@e4779/skills 0.0.0-stage → 0.1.0

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 (57) hide show
  1. package/chrome-devtools/README.md +76 -0
  2. package/chrome-devtools/SKILL.md +277 -0
  3. package/chrome-devtools/references/recipes.md +379 -0
  4. package/chrome-devtools/references/setup.md +138 -0
  5. package/drawio/SKILL.md +395 -0
  6. package/gh-cli/SKILL.md +83 -0
  7. package/grill/SKILL.md +72 -0
  8. package/layout-check/README.md +32 -0
  9. package/layout-check/SKILL.md +416 -0
  10. package/layout-check/scripts/layoutProbe.mjs +539 -0
  11. package/mcpls/SKILL.md +261 -0
  12. package/mcpls/references/configuration.md +131 -0
  13. package/mcpls/references/install.md +13 -0
  14. package/mcporter/SKILL.md +90 -0
  15. package/mindmap/SKILL.md +46 -0
  16. package/mindmap/references/drawio-recipe.md +60 -0
  17. package/mindmap/references/mindmap-design.md +63 -0
  18. package/opensource-contribution/SKILL.md +119 -0
  19. package/package.json +29 -5
  20. package/shell-taste/.shellcheckrc +16 -0
  21. package/shell-taste/SKILL.md +95 -0
  22. package/shell-taste/references/handbook/dialects.md +168 -0
  23. package/shell-taste/references/handbook/dom/01-output-logging.md +245 -0
  24. package/shell-taste/references/handbook/dom/02-errors-exit.md +48 -0
  25. package/shell-taste/references/handbook/dom/03-installation-setup.md +179 -0
  26. package/shell-taste/references/handbook/dom/04-network.md +384 -0
  27. package/shell-taste/references/handbook/dom/05-files-paths.md +74 -0
  28. package/shell-taste/references/handbook/dom/06-checks-validation.md +127 -0
  29. package/shell-taste/references/handbook/dom/07-system-administration.md +103 -0
  30. package/shell-taste/references/handbook/dom/08-privilege-users.md +58 -0
  31. package/shell-taste/references/handbook/dom/09-text-data.md +62 -0
  32. package/shell-taste/references/handbook/dom/10-terminal-ux.md +60 -0
  33. package/shell-taste/references/handbook/dom/11-process-control.md +28 -0
  34. package/shell-taste/references/handbook/dom/12-uncategorized.md +161 -0
  35. package/shell-taste/references/handbook/dom/13-containers.md +13 -0
  36. package/shell-taste/references/handbook/dom/14-agents-ai-tooling.md +13 -0
  37. package/shell-taste/references/handbook/family-index.md +230 -0
  38. package/shell-taste/references/handbook/index.md +41 -0
  39. package/skill-author/SKILL.md +86 -0
  40. package/sobesednik/SKILL.md +98 -0
  41. package/sobesednik/card-template.md +23 -0
  42. package/sobesednik/personas/advocate.md +27 -0
  43. package/sobesednik/personas/analyst.md +29 -0
  44. package/sobesednik/personas/chat.md +42 -0
  45. package/sobesednik/personas/ifs.md +32 -0
  46. package/sobesednik/personas/journal/README.md +6 -0
  47. package/sobesednik/personas/nasreddin.md +29 -0
  48. package/sobesednik/personas/premortem.md +29 -0
  49. package/sobesednik/personas/rogers.md +27 -0
  50. package/sobesednik/personas/socratic.md +29 -0
  51. package/sobesednik/personas/stoic.md +27 -0
  52. package/sobesednik/references/persona-design.md +41 -0
  53. package/zai-mcp/README.md +63 -0
  54. package/zai-mcp/SKILL.md +160 -0
  55. package/zai-mcp/references/recipes.md +273 -0
  56. package/zai-mcp/references/tools.md +266 -0
  57. package/README.md +0 -3
@@ -0,0 +1,63 @@
1
+ # Mindmap design — theory into drawio practice
2
+
3
+ ## Buzan's laws (1974, "Mind Map Mastery") — digital adaptation
4
+
5
+ 1. **Central image** — the idea in the middle, bigger than everything. In
6
+ drawio: the ellipse center node, 1.5–2x the size of branch nodes.
7
+ 2. **Images/symbols** — hand-drawn pictures become shapes/icons; use
8
+ sparingly (one per major branch at most, else noise).
9
+ 3. **Color** — one color per first-level branch. Clusters become scannable;
10
+ recall improves. Apply fillColor per branch subtree.
11
+ 4. **Curved organic branches, thick to thin** — in drawio:
12
+ entityRelationEdgeStyle + curved=1, no arrows; emulate taper by
13
+ strokeWidth 2 on level-1 edges, 1 deeper.
14
+ 5. **One keyword per line** — a node holds 1-3 words, never a sentence.
15
+ Compression forces precise naming; sentences hide their atoms.
16
+ 6. **Each idea gets its own branch** — if you write "X and Y" in one node,
17
+ that is two nodes. Splitting is the thinking.
18
+ 7. **Hierarchy — Basic Ordering Ideas (BOI)** — first level = 5±2 "chapter"
19
+ branches. This is the working-memory budget (Miller 7±2).
20
+ 8. **Radiant structure** — association radiates outward from the center.
21
+ 9. **Personal style** — idiosyncratic phrasing aids memory; do not sanitize
22
+ the user's words.
23
+ 10. **Redraw/revise** — redrawing consolidates understanding. Cheap here:
24
+ re-render after rounds; it is the built-in review.
25
+
26
+ ## Layout patterns — when to use which
27
+
28
+ | Layout | Shape | Best for |
29
+ |---|---|---|
30
+ | Radial "sun" | center + fan in all directions | exploration, brainstorm, balanced association; structure still being discovered |
31
+ | Right-flowing logic tree ("grape bunch") | root at left, chains to the right | reading order, execution plans, argument chains, documentation |
32
+ | Bottom-up tree | conclusion at top, reasons below | argument/justification maps |
33
+
34
+ Both sun and grape bunch are legitimate — they answer different questions
35
+ (sun: "what is related?"; tree: "what follows from what?"). Do not mix
36
+ directions within one level. A hybrid is fine when quadrants stay coherent.
37
+
38
+ ## Sub-topic vs branching (terminology)
39
+
40
+ - **Main branches** = the BOI level: 5±2 top areas, the "chapters" of the map.
41
+ - **Branching** = the act of splitting: one idea per branch. If a node reads
42
+ "X and Y", it is two nodes — the split forces you to name each precisely.
43
+ - **Sub-topic** = any node below level 1. Depth budget: max 4 levels; deeper
44
+ → regroup or spin a sub-map (a drawio page or separate file).
45
+
46
+ ## Cognitive constraints (why the rules exist)
47
+
48
+ - Working memory = 7±2 chunks → first-level fan max 9.
49
+ - Keyword compression = forced encoding: naming an atom beats paraphrasing.
50
+ - Color clustering aids recall and rapid scanning.
51
+ - Semantic adjacency: related branches sit adjacent; crossing edges are a
52
+ layout smell, not a style choice.
53
+
54
+ ## drawio-specific tips
55
+
56
+ - Template styles: ellipse center; rounded=1;arcSize=50 topics;
57
+ partialRectangle leaves (bottom-line only); entityRelationEdgeStyle
58
+ curved edges without arrows.
59
+ - Generate via drawio MCP (get_page → modify → set_page); geometry attribute
60
+ must be as="geometry".
61
+ - Export PNG at -s 2 and review with a vision model: overlaps, truncation,
62
+ edges-through-nodes. Fix coordinates, re-export. One or two iterations
63
+ usually suffice.
@@ -0,0 +1,119 @@
1
+ ---
2
+ name: opensource-contribution
3
+ description: "Writing upstream messages that maintainers actually read: bug reports, PR/MR descriptions, issue comments, mailing-list posts. Anti-AI-slop rules, evidence discipline, structure templates, scale-to-change and style-audit checklists, follow-up etiquette. Not for internal QA defect documentation or tracker automation; MusicBrainz contributions live in musicbrainz-contribute. Использовать, когда готовим сообщение в опенсорс: багрепорт, написать в апстрим, PR description, письмо мейнтейнеру или в рассылку, контрибьюция."
4
+ ---
5
+
6
+ # opensource-contribution — upstream messages without the slop
7
+
8
+ You are writing to a busy maintainer who owes you nothing and reads at
9
+ skimming speed. Every sentence must earn its place. The contribution IS the
10
+ evidence, curated — what you filtered out is half the gift.
11
+
12
+ ## Before writing (non-negotiable)
13
+
14
+ 1. **Dedup**: search THEIR tracker + the web for the symptom. Cite near-misses
15
+ in the report ("related but distinct: #NNN — why"). A duplicate report with
16
+ good analysis is still welcome; a duplicate with none is slop.
17
+ 2. **Read their rules**: CONTRIBUTING.md, issue templates, submission article
18
+ (e.g. FreeBSD's "Writing Problem Reports"). Use THEIR fields and title
19
+ style, not your own scheme.
20
+ 3. **Calibrate to the room**: open 3 recent merged PRs / resolved issues.
21
+ Copy their format, length, and register. A kernel-list mail differs from a
22
+ GitHub issue; both differ from a GitLab MR to a small Wayland project.
23
+
24
+ ## Scale to the change
25
+
26
+ - **One-line fix** → minimal report: repro, diff, why. Resist padding.
27
+ - **Significant change** → scope consult FIRST: comment on the existing issue
28
+ / IRC / list with the design sketch BEFORE the big PR. The issue thread is
29
+ your anchor; reference it everywhere.
30
+ - **Multi-part work** → offer a reviewable series ("core → feature → tests"),
31
+ not a 13-commit blob. Ask maintainers which split they prefer.
32
+ - **Style-audit before PR**: check their code style rules and apply them to
33
+ YOUR code — naming prefixes, allocator APIs, brace style, column limit,
34
+ C standard, no-GNU-extensions. Rename your darlings to their conventions
35
+ (an enum that misses their prefix is a review-speed bump).
36
+ - **A reproducible demo/test harness is the strongest evidence** a report can
37
+ carry. If you built one, lead with it.
38
+
39
+ ## Structure templates
40
+
41
+ **Bug report** — Title: `component: observed behavior` (one line, no drama).
42
+ Then: Environment (exact versions, hw) → Reproduction (deterministic commands,
43
+ not prose) → Observed (evidence excerpts) → Expected (spec/doc citation) →
44
+ Analysis (what you ruled out, with file:line refs) → Suggested direction.
45
+
46
+ **PR/MR** — Problem (why, link the issue) → Fix (what, minimal, why this
47
+ approach) → Testing (exact commands + observed results) → Notes for review
48
+ (risks, rejected alternatives).
49
+
50
+ **Mailing list** — one context paragraph, facts, then the ask. Short.
51
+ Subject line states the conclusion, not the journey.
52
+
53
+ ## Anti-slop rules (the heart of this skill)
54
+
55
+ | Never | Instead |
56
+ |---|---|
57
+ | "Hi team!", "Hope you're doing well", "Thanks in advance!" | Start at the first fact |
58
+ | "delve", "leverage", "comprehensive", "seamless", "robust" | plain verbs: read, use, full, exact |
59
+ | "It's worth noting that", "Interestingly", "Notably" | delete the throat-clearing, keep the fact |
60
+ | "Moreover/Additionally" chains | cut, or renumber real lists |
61
+ | hedging everything ("might", "perhaps", "I believe") | commit to verified claims; mark the rest "unverified" |
62
+ | bullet-point-everything | prose for reasoning, lists for artifacts |
63
+ | restating the report in a closing summary | end at the last fact |
64
+ | apologetic opener about being new/possibly wrong | one honest scope note, if needed |
65
+ | "!" energy, emojis, bold-spam | flat declarative tone; bold for 2-3 real anchors max |
66
+ | pleasing tone: "happy to", "I'd love to", "буду рад", "feel free to", "let me know if you need" | end at the last fact — the work itself is the offer |
67
+ | pasted raw logs (500+ lines) | the 10 lines that matter + "full log attached" |
68
+
69
+ Absolute bans: invented line numbers, versions, quotes, benchmark numbers.
70
+ Every number reproducible by the reader from your repro steps. If an LLM
71
+ generated any part you could not verify, verify it or cut it — this is the
72
+ exact failure mode that got AI reports mass-banned (curl, 2025-2026).
73
+
74
+ ## Evidence discipline
75
+
76
+ - file:line with repo paths, commit hashes, counter deltas
77
+ ("+22 EXCHANGE_ID per 30s"), before/after behavior.
78
+ - Reproduction first, theory second, speculation labeled as such.
79
+ - Say what you EXCLUDED and how — negative results save maintainers hours.
80
+ - Attachments for anything over ~20 lines; name files descriptively.
81
+
82
+ ## Tone
83
+
84
+ - First person singular for what you did; no "we" unless there is a we.
85
+ - Match their language (English for most trackers) and their register.
86
+ - Respect ≠ deference: plain, direct, no honorifics, no self-flagellation.
87
+ - No pleasing tone: zero eagerness signals, no offers of further service.
88
+ The patch/report IS the gesture; servility reads as noise (or as a bot).
89
+ - Disagree with a maintainer by citing evidence, never by volume.
90
+
91
+ ## Follow-up etiquette
92
+
93
+ - Maintainer replied: answer exactly what was asked, nothing else. No
94
+ unasked-for status updates, no enthusiasm about the review.
95
+ - Silence is normal. Do not ping before ~2 weeks (or the project's own norm,
96
+ if stated). One polite ping, never two in a row.
97
+ - Asked for changes/rebase: do them, reply tersely ("Rebased onto main;
98
+ commit X dropped because Y").
99
+ - Abandoning the work: say so in the thread and close/label it yourself.
100
+
101
+ ## Pre-submit checklist
102
+
103
+ - [ ] dedup done, near-misses cited
104
+ - [ ] their template/fields used, their title style
105
+ - [ ] scale right: minimal for small, consult-first for big, series offered
106
+ - [ ] style-audit applied if code is included
107
+ - [ ] repro is deterministic (someone else can run it)
108
+ - [ ] every number/quote reproducible, nothing invented
109
+ - [ ] no greetings, no filler, no AI-tell vocabulary
110
+ - [ ] prose where reasoning, lists where artifacts
111
+ - [ ] attachments named, oversized logs trimmed
112
+ - [ ] read it aloud once: would YOU triage this in 60 seconds?
113
+
114
+ ## Sources
115
+
116
+ - Simon Tatham, "How to Report Bugs Effectively" (structure, "show me" ethos)
117
+ - Daniel Stenberg, "Death by a thousand slops" (2025), "The end of the curl
118
+ bug-bounty" (2026) — what mass low-trust reports do to maintainers
119
+ - FreeBSD, "Writing FreeBSD Problem Reports" (docs.freebsd.org article)
package/package.json CHANGED
@@ -1,6 +1,30 @@
1
1
  {
2
- "name": "@e4779/skills",
3
- "version": "0.0.0-stage",
4
- "stub": true,
5
- "description": "Temporary package placeholder for staged publishing"
6
- }
2
+ "name": "@e4779/skills",
3
+ "version": "0.1.0",
4
+ "description": "Agent-agnostic skill pack (pi manifest inside; plain markdown, other agents can read it too)",
5
+ "keywords": [
6
+ "pi-package"
7
+ ],
8
+ "license": "ISC",
9
+ "author": "e4779",
10
+ "pi": {
11
+ "skills": [
12
+ "./drawio",
13
+ "./mindmap",
14
+ "./skill-author",
15
+ "./gh-cli",
16
+ "./mcpls",
17
+ "./mcporter",
18
+ "./chrome-devtools",
19
+ "./grill",
20
+ "./layout-check",
21
+ "./opensource-contribution",
22
+ "./shell-taste",
23
+ "./sobesednik",
24
+ "./zai-mcp"
25
+ ]
26
+ },
27
+ "publishConfig": {
28
+ "access": "public"
29
+ }
30
+ }
@@ -0,0 +1,16 @@
1
+ # shell-taste: the objective floor for shellcheck (≈ eslint for shell)
2
+ # Usage: shellcheck --rcfile <skill-dir>/.shellcheckrc script.sh
3
+ # Repo-wide: drop a copy at repo root as .shellcheckrc — shellcheck walks
4
+ # parent dirs from each script, so nested scripts inherit it.
5
+ # Inline pragmas work like eslint-disable: # shellcheck disable=SC2086
6
+
7
+ # report everything down to style level
8
+ severity=style
9
+
10
+ # warn on unquoted variables even where the value is "safe"
11
+ enable=quote-safe-variables
12
+
13
+ # uppercase variables must be assigned before use (catches typo'd consts)
14
+ enable=check-unassigned-uppercase
15
+
16
+ # aggressive option (verified in man): enable=all — every optional check
@@ -0,0 +1,95 @@
1
+ ---
2
+ name: shell-taste
3
+ description: "Taste layer for shell code: portable POSIX idioms vs host-shell extras, composability discipline, pre-delivery ritual — plus a generated handbook of measured idioms from ~900 real-world open-source scripts (error handling and die, check_root, usage output, cleanup traps, install logic, menus, logging) with ecosystem-consensus counts and canon exemplars. Use when writing, reviewing, or refactoring any #!/bin/sh or bash script, rc.d-style service script, or CLI glue; when asked how real-world scripts or the ecosystem handle a task, or which idiom is common vs one-off. RU: написать шелл-скрипт, идиоматичный sh, красивый шелл, переписать скрипт на sh, как принято в опенсорсных скриптах, как это делают реальные скрипты, обработка ошибок, check_root, usage, cleanup, install."
4
+ ---
5
+
6
+ # shell-taste
7
+
8
+ How to write shell *well*: dialect discipline, idioms, a ritual.
9
+ The objective floor is shellcheck — the eslint of shell (rule codes SC####,
10
+ severity levels, enable/disable config, inline pragmas) — driven by this
11
+ skill's `.shellcheckrc`.
12
+
13
+ ## 1. Dialect — portable core by default
14
+
15
+ Two tiers; pick deliberately. Host-local scripts mark `# dialect: host` on line 2.
16
+
17
+ - **posix** (default, unmarked): POSIX Issue 7 + `local`. Runs on dash,
18
+ FreeBSD sh, bash, busybox ash, mksh. No pipefail, no $'..'.
19
+ - **host**: extras of the host shell. On FreeBSD /bin/sh (probed 2026-11):
20
+ pipefail, $'..', `local` with dynamic scoping, `local -` (options scoped
21
+ to the function, restored on return). bash: same set.
22
+
23
+ Portability matrix (probed live 2026-11; dash column from upstream docs):
24
+
25
+ | feature | POSIX Iss.7 | FreeBSD sh | bash | dash | zsh |
26
+ |---|---|---|---|---|---|
27
+ | `local` | no (rationale left it for "a future version") | yes | yes | yes | yes |
28
+ | `pipefail`| no | yes | yes | NO | yes |
29
+ | `$'..'` | no | yes | yes | NO | yes |
30
+ | `local -` | no | yes | yes | — | (use `setopt local_options`) |
31
+
32
+ ksh93 has no `local` (use `typeset`).
33
+
34
+ If correctness *needs* pipefail: restructure — check exit codes explicitly
35
+ (`cmd || exit`, while-read loops) instead of leaning on a non-portable option.
36
+
37
+ ## 2. Idioms (the canon)
38
+
39
+ Verbs as functions returning via stdout; case = pattern dispatch;
40
+ `${var#pat}` / `${var%%pat}` before sed; heredocs for generated files;
41
+ pipelines as dataflow; `set --` as a worklist; metadata-vars + naming
42
+ convention (rc.subr style) for multi-script projects.
43
+
44
+ ```sh
45
+ log() { printf '%s: %s\n' "${0##*/}" "$*" >&2; }
46
+ die() { log "error: $*"; exit 1; }
47
+ cleanup() { [ -n "$tmp" ] && rm -rf "$tmp"; }
48
+ trap cleanup EXIT INT TERM
49
+ tmp=$(mktemp -d) || die mktemp
50
+ ```
51
+
52
+ ## 3. Ritual before delivering
53
+
54
+ 1. `shfmt -ln posix -s` — apply its rewrite; it enforces idiom mechanically
55
+ 2. shellcheck with this skill's rc (explicit path, nothing to copy):
56
+ `shellcheck --rcfile <this-skill-dir>/.shellcheckrc script.sh`
57
+ - portable tier: add `-s dash` — the strictest dialect, catches bashisms
58
+ (SC2039 class "In POSIX sh, ... is undefined")
59
+ - host tier: `-s bash`, or let the shebang deduce
60
+ - many scripts in one repo: drop the rc at repo root as `.shellcheckrc` —
61
+ discovered from each script's directory upward, editors inherit it too
62
+ 3. Read every unquoted $var aloud; cannot justify it → quote it
63
+ 4. If the user is taste-learning: offer an A/B variant (toolbox vs loop) and let them pick
64
+
65
+ Inline pragmas work like eslint-disable (`# shellcheck disable=SC2086`) —
66
+ use sparingly, with the reason on the same line.
67
+
68
+ ## 4. Canon (consult on demand)
69
+
70
+ pure-sh-bible (dylanaraps) · Wooledge wiki (BashPitfalls, BashFAQ/105) · kiss,
71
+ `/etc/rc.d/*` on this host.
72
+
73
+ ## 5. Handbook (generated)
74
+
75
+ Measured companion: idioms harvested from ~900 real-world open-source
76
+ scripts (die/error handling, check_root, usage output, cleanup traps,
77
+ install logic, menus, logging), with ecosystem-consensus counts and canon
78
+ exemplars. Rebuild after atlas data changes: `mk -C ~/projects/shell-atlas handbook`.
79
+
80
+ | You ask | Open | You get |
81
+ |---|---|---|
82
+ | error handling, die, exit codes | references/handbook/dom/02-errors-exit.md | family entries + copy-paste queries |
83
+ | how do real-world scripts do X / what does the ecosystem use | domain page via the slug line below | consensus counts + canon kin |
84
+ | the proper way / show the reference | domain page, canon line | top-ranked exemplar body |
85
+ | how common is this practice | domain page, consensus lines; thresholds in references/handbook/index.md | wild family names + similarity scores |
86
+ | two live variants of one task | references/handbook/dialects.md | JOB/SAME/DIFF card, bodies a/b |
87
+ | what is family X / where does check_root appear | references/handbook/family-index.md | anchor → domain page → query |
88
+
89
+ Domains: `output-logging` `errors-exit` `installation-setup` `network` `files-paths` `checks-validation` `system-administration` `privilege-users` `text-data` `terminal-ux` `process-control` `containers` `agents-ai-tooling` `uncategorized` — one page per domain at `references/handbook/dom/<NN>-<slug>.md`.
90
+
91
+ Reading rules: counts printed on the pages are navigation aids, stamped
92
+ as-of — never quote them as universal facts. Canon and wild family names
93
+ differ ("Check Root Privileges" vs "Privilege Check"): grep family-index
94
+ by word root and desc words, not by exact name.
95
+
@@ -0,0 +1,168 @@
1
+ <!-- GENERATED by shell-atlas bin/handbook (mk handbook). Hand edits WILL be overwritten. as-of: 5539bbbe9ebc 2026-11-08 -->
2
+
3
+ # Dialect cards — same job, different spelling
4
+
5
+ 18 measured pairs (top archetypes of the dialect band). Verdict lines are verbatim from work/wild/dialect-cards.toon (audit diffs them against the jsonl mirror); bodies resolve by coordinate under work/wild/units/.
6
+
7
+ ### change_color
8
+
9
+ sim 0.8996 (observed pair similarity).
10
+
11
+ JOB: Update configuration files for colors | SAME: Both functions modify polybar and rofi configurations | DIFF: Function A replaces specific color values using sed and overwrites the entire rofi configuration file | Function B replaces specific color values using sed for polybar and overwrites the entire rofi configuration file | Function C is not present. | Function D is not present. | Function
12
+
13
+ - a: `work/wild/units/adi1090x-polybar-themes/0158-change_color.sh`
14
+ - b: `work/wild/units/thegoodhackertv-hackerpwm/0003-change_color.sh`
15
+
16
+ ### rofi_cmd
17
+
18
+ sim 0.8901 (observed pair similarity).
19
+
20
+ JOB: Launching a rofi menu | SAME: rofi -dmenu with user prompt and uptime message | DIFF: The functions are identical in logic and parameters. They differ only in the way they are formatted and indented.
21
+
22
+ - a: `work/wild/units/adi1090x-rofi/0054-rofi_cmd.sh`
23
+ - b: `work/wild/units/stringhost-kalighost/0001-rofi_cmd.sh`
24
+
25
+ ### error
26
+
27
+ sim 0.8781 (observed pair similarity).
28
+
29
+ JOB: Error handling | SAME: Exit with status 1 | DIFF: A uses a ANSI color code variable, while B uses raw escape sequences.
30
+
31
+ - a: `work/wild/units/alirezarezvani-claude-code-tresor/0003-error.sh`
32
+ - b: `work/wild/units/adam-sizzler-marz-reverse-proxy/0004-error.sh`
33
+
34
+ ### die
35
+
36
+ sim 0.8996 (observed pair similarity).
37
+
38
+ JOB: Terminate script | SAME: Prints to stderr and exits with 1 | DIFF: A uses ANSI color codes for formatting.
39
+
40
+ - a: `work/wild/units/andrewnakas-skate3-ios-setup/0008-die.sh`
41
+ - b: `work/wild/units/solaceproducts-solace-agent-mesh/0003-die.sh`
42
+
43
+ ### fail
44
+
45
+ sim 0.8782 (observed pair similarity).
46
+
47
+ JOB: Terminate script execution | SAME: Exit with status 1 | DIFF: B redirects output to stderr and A does not.
48
+
49
+ - a: `work/wild/units/automazeio-ccpm/0011-fail.sh`
50
+ - b: `work/wild/units/ublue-os-ucore/0001-fail.sh`
51
+
52
+ ### cleanup
53
+
54
+ sim 0.8971 (observed pair similarity).
55
+
56
+ JOB: Deleting worktrees and branches | SAME: Removing directories and branches | DIFF: A is a simple directory deletion, while B uses git-specific commands to remove worktrees and branches.
57
+
58
+ - a: `work/wild/units/binpash-try/0010-cleanup.sh`
59
+ - b: `work/wild/units/fletchgqc-agentbox/0001-cleanup.sh`
60
+
61
+ ### _request
62
+
63
+ sim 0.895 (observed pair similarity).
64
+
65
+ JOB: Executing a curl request | SAME: Docker container status check and curl command structure | DIFF: Container linking and target hostnames for the request.
66
+
67
+ - a: `work/wild/units/docker-library-official-images/0079-_request.sh`
68
+ - b: `work/wild/units/nginx-docker-nginx/0010-_request.sh`
69
+
70
+ ### info
71
+
72
+ sim 0.878 (observed pair similarity).
73
+
74
+ JOB: Logging status messages | SAME: Both use printf to output a formatted string | DIFF: A uses a variable for color codes and A uses a shell variable for color codes. (Wait, I need to make sure I'm correct. A uses `_CLR_GREEN` and `_CLR_RESET`, while B uses `GREEN` and `GREEN`? No, B uses `GREEN` and `RESET`. Let me re-read.) 1. A uses `_CLR_GREEN` and `_CLR_RESET`. 2. B uses `GREEN` and `RESET`
75
+
76
+ - a: `work/wild/units/prismml-eng-bonsai-demo/0014-info.sh`
77
+ - b: `work/wild/units/jnmetacode-agency-agents-zh/0001-info.sh`
78
+
79
+ ### fetch
80
+
81
+ sim 0.8978 (observed pair similarity).
82
+
83
+ JOB: Downloading files | SAME: curl and die commands | DIFF: A pipes to tar for extraction, while B downloads a single file to a disk.
84
+
85
+ - a: `work/wild/units/ravindu644-android-kernel-tutorials/0003-fetch.sh`
86
+ - b: `work/wild/units/xiaolegun-kernelsu_action/0026-fetch.sh`
87
+
88
+ ### usage
89
+
90
+ sim 0.8907 (observed pair similarity).
91
+
92
+ JOB: Display usage instructions | SAME: sed commands for parsing and replacing | DIFF: There is no substantive difference between the two functions. They are identical.
93
+
94
+ - a: `work/wild/units/sobolevn-git-secret/0008-usage.sh`
95
+ - b: `work/wild/units/bats-core-bats-core/0008-usage.sh`
96
+
97
+ ### warn
98
+
99
+ sim 0.8997 (observed pair similarity).
100
+
101
+ JOB: Issue a warning | SAME: Both use printf to output a colorized warning message | DIFF: A uses a character-style prefix and character-style prefix, while B uses a bracketed "!!" prefix. (Wait, I need to make sure the shell script is thesaurus-ly correct and concise.) Let me re-read the "DIFF" part. A uses `!`, B uses `[!!]`. 100% correct. Let me re- the "DIFF" part. A uses `!`, B uses `[!!]`. 1
102
+
103
+ - a: `work/wild/units/peonping-peon-ping/0029-warn.sh`
104
+ - b: `work/wild/units/jnmetacode-agency-agents-zh/0002-warn.sh`
105
+
106
+ ### err
107
+
108
+ sim 0.8859 (observed pair similarity).
109
+
110
+ JOB: Error reporting | SAME: Output to stderr | DIFF: Variable names for color codes differ.
111
+
112
+ - a: `work/wild/units/jnmetacode-agency-agents-zh/0043-err.sh`
113
+ - b: `work/wild/units/prismml-eng-bonsai-demo/0016-err.sh`
114
+
115
+ ### command_exists
116
+
117
+ sim 0.8975 (observed pair similarity).
118
+
119
+ JOB: Check if a command exists | SAME: command -v "$1" >/dev/null 2>&1 | DIFF: There is no substantive difference between the two functions. They are functionally identical.
120
+
121
+ - a: `work/wild/units/lmc999-regionrestrictioncheck/0002-command_exists.sh`
122
+ - b: `work/wild/units/yeahwu-v2ray-wss/0022-command_exists.sh`
123
+
124
+ ### main
125
+
126
+ sim 0.8429 (observed pair similarity).
127
+
128
+ JOB: Software installation | SAME: Automated script execution | DIFF: Script A handles complex configuration and service management, while Script B focuses on a linear, file-based binary installation.
129
+
130
+ - a: `work/wild/units/xtls-xray-install/0056-main.sh`
131
+ - b: `work/wild/units/lewislulu-terminal-setup/0001-main.sh`
132
+
133
+ ### pass
134
+
135
+ sim 0.8277 (observed pair similarity).
136
+
137
+ JOB: Print success messages | SAME: Output format and content | DIFF: Function A includes a counter and increments it.
138
+
139
+ - a: `work/wild/units/games-on-whales-gow/0003-pass.sh`
140
+ - b: `work/wild/units/0xax-asm/0002-pass.sh`
141
+
142
+ ### ok
143
+
144
+ sim 0.885 (observed pair similarity).
145
+
146
+ JOB: Print status message | SAME: Output format with prefix and arguments | DIFF: A uses a plain text output with a added counter, but B uses ANSI escape codes for green color.
147
+
148
+ - a: `work/wild/units/games-on-whales-gow/0008-ok.sh`
149
+ - b: `work/wild/units/andrewnakas-skate3-ios-setup/0006-ok.sh`
150
+
151
+ ### log
152
+
153
+ sim 0.8747 (observed pair similarity).
154
+
155
+ JOB: Logging messages | SAME: Output format with timestamp | DIFF: B uses printf for formatting and a different date format, while A uses echo and a1zmc format. (Wait, I need to meed the strict format) -> JOB: Logging messages | SAME: Output format with timestamp | DIFF: B uses printf for formatting and a different date format, while A uses echo and a1zmc format. (Wait, I meed to meed the strict f
156
+
157
+ - a: `work/wild/units/hamelsmu-claude-review-loop/0007-log.sh`
158
+ - b: `work/wild/units/aaddrick-claude-desktop-debian/0001-log.sh`
159
+
160
+ ### check_root
161
+
162
+ sim 0.8675 (observed pair similarity).
163
+
164
+ JOB: Root privilege check | SAME: Checking the current user's UID | DIFF: A uses exiterr function and a custom message, while B uses printf with a ANSI-colored string and a hardcoded exit code.
165
+
166
+ - a: `work/wild/units/hwdsl2-setup-ipsec-vpn/0151-check_root.sh`
167
+ - b: `work/wild/units/jinqians-snell/0102-check_root.sh`
168
+