nobrainer-tech-flow 0.0.0-stage → 2.4.0

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.
Files changed (149) hide show
  1. package/.agents/plugins/marketplace.json +20 -0
  2. package/.claude-plugin/marketplace.json +18 -0
  3. package/.claude-plugin/plugin.json +21 -0
  4. package/.codex-plugin/plugin.json +47 -0
  5. package/.opencode/INSTALL.md +37 -0
  6. package/.opencode/plugins/nobrainer-tech-skills.js +66 -0
  7. package/.pi/extensions/nobrainer-tech-skills.js +64 -0
  8. package/AGENTS.md +204 -0
  9. package/CLAUDE.md +204 -0
  10. package/CONTRIBUTING.md +70 -0
  11. package/GEMINI.md +1 -0
  12. package/LICENSE +21 -0
  13. package/README.md +658 -2
  14. package/SECURITY.md +33 -0
  15. package/adapters/bootstrap.md +27 -0
  16. package/assets/nobrainer-tech-logo.svg +1 -0
  17. package/assets/nobrainer-workflow.svg +155 -0
  18. package/assets/plugin-logo.png +0 -0
  19. package/bin/nobrainer-tech-flow.mjs +48 -0
  20. package/docs/BOUNDED_RUNNER.md +60 -0
  21. package/docs/CHAT_PLUGINS.md +29 -0
  22. package/docs/COMPATIBILITY.md +371 -0
  23. package/docs/INSTALL.md +626 -0
  24. package/docs/MIGRATION_TO_FLOW.md +117 -0
  25. package/docs/NAMING.md +39 -0
  26. package/docs/PRIVACY.md +9 -0
  27. package/docs/SESSION_RESTART.md +70 -0
  28. package/docs/SKILL_CURATION.md +124 -0
  29. package/docs/TESTING.md +232 -0
  30. package/docs/TRY_IT.md +109 -0
  31. package/docs/releases/v1.0.0.md +81 -0
  32. package/docs/releases/v1.1.0.md +85 -0
  33. package/docs/releases/v1.10.0.md +34 -0
  34. package/docs/releases/v1.11.0-publication-readback.md +25 -0
  35. package/docs/releases/v1.11.0.md +42 -0
  36. package/docs/releases/v1.12.0.md +30 -0
  37. package/docs/releases/v1.13.0.md +46 -0
  38. package/docs/releases/v1.14.0.md +43 -0
  39. package/docs/releases/v1.14.1.md +21 -0
  40. package/docs/releases/v1.2.0.md +87 -0
  41. package/docs/releases/v1.2.1.md +92 -0
  42. package/docs/releases/v1.3.0-publication-readback.md +69 -0
  43. package/docs/releases/v1.3.0.md +66 -0
  44. package/docs/releases/v1.3.1-publication-readback.md +76 -0
  45. package/docs/releases/v1.3.1.md +59 -0
  46. package/docs/releases/v1.4.0-publication-readback.md +81 -0
  47. package/docs/releases/v1.4.0.md +43 -0
  48. package/docs/releases/v1.5.0-publication-readback.md +76 -0
  49. package/docs/releases/v1.5.0.md +54 -0
  50. package/docs/releases/v1.6.0.md +64 -0
  51. package/docs/releases/v1.6.1-publication-readback.md +53 -0
  52. package/docs/releases/v1.6.1.md +82 -0
  53. package/docs/releases/v1.7.0.md +42 -0
  54. package/docs/releases/v1.7.1-publication-readback.md +64 -0
  55. package/docs/releases/v1.7.1.md +49 -0
  56. package/docs/releases/v1.8.0.md +63 -0
  57. package/docs/releases/v1.8.1.md +32 -0
  58. package/docs/releases/v1.9.0.md +33 -0
  59. package/docs/releases/v2.0.0-candidate.md +28 -0
  60. package/docs/releases/v2.0.0.md +53 -0
  61. package/docs/releases/v2.0.1.md +135 -0
  62. package/docs/releases/v2.0.2.md +189 -0
  63. package/docs/releases/v2.1.0.md +64 -0
  64. package/docs/releases/v2.1.1.md +43 -0
  65. package/docs/releases/v2.2.0.md +64 -0
  66. package/docs/releases/v2.2.1.md +11 -0
  67. package/docs/releases/v2.3.0.md +97 -0
  68. package/docs/releases/v2.3.1.md +71 -0
  69. package/docs/releases/v2.3.2.md +242 -0
  70. package/docs/releases/v2.4.0.md +126 -0
  71. package/gemini-extension.json +6 -0
  72. package/hooks/claude-hooks.json +17 -0
  73. package/hooks/hooks-cursor.json +10 -0
  74. package/hooks/run-hook.cmd +36 -0
  75. package/hooks/session-start +38 -0
  76. package/package.json +69 -4
  77. package/plugin.json +51 -0
  78. package/scripts/build_plugin.py +50 -0
  79. package/scripts/check_flow_update.py +18 -0
  80. package/scripts/check_public_commit.py +86 -0
  81. package/scripts/fetch_public_refs.py +69 -0
  82. package/scripts/install.py +600 -0
  83. package/scripts/install.sh +1392 -0
  84. package/scripts/install_guards.py +279 -0
  85. package/scripts/install_personalization.py +527 -0
  86. package/scripts/install_skills.py +951 -0
  87. package/scripts/npm_install.py +85 -0
  88. package/scripts/recommend_flow_setup.py +1522 -0
  89. package/scripts/validate_skills.py +465 -0
  90. package/skills/nobrainer-auto-fine-tune/SKILL.md +79 -0
  91. package/skills/nobrainer-auto-fine-tune/references/capacity-audit.md +49 -0
  92. package/skills/nobrainer-autoimprove/SKILL.md +212 -0
  93. package/skills/nobrainer-autoimprove/scripts/count_words.py +22 -0
  94. package/skills/nobrainer-browser/SKILL.md +234 -0
  95. package/skills/nobrainer-browser/references/cdp-profile-restart.md +93 -0
  96. package/skills/nobrainer-build/SKILL.md +173 -0
  97. package/skills/nobrainer-codex-context/SKILL.md +111 -0
  98. package/skills/nobrainer-codex-context/scripts/codex_context.py +175 -0
  99. package/skills/nobrainer-decide/SKILL.md +198 -0
  100. package/skills/nobrainer-dispatcher/SKILL.md +256 -0
  101. package/skills/nobrainer-dispatcher/agents/openai.yaml +4 -0
  102. package/skills/nobrainer-rca/SKILL.md +163 -0
  103. package/skills/nobrainer-research/SKILL.md +131 -0
  104. package/skills/nobrainer-review/SKILL.md +195 -0
  105. package/skills/nobrainer-security/SKILL.md +166 -0
  106. package/skills/nobrainer-sessions/SKILL.md +223 -0
  107. package/skills/nobrainer-sessions/references/evidence-receipt.md +55 -0
  108. package/skills/nobrainer-sessions/references/protocol.md +337 -0
  109. package/skills/nobrainer-sessions/references/session-restart.md +255 -0
  110. package/skills/nobrainer-sessions/scripts/evidence_receipt.py +373 -0
  111. package/skills/nobrainer-sessions/scripts/restart_gate.py +140 -0
  112. package/skills/nobrainer-sessions/scripts/run_bounded.py +390 -0
  113. package/skills/nobrainer-sessions/scripts/session_title.py +43 -0
  114. package/skills/nobrainer-skill-doctor/SKILL.md +149 -0
  115. package/skills/nobrainer-skill-doctor/references/astra.md +19 -0
  116. package/skills/nobrainer-skill-doctor/scripts/audit.py +100 -0
  117. package/skills/nobrainer-skill-doctor/scripts/context_candidates.py +133 -0
  118. package/skills/nobrainer-spec-driven-development/SKILL.md +132 -0
  119. package/skills/nobrainer-spec-driven-development/references/spec-template.md +114 -0
  120. package/skills/nobrainer-team/SKILL.md +165 -0
  121. package/skills/nobrainer-team/references/team-plan.md +64 -0
  122. package/skills/nobrainer-tech-flow/SKILL.md +239 -0
  123. package/skills/nobrainer-tech-flow/VERSION +1 -0
  124. package/skills/nobrainer-tech-flow/references/autopilot.md +72 -0
  125. package/skills/nobrainer-tech-flow/references/codex-routing.md +21 -0
  126. package/skills/nobrainer-tech-flow/references/communication.md +108 -0
  127. package/skills/nobrainer-tech-flow/references/correction-hooks.md +110 -0
  128. package/skills/nobrainer-tech-flow/references/daily-update.md +28 -0
  129. package/skills/nobrainer-tech-flow/references/delivery.md +243 -0
  130. package/skills/nobrainer-tech-flow/references/idea-to-delivery.md +82 -0
  131. package/skills/nobrainer-tech-flow/references/long-run-state.md +209 -0
  132. package/skills/nobrainer-tech-flow/references/model-routing.md +83 -0
  133. package/skills/nobrainer-tech-flow/references/optional-decisions.md +83 -0
  134. package/skills/nobrainer-tech-flow/references/project-start.md +22 -0
  135. package/skills/nobrainer-tech-flow/references/routing.md +125 -0
  136. package/skills/nobrainer-tech-flow/references/setup.md +180 -0
  137. package/skills/nobrainer-tech-flow/references/worktrees.md +138 -0
  138. package/skills/nobrainer-tech-flow/scripts/check_flow_update.py +298 -0
  139. package/skills/nobrainer-tech-flow/scripts/cleanup_worktree.py +327 -0
  140. package/skills/nobrainer-tech-flow/scripts/decision_config.py +107 -0
  141. package/skills/nobrainer-tech-flow/scripts/flow_guard.py +252 -0
  142. package/skills/nobrainer-tech-flow/scripts/typed_decisions.py +210 -0
  143. package/skills/nobrainer-wiki/SKILL.md +164 -0
  144. package/skills/nobrainer-wiki/references/setup.md +88 -0
  145. package/skills/nobrainer-writing/SKILL.md +272 -0
  146. package/skills/nobrainer-writing/agents/openai.yaml +4 -0
  147. package/skills/nobrainer-writing/references/brief-artifacts.md +357 -0
  148. package/skills/nobrainer-writing/references/plain-technical-english.md +76 -0
  149. package/skills/nobrainer-writing/references/research.md +86 -0
package/docs/TRY_IT.md ADDED
@@ -0,0 +1,109 @@
1
+ # Try nobrainer-tech-flow on one task
2
+
3
+ Install from the [reviewed source and dry-run](INSTALL.md), restart your client,
4
+ and check that it can discover `nobrainer-tech-flow`. A successful file installation
5
+ is not a successful model run. Pick one small trial and inspect the artifact.
6
+
7
+ ## An everyday task
8
+
9
+ In an ordinary empty folder, ask:
10
+
11
+ > Use NBFlow. Create invitation.md for a free board-game evening on 18 September
12
+ > at 18:00, ending at 21:00, at the community room. Bring a game if you have one;
13
+ > newcomers are welcome. Also make checklist.md with the organizer's preparation
14
+ > tasks. Use only these supplied details. Do not initialize Git or add frameworks.
15
+
16
+ Accept when both files exist, the invitation preserves every supplied fact and
17
+ the checklist is usable. Reject invented addresses, prices or contact details,
18
+ unrequested project scaffolding, and claims about checks that were not performed.
19
+
20
+ ## A small code correction
21
+
22
+ In a disposable copy of a project with a known failing test, ask:
23
+
24
+ > Use NBF. Reproduce the failing test, find the cause and make the smallest
25
+ > correction. Run the affected test and the project's required checks. Keep
26
+ > unrelated files unchanged. Report the actual results and remaining uncertainty.
27
+
28
+ Accept when the original failure is reproduced, the correction addresses it,
29
+ the relevant checks pass and the diff stays in scope. A plausible explanation
30
+ or a green unrelated test is insufficient. Do not use a live incident as a demo.
31
+
32
+ ## From an idea to a working result
33
+
34
+ In a disposable folder, ask:
35
+
36
+ > Use NBFlow. I have an idea for an offline CLI that turns Markdown meeting
37
+ > notes into an action-item list. Use SDD and deliver it end to end. Use Python
38
+ > standard library only, keep input files unchanged, and make no network calls.
39
+ > Choose routine reversible details. Ask only when a missing choice changes
40
+ > scope or acceptance. Do not initialize Git, publish or send messages.
41
+
42
+ Accept when the reviewed spec connects each requirement to an observable check,
43
+ the CLI works on the agreed examples, and every required acceptance ID has
44
+ inspected evidence. Assumptions must be visible. A spec alone is not delivery;
45
+ a spec-only request must not start implementation.
46
+
47
+ ## Make a technical procedure easier to follow
48
+
49
+ > Use nb-write. Rewrite this English runbook as a clear procedure.
50
+ > Keep commands and identifiers exact. Preserve prerequisites, permissions,
51
+ > uncertainty and safety conditions. Return the finished text.
52
+
53
+ With no other mode selected, English technical artifacts use TECHNICAL by
54
+ default. Accept when the instructions are easier to follow and no material
55
+ meaning has changed. Test conditions and exceptions, not just sentence length. This mode
56
+ uses selected clarity principles informed by ASD-STE100, not a full compliance
57
+ checker. It does not replace qualified review of safety-critical instructions.
58
+
59
+ ## Carry an approved plan through autopilot
60
+
61
+ Use a small, approved plan in a disposable project. Give each task a checkable
62
+ result, name the required validation commands, and state any commit or publication
63
+ authority separately. Then ask:
64
+
65
+ > Use nobrainer-tech-flow in autopilot to execute this approved plan through its
66
+ > full acceptance criteria. Keep the goal and remaining work in the existing plan.
67
+ > Execute the next ready unit, run its relevant checks, fix confirmed defects and
68
+ > continue without asking me to say "continue". Use independent native workers
69
+ > and review when useful and available. Keep unrelated work unchanged. Ask only
70
+ > for a material missing decision or an action that still needs authorization.
71
+ > Finish with the delivered result, actual checks and anything still unverified.
72
+
73
+ Accept when every required result has inspected evidence, corrections have been
74
+ rechecked, and no required task quietly disappears. If one unit is blocked,
75
+ independent ready work should continue. An exhausted retry or missing approval
76
+ keeps the affected result incomplete; it is not a reason to invent success.
77
+
78
+ Autopilot guides the active client. It does not install an unattended runner,
79
+ reopen a closed client or create a scheduler. Checkpoints support a later resume;
80
+ automatic continuation needs a supported host mechanism and its own authorization.
81
+ See [delivery](../skills/nobrainer-tech-flow/references/delivery.md) and
82
+ [long-run state](../skills/nobrainer-tech-flow/references/long-run-state.md).
83
+ The [autopilot loop](../skills/nobrainer-tech-flow/references/autopilot.md) defines
84
+ the next-unit decisions and whole-goal completion check.
85
+
86
+ ## Keep agent coordination and user language separate
87
+
88
+ > Use NBFlow. Review this small project without changing files. If useful,
89
+ > delegate one bounded review. Answer me in Polish. Preserve exact commands,
90
+ > error messages and test evidence, and do not claim a check ran unless it did.
91
+
92
+ Agent assignments and reports default to plain technical English. The final
93
+ answer should be natural Polish with the same findings, constraints and
94
+ uncertainty. Explicitly request another worker-report language to override that
95
+ default. Required schemas and quoted text must remain intact. No translation
96
+ provider or English draft pass is needed.
97
+
98
+ ## Evaluate the result
99
+
100
+ Record the client and version, model and effort, source commit, task, files
101
+ changed, human interventions, checks and elapsed time. Record tokens or cost
102
+ only when the client reports them; otherwise use UNKNOWN.
103
+
104
+ Compare the same task with the same client/model/settings without this suite.
105
+ Keep failed runs and repeated trials. A single satisfying result is useful
106
+ feedback, not evidence of universal superiority or cheaper-model equivalence.
107
+
108
+ If a task hangs or floods output, see the optional [bounded command runner](BOUNDED_RUNNER.md).
109
+ It controls a command's execution; the task's acceptance still needs inspection.
@@ -0,0 +1,81 @@
1
+ # v1.0.0 publication and current-tag readback
2
+
3
+ Original release verified at: `2026-08-28T08:02Z`. Current tag and archive
4
+ reverified at: `2026-08-30T21:03Z`.
5
+
6
+ ```text
7
+ VERSION: 1.0.0
8
+ TAG: v1.0.0
9
+ COMMIT_SHA: 55c49f40d7dc4ebe900f139711cd46617c706233
10
+ TREE_SHA: 4d2f02aae304e8c2ff88975fc194d115f7a1c57a
11
+ RELEASE_URL: https://github.com/nobrainer-tech/nobrainer-tech-skills/releases/tag/v1.0.0
12
+ TARBALL_URL: https://api.github.com/repos/nobrainer-tech/nobrainer-tech-skills/tarball/v1.0.0
13
+ TARBALL_FILENAME: v1.0.0.tar.gz
14
+ TARBALL_DOWNLOADED_AT: 2026-08-28T08:02Z
15
+ TARBALL_SHA256: 1d632f98c57b39246ab7c377da729cf1327dd32a9d0feb5bfe465dfbce36849b
16
+ PR_CI_RUN: 33153619658
17
+ MAIN_CI_RUN: 33153660970
18
+ GITHUB_RELEASE_IMMUTABLE: false
19
+ ARCHIVE_FILE_MATCH: PASS
20
+ ARCHIVE_SKILL_COUNT: 9
21
+ ARCHIVE_TESTS: 57/57 PASS
22
+ INSTALL_CLIENT: codex
23
+ INSTALL_READBACK: PASS
24
+ ```
25
+
26
+ ## GitHub state
27
+
28
+ - Release: [NoBrainer Tech Skills 1.0](https://github.com/nobrainer-tech/nobrainer-tech-skills/releases/tag/v1.0.0)
29
+ - Published at: `2026-08-28T08:01:36Z`
30
+ - Tag: `v1.0.0`
31
+ - Current tag: `55c49f40d7dc4ebe900f139711cd46617c706233`
32
+ - The release API retains a historical target from an earlier metadata rewrite;
33
+ it is not the current tag ref.
34
+ - Git tree: `4d2f02aae304e8c2ff88975fc194d115f7a1c57a`
35
+ - Source PR: [#21](https://github.com/nobrainer-tech/nobrainer-tech-skills/pull/21)
36
+ - PR CI: [run 33153619658](https://github.com/nobrainer-tech/nobrainer-tech-skills/actions/runs/33153619658), Ubuntu and macOS passed
37
+ - Merged-main CI: [run 33153660970](https://github.com/nobrainer-tech/nobrainer-tech-skills/actions/runs/33153660970), the tree-equivalent pre-rewrite commit passed
38
+ - GitHub API flags: `isDraft: false`, `isPrerelease: false`, `isImmutable: false`
39
+
40
+ The release and tag are not immutable according to GitHub's API. The current
41
+ tag preserves the verified tree across a metadata-only history rewrite. The
42
+ historical CI binds the same tree, not the current commit identity. Pin the full
43
+ current commit SHA above when a moving tag would be an unacceptable
44
+ supply-chain risk.
45
+
46
+ ## Downloaded source archive
47
+
48
+ - Tarball SHA-256 at the current readback:
49
+ `1d632f98c57b39246ab7c377da729cf1327dd32a9d0feb5bfe465dfbce36849b`
50
+ - Every extracted file matched the corresponding path in the release Git tree.
51
+ - No tracked path was missing and no extra file was present.
52
+ - Exactly nine `SKILL.md` files were present.
53
+ - Both repository validators passed from the extracted archive.
54
+ - Unit suite from the extracted archive: `57/57 PASS`.
55
+ - The public archive's installer completed an isolated Codex-target install of
56
+ all nine skills; a second apply reported nine unchanged links, and readback
57
+ resolved every `SKILL.md` to the extracted release source.
58
+
59
+ Download and checksum command used for this readback:
60
+
61
+ ```bash
62
+ curl --fail --location --proto '=https' --tlsv1.2 \
63
+ --output v1.0.0.tar.gz \
64
+ https://api.github.com/repos/nobrainer-tech/nobrainer-tech-skills/tarball/v1.0.0
65
+ shasum -a 256 v1.0.0.tar.gz
66
+ ```
67
+
68
+ The tarball checksum records this download, but the full commit SHA remains the
69
+ canonical content identity.
70
+
71
+ ## Proof boundary
72
+
73
+ This proves publication, source download and isolated portable installation from
74
+ the GitHub release. It does not prove npm publication, client marketplace
75
+ distribution, native client loading, clean-session routing, production behavior
76
+ or buyer usefulness. Those levels remain separate in
77
+ [the compatibility matrix](../COMPATIBILITY.md).
78
+
79
+ Rollback for an installed copy is to remove only links created from this source
80
+ or restore the project's previously reviewed Git ref. There is no earlier stable
81
+ NoBrainer Tech Skills release to pin automatically.
@@ -0,0 +1,85 @@
1
+ # v1.1.0 publication and current-tag readback
2
+
3
+ Original release verified at: `2026-08-28T14:19Z`. Current tag and archive
4
+ reverified at: `2026-08-30T21:03Z`.
5
+
6
+ ```text
7
+ VERSION: 1.1.0
8
+ TAG: v1.1.0
9
+ COMMIT_SHA: 711be31d654835a04ef8c70674c3e493aeb2da8a
10
+ TREE_SHA: 085e6d083b87f85ee5dc89816bd6818fa5aa6f1d
11
+ RELEASE_URL: https://github.com/nobrainer-tech/nobrainer-tech-skills/releases/tag/v1.1.0
12
+ TARBALL_URL: https://github.com/nobrainer-tech/nobrainer-tech-skills/archive/refs/tags/v1.1.0.tar.gz
13
+ TARBALL_FILENAME: v1.1.0.tar.gz
14
+ TARBALL_DOWNLOADED_AT: 2026-08-28T14:18Z
15
+ TARBALL_SHA256: 1a05186234487df1fa2381a291b53141a5fad6347d24f1638fc86b961abb7f5e
16
+ PR_CI_RUN: 33179388586
17
+ MAIN_CI_RUN: 33179433836
18
+ GITHUB_RELEASE_IMMUTABLE: false
19
+ ARCHIVE_FILE_MATCH: PASS
20
+ ARCHIVE_SKILL_COUNT: 13
21
+ ARCHIVE_TESTS: 70/70 PASS
22
+ ARCHIVE_QUICK_VALIDATE: 13/13 PASS
23
+ ARCHIVE_SECRET_SCAN: PASS
24
+ INSTALL_CLIENT: agents
25
+ INSTALL_MODE: copy
26
+ INSTALL_READBACK: PASS
27
+ ```
28
+
29
+ ## GitHub state
30
+
31
+ - Release: [NoBrainer Tech Skills v1.1.0](https://github.com/nobrainer-tech/nobrainer-tech-skills/releases/tag/v1.1.0)
32
+ - Published at: `2026-08-28T14:18:00Z`
33
+ - Current tag and source commit:
34
+ `711be31d654835a04ef8c70674c3e493aeb2da8a`
35
+ - The release API retains its historical target; it is not the current tag ref
36
+ after the metadata-only history rewrite.
37
+ - Git tree: `085e6d083b87f85ee5dc89816bd6818fa5aa6f1d`
38
+ - Source PR: [#23](https://github.com/nobrainer-tech/nobrainer-tech-skills/pull/23)
39
+ - PR CI: [run 33179388586](https://github.com/nobrainer-tech/nobrainer-tech-skills/actions/runs/33179388586), Ubuntu and macOS passed
40
+ - Merged-main CI: [run 33179433836](https://github.com/nobrainer-tech/nobrainer-tech-skills/actions/runs/33179433836), the tree-equivalent pre-rewrite commit passed
41
+ - GitHub API flags: `draft: false`, `prerelease: false`, `immutable: false`
42
+
43
+ The release and tag are not immutable according to GitHub's API. The current
44
+ tag preserves the verified tree across a metadata-only history rewrite. The
45
+ historical CI binds the same tree, not the current commit identity. Pin the full
46
+ current commit SHA above when a moving tag would be an unacceptable
47
+ supply-chain risk.
48
+
49
+ ## Downloaded source archive
50
+
51
+ - Tarball SHA-256 at the current readback:
52
+ `1a05186234487df1fa2381a291b53141a5fad6347d24f1638fc86b961abb7f5e`
53
+ - Every extracted file matched the corresponding file from `git archive v1.1.0`;
54
+ no path was missing or added.
55
+ - Exactly thirteen active skill directories and `SKILL.md` files were present.
56
+ - The canonical suite validator and unit suite passed from the archive:
57
+ `70/70 PASS`.
58
+ - All thirteen installed copies passed the platform `quick_validate.py`.
59
+ - The archive passed a fresh `gitleaks dir` scan.
60
+ - The public archive installer completed an isolated `agents` copy install of
61
+ all thirteen skills, and exact inventory/readback passed.
62
+
63
+ Download and checksum command used for this readback:
64
+
65
+ ```bash
66
+ curl --fail --location \
67
+ --output v1.1.0.tar.gz \
68
+ https://github.com/nobrainer-tech/nobrainer-tech-skills/archive/refs/tags/v1.1.0.tar.gz
69
+ shasum -a 256 v1.1.0.tar.gz
70
+ ```
71
+
72
+ The tarball checksum records this download, while the full commit SHA remains
73
+ the canonical content identity.
74
+
75
+ ## Proof boundary
76
+
77
+ This proves GitHub publication, tag-to-commit identity, source download, file
78
+ parity and isolated portable installation. It does not prove npm publication,
79
+ client marketplace distribution, native client loading, clean-session routing,
80
+ production behavior or buyer usefulness. Those levels remain separate in
81
+ [the compatibility matrix](../COMPATIBILITY.md).
82
+
83
+ Rollback is to pin `v1.0.0` or another reviewed full commit, remove only targets
84
+ created by the exact installation, and inspect any reported migration/rollback
85
+ recovery claim before exact manual cleanup.
@@ -0,0 +1,34 @@
1
+ # Version 1.10.0 - session care, naming and delegation boundaries
2
+
3
+ This release extends the existing Sessions and Ultra contracts without adding a
4
+ new permanent skill, visible conversation or background daemon.
5
+
6
+ ## Scope
7
+
8
+ - Explicit NoBrainer.Tech Flow entry invokes startup naming and health
9
+ assessment using verified session metadata where the host provides it.
10
+ - A deterministic title helper formats the observed start date and timezone but
11
+ never creates or renames a session.
12
+ - Context pressure and a materially smaller startup can qualify a safe health
13
+ rotation without inventing billing or token-savings proof.
14
+ - Native subagents are the bounded delegation path. Visible conversations require
15
+ an explicit owner request or authorized MAIN restart.
16
+ - Maintained public language uses NoBrainer.Tech Flow; technical identifiers and
17
+ package compatibility remain unchanged.
18
+ - `nobrainer-wiki` records adoption gates for external memory services: scope,
19
+ provenance, freshness, conflict handling, deletion/export, privacy, cost and
20
+ file fallback.
21
+
22
+ ## Verification boundary
23
+
24
+ The release candidate passed both validators, 165 deterministic tests,
25
+ session-title and restart-gate tests, syntax checks, whitespace checks and a
26
+ staged secret scan. The actual client startup and native fresh-session takeover
27
+ remain host-specific; unsupported hosts use a manual handoff. No universal
28
+ context-saving, billing, model-quality or automatic-transport claim is made.
29
+
30
+ ## Rollback
31
+
32
+ The rollback anchor is the published `v1.9.0` tag. Revert to that exact source
33
+ after verifying the tag and rerun its validators. Existing package/plugin IDs,
34
+ skill paths and compatibility aliases remain unchanged.
@@ -0,0 +1,25 @@
1
+ # v1.11.0 publication readback - 2026-09-11
2
+
3
+ NoBrainer.Tech Flow v1.11.0 is published at
4
+ <https://github.com/nobrainer-tech/nobrainer-tech-flow/releases/tag/v1.11.0>.
5
+
6
+ ## Source
7
+
8
+ - [PR #57](https://github.com/nobrainer-tech/nobrainer-tech-flow/pull/57)
9
+ merged at `2026-09-11T18:47:59Z`.
10
+ - Merge commit: `134788181e7d0745e94d9b1052c1605de810259c`.
11
+ - Annotated tag object: `464eaea68dc777de1f89caf3cc5aafa7db826459`.
12
+ - Release: [NoBrainer.Tech Flow v1.11.0](https://github.com/nobrainer-tech/nobrainer-tech-flow/releases/tag/v1.11.0), published at `2026-09-11T18:49:32Z`, not draft and not prerelease.
13
+ - Both Linux and macOS validation jobs for PR #57 passed. The source archive readback contained `skills/nobrainer-codex-context/SKILL.md`, its probe tests and the v1.11.0 release record.
14
+ - Downloaded source archive SHA-256: `09903b647e6eea141b83d210013c64899ee71e936e04ae53e05e52e3df5ca0e5`.
15
+
16
+ ## Limits and rollback
17
+
18
+ The local Codex CLI readback confirms that `AGENTS.md` is included by
19
+ `codex debug prompt-input`. It does not prove automatic discovery of the newly
20
+ added skill in every client because this release was not globally installed.
21
+ Model behavior, client marketplace distribution and universal runtime support
22
+ remain unverified.
23
+
24
+ Rollback source: the `v1.10.0` tag. Restore that exact ref only after checking
25
+ current working state and rerun the validators and tests.
@@ -0,0 +1,42 @@
1
+ # Version 1.11.0 - Codex project context
2
+
3
+ This source release adds one original, Codex-specific skill without copying a
4
+ third-party orchestrator or changing the existing NoBrainer.Tech Flow lifecycle.
5
+
6
+ ## Scope
7
+
8
+ - `nobrainer-codex-context` owns project-local Codex instruction discovery and
9
+ setup: root markers, `AGENTS.override.md`/`AGENTS.md`, configured fallback
10
+ names, byte-budget/truncation risk and post-edit readback.
11
+ - The bundled Python probe is read-only and returns a machine-readable context
12
+ inventory. It is not a client hook and does not mutate `~/.codex/config.toml`.
13
+ - Setup may add or update one `NOBRAINER-CODEX-CONTEXT` block in the appropriate
14
+ project `AGENTS.md` while preserving existing content and managed markers.
15
+ - Model choice, Astra/Luna role topology, session transfer, installation and
16
+ publication remain owned by existing skills or explicit owner gates.
17
+
18
+ ## Verification boundary
19
+
20
+ The candidate must pass the canonical validator, full deterministic test suite,
21
+ the three Codex-context probe fixtures, syntax checks, whitespace checks and a
22
+ release-gate review. A passing local probe proves only repository-level context
23
+ inventory behavior. It does not prove that a particular Codex client injected or
24
+ followed the discovered files; a fresh `codex debug prompt-input` or
25
+ `codex debug agents-md` readback is required for that claim.
26
+
27
+ ## Release state
28
+
29
+ ```text
30
+ SOURCE_VERSION: 1.11.0
31
+ PORTFOLIO: SIXTEEN_CANONICAL_NOBRAINER_SKILLS
32
+ GITHUB_RELEASE: CREATED
33
+ CLIENT_RUNTIME: NOT_VERIFIED
34
+ MARKETPLACE_DISTRIBUTION: NOT_VERIFIED
35
+ ```
36
+
37
+ ## Rollback
38
+
39
+ The rollback anchor is the `v1.10.0` tag. Restore that exact ref only after
40
+ checking the current working tree and rerun its validator and tests. The new
41
+ skill is additive; existing skill directories, aliases and adapters remain
42
+ unchanged.
@@ -0,0 +1,30 @@
1
+ # v1.12.0 - goal-driven delivery
2
+
3
+ This release adds GDD as an execution pattern inside the existing sixteen-skill
4
+ portfolio. The [specification](../specs/goal-driven-delivery.spec.md) defines full
5
+ outcomes, milestone acceptance, visible TODO, BUDDY authorization, operational
6
+ obstacle resolution and coordinated native workers. No separate GDD skill,
7
+ daemon or global permission override is introduced.
8
+
9
+ Codex workers use the documented Luna/max preset when supported and not overridden
10
+ by the owner. MAIN retains its selected model. Other clients retain portable
11
+ model policy. Parallelism is limited by ready independent work, native capacity
12
+ and 15 live descendants, with uncertain spawns reserving capacity and write scope.
13
+
14
+ Maintained display copy is `nobrainer-tech-flow`. Existing package IDs, paths,
15
+ aliases and historical evidence remain intact.
16
+
17
+ ## Verification boundary
18
+
19
+ The release requires both source validators, the full deterministic suite,
20
+ source secret scan and bounded independent scenario review. These demonstrate
21
+ source contracts and evaluated scenario responses; they do not establish faster
22
+ completion or reliable unattended execution across all models and clients.
23
+ Actual deployment UI, physical MFA, host-only START and account quotas remain
24
+ subject to each host's capabilities and authority.
25
+
26
+ ## Recovery
27
+
28
+ Rollback source is v1.11.0. Restore a verified source in an isolated checkout,
29
+ preserve unrelated working changes and read back installed links. No consumer
30
+ production or global configuration migration is part of this source release.
@@ -0,0 +1,46 @@
1
+ # v1.13.0 - skill doctor and evidence-bound fixes
2
+
3
+ This is the next-minor release candidate after the published `v1.12.0` source
4
+ release. It adds `nobrainer-skill-doctor` as the 17th portable skill inside
5
+ NoBrainer.Tech Flow. The doctor owns cross-project audits of skills, project
6
+ instructions and task prompts, including trigger overlap, excessive or
7
+ conflicting rules, coverage and minimal portfolio repair planning. It does not
8
+ replace Review's exact-change evidence gate or Codex-context's one-project
9
+ context setup boundary.
10
+
11
+ The candidate also carries accepted local fixes for uncertain session creation,
12
+ SDD state/template vocabulary, generic Security mode selection, Autoimprove
13
+ strict-cap lower bounds, Writing trigger scope and Browser local-first
14
+ capability discovery. Historical evaluation receipts remain internally
15
+ consistent with their original source bytes.
16
+
17
+ The bundled doctor inventory helper is read-only. It reports metadata and
18
+ candidate locations, records skipped symlink directories and traversal errors,
19
+ and does not establish semantic findings, client loading or model behavior.
20
+
21
+ ## Verification boundary
22
+
23
+ Static and repository evidence for this candidate:
24
+
25
+ - 17 canonical skill directories and matching validator/installer/adapter
26
+ inventories;
27
+ - 186/186 deterministic tests passed;
28
+ - both `scripts/validate_skills.py` modes passed;
29
+ - the helper inventory found 17 public `SKILL.md` files with `status: inventoried`
30
+ and `skipped: []`;
31
+ - Python, shell and Node syntax checks passed;
32
+ - `git diff --check` and the configured Gitleaks scan passed;
33
+ - isolated temporary copy-install and source-archive readbacks passed.
34
+
35
+ These checks prove the reviewed source tree and local contracts only. Clean
36
+ client discovery, model behavior, client loading, marketplace distribution,
37
+ production behavior and user usefulness remain `UNVERIFIED` until each exact
38
+ client or external system supplies its own readback. No real client installation,
39
+ commit, push, tag or GitHub release is part of this candidate preparation.
40
+
41
+ ## Rollback
42
+
43
+ Use the published `v1.12.0` source release as the rollback anchor. Restore only
44
+ the reviewed candidate files in an isolated checkout, preserve unrelated dirty
45
+ work and verify the previous 16-skill source and manifest state before any
46
+ client-side rollback.
@@ -0,0 +1,43 @@
1
+ # v1.14.0 - optional Jev and Laya decisions
2
+
3
+ Same seventeen-skill workflow, with an optional
4
+ typed-decision layer. Core runs with the host-selected model and no provider
5
+ calls. Setup remembers core/Jev/Laya, and explicit settings can change it later.
6
+ Remote Jev and cached local Laya return advisory suggestions with timeout,
7
+ attempt-limit and malformed-response fallback. They cannot execute actions or
8
+ approve completion.
9
+
10
+ The top-ten review
11
+ maps existing coverage and identifies optional graph, scientific and research
12
+ capabilities without importing a second workflow or requiring a skill catalog.
13
+
14
+ ## Evidence
15
+
16
+ - Baseline candidate: 198 deterministic tests passed before review fixes.
17
+ - After review fixes: 12 focused optional-decision tests passed. Full 201-test
18
+ reruns encountered existing wall-clock runner failures on a heavily loaded
19
+ host; the identical-to-1.13.0 runner suite passed all 14 tests in isolation.
20
+ Clean Linux and macOS CI subsequently passed all 201 tests on implementation
21
+ commit `8425ceafcf6bdcce3d9900c69b82fcc4c036e4f6`.
22
+ - Both validators, syntax checks and clean-tree checks passed in
23
+ [CI run 35705688851](https://github.com/nobrainer-tech/nobrainer-tech-flow/actions/runs/35705688851).
24
+ The Linux current-tree secret scan passed too.
25
+ - Live direct Jev 1.13.0 and offline cached Laya multilingual returned validated
26
+ Noul, Choice and Score on synthetic input. Apple Silicon/macOS local evidence
27
+ only. Laya's middling urgency result limits semantic quality claims.
28
+ - Workflow SVG rendered and inspected at 1600x900; no clipped new caption.
29
+ - Independent audit found setup, failure-model metadata and local-diagnostic
30
+ gaps; fixes were rechecked. The final prompt-label correction is applied.
31
+ Local cached-path loading was reverified after the fix with all three types.
32
+
33
+ Helper runtime evidence does not prove every harness loads or follows Flow.
34
+ Clean-session routing, package installation and public distribution each need
35
+ separate readback. No paid provider, download or second model is necessary for core.
36
+
37
+ ## Rollback
38
+
39
+ Set decision mode off or provider core to disable optional calls. Reinstall
40
+ the verified v1.13.0 source if the release itself needs rollback; preserve
41
+ project instructions and unrelated changes. Website preimages must be retained
42
+ before deployment. GitHub release, public website and X permalink require their
43
+ own publication readback; CI does not prove those external states.
@@ -0,0 +1,21 @@
1
+ # v1.14.1 - portable coverage and explicit limits
2
+
3
+ This follow-up incorporates three portable gaps from the top-ten source review
4
+ into existing skill owners. Team checks applicable capability coverage and
5
+ retains selected/rejected source evidence; Research reports source-by-source
6
+ coverage instead of treating missing access as an empty result; Build records
7
+ known simplification ceilings and concrete revisit triggers.
8
+
9
+ The seventeen-skill portfolio and one-model core remain unchanged. Jev and Laya
10
+ Multilingual are supported when wanted, with the existing explicit opt-in and
11
+ advisory boundary. No graph, crawler, scientific engine or proxy is claimed as
12
+ a Flow implementation.
13
+
14
+ The adoption map separates the new
15
+ contracts from existing coverage and specialist capabilities. Independent
16
+ source-contract review exercised Team, Research and Build scenarios, including
17
+ small-task and unavailable-model cases. Deterministic tests and validators
18
+ cover repository regressions; neither establishes a cross-model quality win.
19
+
20
+ Rollback: use v1.14.0 source and retain the existing project plan/evidence.
21
+ No migration, dependency installation, credential change or mode reset is needed.
@@ -0,0 +1,87 @@
1
+ # v1.2.0 publication and current-tag readback
2
+
3
+ Original release verified at: `2026-08-28T21:52Z`. Current tag and archive
4
+ reverified at: `2026-08-30T21:03Z`.
5
+
6
+ ```text
7
+ VERSION: 1.2.0
8
+ TAG: v1.2.0
9
+ COMMIT_SHA: 46feb1e95567db6967ea718cb75051c507ada02f
10
+ TREE_SHA: 4288d4cd89a4a10cad55b790793ed1353f711ebc
11
+ RELEASE_URL: https://github.com/nobrainer-tech/nobrainer-tech-skills/releases/tag/v1.2.0
12
+ TARBALL_URL: https://github.com/nobrainer-tech/nobrainer-tech-skills/archive/refs/tags/v1.2.0.tar.gz
13
+ TARBALL_FILENAME: v1.2.0.tar.gz
14
+ TARBALL_SHA256: 896ddd4a2fd77839a1bf9599d85fc2306cb135bddf3b2dcf521f11e57ddcb39f
15
+ PR_CI_RUN: 33214075697
16
+ MAIN_CI_RUN: 33214127416
17
+ GITHUB_RELEASE_IMMUTABLE: false
18
+ TAG_PROTECTION_STATUS: NOT_VERIFIED
19
+ ARCHIVE_FILE_MATCH: PASS
20
+ ARCHIVE_SKILL_COUNT: 15
21
+ REPOSITORY_TESTS_PASSED: 86
22
+ REPOSITORY_TESTS_FAILED: 0
23
+ REPOSITORY_TESTS_TOTAL: 86
24
+ REPOSITORY_TESTS_STATUS: PASS
25
+ ARCHIVE_TESTS_PASSED: 85
26
+ ARCHIVE_TESTS_FAILED: 1
27
+ ARCHIVE_TESTS_TOTAL: 86
28
+ ARCHIVE_TESTS_STATUS: FAIL
29
+ ARCHIVE_TEST_FAILURE: frozen attributes test required the source repository .git directory
30
+ ARCHIVE_QUICK_VALIDATE: 15/15 PASS
31
+ ARCHIVE_SOURCE_SECRET_SCAN: PASS
32
+ INSTALL_READBACK: PASS
33
+ ACCEPTANCE: FAIL
34
+ PLANNED_REMEDIATION: v1.2.1
35
+ PLANNED_REMEDIATION_STATUS_AT_READBACK: UNVERIFIED_RELEASE_CANDIDATE
36
+ SUPERSEDED_BY: v1.2.1
37
+ SUPERSEDING_RELEASE_ACCEPTANCE: PASS
38
+ ROLLBACK_COMMIT_SHA: 711be31d654835a04ef8c70674c3e493aeb2da8a
39
+ ROLLBACK_TARBALL_SHA256: 1a05186234487df1fa2381a291b53141a5fad6347d24f1638fc86b961abb7f5e
40
+ ```
41
+
42
+ ## GitHub state
43
+
44
+ - Release: [NoBrainer Tech Skills v1.2.0](https://github.com/nobrainer-tech/nobrainer-tech-skills/releases/tag/v1.2.0)
45
+ - Published at: `2026-08-28T21:49:31Z`
46
+ - Current tag and source commit:
47
+ `46feb1e95567db6967ea718cb75051c507ada02f`
48
+ - The release API retains its historical target; it is not the current tag ref
49
+ after the metadata-only history rewrite.
50
+ - Git tree: `4288d4cd89a4a10cad55b790793ed1353f711ebc`
51
+ - Source PR: [#25](https://github.com/nobrainer-tech/nobrainer-tech-skills/pull/25)
52
+ - PR CI: [run 33214075697](https://github.com/nobrainer-tech/nobrainer-tech-skills/actions/runs/33214075697), Ubuntu and macOS passed
53
+ - Merged-main CI: [run 33214127416](https://github.com/nobrainer-tech/nobrainer-tech-skills/actions/runs/33214127416), the tree-equivalent pre-rewrite commit passed
54
+ - GitHub release API flags: `draft: false`, `prerelease: false`,
55
+ `immutable: false`
56
+
57
+ The release object is not immutable according to GitHub's API. Tag protection
58
+ was not independently verified. The current tag preserves the same tree across
59
+ a metadata-only history rewrite; historical CI binds that tree, not the current
60
+ commit identity. GitHub documents that tag archives can change when a tag
61
+ moves, so pin the full current commit SHA above when that risk is unacceptable.
62
+
63
+ ## Downloaded source archive
64
+
65
+ The downloaded tarball matched `git archive v1.2.0` exactly, contained all
66
+ fifteen skills, passed the canonical validator, passed all fifteen installed
67
+ copy validations and passed a fresh source-only Gitleaks scan. One of 86 unit
68
+ tests failed because it ran `git check-attr` in the archive root even though a
69
+ GitHub source archive intentionally contains no `.git` directory.
70
+
71
+ The repository suite passed all 86 tests before publication; the downloaded
72
+ archive then exposed the portability defect. This is not evidence that the
73
+ frozen payload policy itself was wrong. `v1.2.1` preserves the actual
74
+ `git check-attr` assertion in an isolated Git repository, adds a full
75
+ `.git`-less acceptance regression and passed its own independent archive
76
+ readback; see [the v1.2.1 evidence](v1.2.1.md).
77
+
78
+ ## Proof boundary
79
+
80
+ `v1.2.0` is published but fails the package's full acceptance contract because
81
+ the archive-native unit suite is not green. Use the accepted `v1.2.1` release
82
+ at full commit `0010140d19a7ff847dff776569772ef04d82c314`, or roll back to the
83
+ accepted `v1.1.0` commit `711be31d654835a04ef8c70674c3e493aeb2da8a`.
84
+ The commit SHA is the content identity; GitHub may regenerate a compressed
85
+ archive with different outer bytes. This evidence does not claim client
86
+ marketplace distribution, native client loading, clean-session routing,
87
+ production behavior or buyer usefulness.