked row at its first such route, with the count. Auth is not judged for those files; the row says to read the handlers. A Django app with only urlconf routes no longer reads "no HTTP handlers". `route-auth (SKIPPED: no HTTP handlers)` now shows only when there are no handlers and no unjudged routes.
32+
- `route_auth_map.py` reads `.Methods(http.MethodPost)`-style gorilla method constants, so a `http.MethodGet` route is no longer reported with an empty method.
33+
34+
### Added
35+
- `review_checks.sh` drops a trufflehog row in notebook and css-family blob types (`.ipynb`, `.css`, `.scss`, `.less`, `.nb`) only when the line is a blob: in `.ipynb` an output mime key at line start (`"image/<type>":`) or a `data:<mime>;base64,` URI, so a code-cell base64 or hex literal is kept; in the other types a run of 200 or more unbroken base64-alphabet characters, a wrapped base64 chunk, or a line over 2000 characters. Rows are also dropped for a digest manifest (`.sprockets-manifest-<hex>.json`, asset digests) and for a `X.gz` whose decompressed bytes equal its `X`. Base64 lines in yaml, env, json, source and `.map` files are kept: trufflehog decodes base64, so such a line can be a real encoded secret, and `.map` files can carry `sourcesContent`. A row with line number below 1 is kept, and a row whose line cannot be re-read is kept.
36+
637
## [1.652.0] — 2026-11-06
738

839
### Added

‎SHA256SUMS‎

Lines changed: 8 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -20,8 +20,8 @@ ef9e423362b12d0dd73c8e165faf32e9f53c4d4ac215b57c2e34002a88713838 .claude/skills
2020
9521b057ff3db92d1e60923ebd88170713914f7e235ceb45717eb39419daf811 .claude/skills/agentic-ceo/scripts/weekly_receipt.py
2121
81e857c1c5462222ebf737517c1d8639eac3d56b564854af8a7340537752adc1 .claude/skills/agentic-delivery/INDEX-scripts.md
2222
39b5ff201f1ef82cb7fb95f8f69f407c91eaf024ce8ed7c6496baed89f481a15 .claude/skills/agentic-delivery/INDEX.md
23-
f7f645029c4a3c8481250af8ee342fb6b8b3336f5054e1dea408d8eeccc279d4 .claude/skills/agentic-delivery/SKILL.md
24-
d309a8dea955ecc8300b5aabf441deafce0cd52b575192b2e02232fd20283cd2 .claude/skills/agentic-delivery/VERSION
23+
c63d96d07394cdd21673604b31c93844fe2e1c876c0655559b7ec00b9c414ec7 .claude/skills/agentic-delivery/SKILL.md
24+
aa700942c5e945df4932c4fabdeafe698f59b200aa014dc15381eb09d4eab6f4 .claude/skills/agentic-delivery/VERSION
2525
51194479002df37b742c31c7ff8e547dd956e6a4b487ab2d8f150b365c54a814 .claude/skills/agentic-delivery/evals/evals.json
2626
fba7699a2d0836b51b7c838d019b1f0cd20570522a362e8a32b8dacfb911298e .claude/skills/agentic-delivery/evals/triggers.json
2727
d7ab95bb614e70bec6d97bf16f06ffd510d4f9472139519bd8c93dd5061e69ba .claude/skills/agentic-delivery/references/cost-quality-guardrails.md
@@ -206,8 +206,8 @@ d98f3e4380da3496061ba63be7beebe75017ab4ee1734eda9a3e798e19007b0e .claude/skills
206206
18f20db537ab784460ad33cf218f7a01ac0a172fc38413f14c0e19e1ce6821cc .claude/skills/contribution/scripts/share_learning.py
207207
b376ccf4f6e59b46c077455c81de26c3845af14da01f34741bb2dc6c0bf6cfe2 .claude/skills/deep-code-review/INDEX-scripts.md
208208
18e85b024feb3160322022697752939f38a033895be63bd92d46fd43626e7f75 .claude/skills/deep-code-review/INDEX.md
209-
26b4d0865fd96fb0bd0978167dc40f69edb7024ca620b2283601101b26c2163d .claude/skills/deep-code-review/SKILL.md
210-
d309a8dea955ecc8300b5aabf441deafce0cd52b575192b2e02232fd20283cd2 .claude/skills/deep-code-review/VERSION
209+
7505ee5c7f06c8f94eb88edcfd612c4eabaaa5706c12477eb3bfd0464ebf47ff .claude/skills/deep-code-review/SKILL.md
210+
aa700942c5e945df4932c4fabdeafe698f59b200aa014dc15381eb09d4eab6f4 .claude/skills/deep-code-review/VERSION
211211
6d0c4e43596b5ff12d55d4d1f2d9403706a4ed1e17e4141941b1de999f8734c0 .claude/skills/deep-code-review/evals/evals.json
212212
13dbb52e4bf61b1093302e094151def167371d26e7cf67a026f3beeb09d34378 .claude/skills/deep-code-review/evals/triggers.json
213213
b84e3a2b88dc189f041784e76b97e68ad879915814e00f4b37cd27513746d6f1 .claude/skills/deep-code-review/references/a11y-aria.md
@@ -361,12 +361,12 @@ b4d0333271317c881d175dca1d06092cc1dd4859e33f0b375eb870a152a80e6c .claude/skills
361361
47bb118346969518b3688f08fe1c28202d4b7c9d5ac8b7d09e1240af931121cb .claude/skills/deep-code-review/scripts/ci_cost_advisor.py
362362
a0301aaa1b602fd8ca44955eaf66b17893905f99aee8fe7d8db0be981bcc25c2 .claude/skills/deep-code-review/scripts/ci_cost_lint.py
363363
ed0f0c1eebfdca4dcbaafbf5c2f7b5cc93884de38715b57db894e0f210a811f9 .claude/skills/deep-code-review/scripts/closes_lint.py
364-
5cb0f694762296e7f05a183538afa2264f6f5627555264a7ce4e663821f31302 .claude/skills/deep-code-review/scripts/committed_personal_data.py
364+
f8d10be1934b7e69ab1eb19f5aef2bbfd21e16676cbb6cdcd27aaff476f22e9a .claude/skills/deep-code-review/scripts/committed_personal_data.py
365365
6632d4fa969875bb8a26fd8fd1677c3a5eaeeadbea00cb28c88a22d3d9dd4138 .claude/skills/deep-code-review/scripts/context_pack.py
366366
d682969dcd1adb66f2fcfb986dfe3b3c3090a0ea6adf35847c31dce84d6bac32 .claude/skills/deep-code-review/scripts/cost_limit_map.py
367367
c5b40f3f0d3fb174e6c9490150c7b85723de33f95d3fab87d60214d4939401a8 .claude/skills/deep-code-review/scripts/coverage_ledger.py
368368
c724de8ef9b9a028c7bd5e5c4198caf92bb1bad9a579deb3e6c68d7f413c2b8e .claude/skills/deep-code-review/scripts/csrf_cookie_map.py
369-
10155df445d54b8cd517052ddc5583277b38f1c525c977c066081a8b135917f8 .claude/skills/deep-code-review/scripts/dependency_currency.py
369+
4af7fdb3f0914357d21f87f965b931d4a9c6098a969e4538a6cac76a466c9f00 .claude/skills/deep-code-review/scripts/dependency_currency.py
370370
36c1447a65135496c519397c04c6a74ef066bd01399d593907e598b8d2038498 .claude/skills/deep-code-review/scripts/derived_numbers.py
371371
662031286d4a29aff2e1aaa5cb4fa287ab99f9c7b5124085ea47da3dc66622cc .claude/skills/deep-code-review/scripts/design_drift.py
372372
254f1016d76feb5fc2c26925b74ccc1158343d5075efcf875ef1a60305bab827 .claude/skills/deep-code-review/scripts/dynamic_probe.py
@@ -436,8 +436,8 @@ e951eaa63fd0d1870701d69c89c29ef4fe74dbad9b680ab6ef0a4e993fb8bc20 .claude/skills
436436
fd316c63bbf1cc0ea64322d7d923d66692eaf7b545c056367bfecca35146c862 .claude/skills/growth-analytics/references/full-map.md
437437
2f1d7499e9ed7ae59f550b5c00043e465b2a4e885821ad153438e2e33fcccb45 .claude/skills/growth-analytics/scripts/readout_check.py
438438
8a1f63ae9aed7ab6a092142886a0deecd1981e2e240e714106f9f17d469da220 .claude/skills/idea-critic/INDEX.md
439-
48fd46abd26a953147423ea17a626f19f2881c36ba7dbe95977c4a17530b7073 .claude/skills/idea-critic/SKILL.md
440-
d309a8dea955ecc8300b5aabf441deafce0cd52b575192b2e02232fd20283cd2 .claude/skills/idea-critic/VERSION
439+
044e982f2ff2a232c6462d244f37166e9ac132f2bba227380cd7e53fdca8b1e8 .claude/skills/idea-critic/SKILL.md
440+
aa700942c5e945df4932c4fabdeafe698f59b200aa014dc15381eb09d4eab6f4 .claude/skills/idea-critic/VERSION
441441
9f77d5d6d13c99fd16cd8c398795d8be226d74bae5a5bf65369ed94a59fa56b7 .claude/skills/idea-critic/evals/evals.json
442442
0e1c0db300661a1ab1a427a7157c75e2ed67130ab24d5c6cb02ec3ff9e793f58 .claude/skills/idea-critic/evals/heldout.json
443443
d8098c5ced63e8b8608867dd144091049563a0553cf62d27ce05b9d52ac54266 .claude/skills/idea-critic/evals/triggers.json

0 commit comments

Comments
 (0)