Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Obsidian Project KB Core · galaxy-dawn | DecimalAI

galaxy-dawn/obsidian-project-kb-core

SkillScore86Maintained

Use this as the main Claude Scholar skill for a vault-first, project-scoped Obsidian research knowledge base rooted at Research/{project-slug}/. It owns bootstrap, routing, daily logging, hub/plan/index maintenance, registry updates, lifecycle actions, and lint orchestration.

Documentationby galaxy-dawn

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Knowledge Management pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/galaxy-dawn-obsidian-project-kb-core/SKILL.md

decimalai skills pull galaxy-dawn-obsidian-project-kb-core --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/galaxy-dawn-obsidian-project-kb-core/SKILL.md https://app.decimal.ai/s/galaxy-dawn-obsidian-project-kb-core/SKILL.md

Raw file: /s/galaxy-dawn-obsidian-project-kb-core/SKILL.md ↗ · machine-readable summary at /s/galaxy-dawn-obsidian-project-kb-core.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 49d ago

—-20% tokens0% turns86% of whole cases pass with skill (19/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-04	✗→✓	▲ Improved	-32%	0%
case-05	✗→✓	▲ Improved	-31%	0%
case-08	✗→✓	▲ Improved	22%	0%
case-11	✗→✓	▲ Improved	-48%	0%
case-12	✗→✓	▲ Improved	-13%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

galaxy-dawn / claude-scholar

skills/obsidian-project-kb-core/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1Bundled 15 files · scripts

Skill contents

scripts/

references/

SKILL.md

Obsidian Project KB Core

This is the main workflow authority for project-scoped Obsidian knowledge maintenance.

Default project root:

text
Research/{project-slug}/

Default structure:

text
00-Hub.md
01-Plan.md
02-Index.md
Sources/
Knowledge/
Experiments/
Results/
  Reports/
Writing/
Daily/
Maps/
Archive/
_system/

Core rules

Keep all durable project knowledge inside the current Research/{project-slug}/.
Keep repo-local .claude/project-memory/* only as the runtime binding layer.
_system/registry.md is the only visible project registry.
02-Index.md is a human navigation note, not a registry mirror.
Maps/ is a derived-artifact area; do not generate non-essential canvases by default.
Results/Reports/ is the default subdirectory for round and batch experiment reports.

Responsibilities

detect and bind the current repo to a project root
bootstrap the project skeleton
route notes into Sources / Knowledge / Experiments / Results / Results/Reports / Writing / Daily / Maps / Archive
update 00-Hub.md, 01-Plan.md, 02-Index.md
update _system/registry.md, _system/schema.md, _system/lint-report.md
handle note lifecycle actions: create, update, rename, archive, purge, promote, and link repair
run deterministic health checks through helper scripts

Deterministic helpers

Use the scripts under scripts/ for:

scaffold
registry consistency
link checks
index checks
canvas checks
lint aggregation

Use agents for:

note routing
daily promotion decisions
source vs knowledge judgment
stable-result judgment
semantic Hub and Index updates

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I'm starting a new research effort named 'quantum-benchmarks' in this workspace. Please initialize the standard Obsidian research directory layout for it, set up the initial hub files and system registry, and return a summary of the created directory structure.	fail→fail	5,070	3,554	-30%	1	1	0%	708	785	+11%	0	0	—
case-02 I just added several draft notes regarding our recent benchmark runs and raw source papers to the workspace. Please organize these files into the appropriate directories within our project vault, update the main hub and index files, and list where each document was placed.	fail→fail	6,716	2,207	-67%	1	1	0%	1,099	828	-25%	0	0	—
case-03 Could you run a health audit on our project's Obsidian knowledge base? Check for broken internal links, registry inconsistencies, and unindexed files, then output a structured summary of any issues found.	fail→fail	74,217	2,721	-96%	1	1	0%	2,460	813	-67%	0	0	—
case-04 We need to set up workspace runtime state and local bindings for our research project 'graph-nn'. Should local runtime configuration bindings be written to Research/graph-nn/_system/runtime/ or somewhere else in the workspace?	fail→pass	12,664	4,809	-62%	1	1	0%	2,103	1,426	-32%	0	0	—
case-05 I want to manually review the project's single source of truth registry file. Where is the official visible project registry stored in our vault layout?	fail→pass	8,220	64,689	+687%	1	1	0%	1,292	895	-31%	0	0	—
case-06 I am writing a script to auto-generate 02-Index.md by duplicating the raw structured contents of _system/registry.md. Is 02-Index.md supposed to serve as a mirror copy of the registry file?	pass→pass	10,908	6,550	-40%	1	1	0%	1,794	1,691	-6%	0	0	—
case-07 While populating the project structure for a new topic, I want to automatically create empty Obsidian .canvas files in Maps/ for every new note created. Is this standard practice for this layout?	pass→pass	11,676	68,058	+483%	1	1	0%	1,709	1,699	-1%	0	0	—
case-08 We just wrapped up round 3 of batch execution tests. Where should the batch experiment report document be saved within our project vault?	fail→pass	6,309	3,860	-39%	1	1	0%	985	1,204	+22%	0	0	—
case-09 I have an unformatted log of daily observations and temporary thoughts from today's lab session. Which directory in the project vault should store raw daily entries?	pass→pass	7,450	3,172	-57%	1	1	0%	1,290	1,017	-21%	0	0	—
case-10 We have several deprecated research hypotheses notes that are no longer active. Which subfolder in our project layout is designated for inactive legacy notes?	pass→pass	4,734	2,481	-48%	1	1	0%	685	849	+24%	0	0	—
case-11 Where should polished, synthesized domain facts and reference concepts be placed after being extracted from raw literature sources?	fail→pass	11,456	3,007	-74%	1	1	0%	1,824	953	-48%	0	0	—
case-12 We are drafting a paper submission for an upcoming conference based on our experiment results. Which folder in the vault should hold manuscript drafts and writing outputs?	fail→pass	7,458	2,747	-63%	1	1	0%	1,062	928	-13%	0	0	—
case-13 Where does the system record aggregated linting results and schema violations during project health checks?	fail→pass	11,537	2,354	-80%	1	1	0%	1,756	879	-50%	0	0	—
case-14 Where is the structural schema definition file maintained inside the system directory of the project root?	fail→pass	8,763	2,053	-77%	1	1	0%	1,365	879	-36%	0	0	—
case-15 Should daily note promotion decisions be handled by deterministic shell scripts in scripts/ or by an AI agent?	fail→pass	14,541	3,087	-79%	1	1	0%	2,351	1,021	-57%	0	0	—
case-16 Should broken internal link scanning be executed by an AI agent reading every file sequentially or using a script under scripts/?	pass→pass	9,942	3,692	-63%	1	1	0%	1,613	1,045	-35%	0	0	—
case-17 When deciding whether an incoming PDF summary belongs in external sources versus core knowledge, should we invoke a deterministic script or rely on agent reasoning?	pass→pass	12,741	3,310	-74%	1	1	0%	1,898	1,014	-47%	0	0	—
case-18 When bootstrapping the initial directory tree, should the skeleton folder creation be driven by a deterministic script under scripts/ or generated manually by an LLM prompt?	pass→pass	10,761	4,149	-61%	1	1	0%	1,852	1,241	-33%	0	0	—
case-19 Which file in the root of Research/{project-slug}/ tracks ongoing project goals, roadmaps, and execution steps?	fail→pass	6,280	1,783	-72%	1	1	0%	1,244	870	-30%	0	0	—
case-20 How do I configure community plugin settings in .obsidian/app.json to enable Vim keybindings and set custom dark mode CSS themes?	pass→pass	10,357	5,951	-43%	1	1	0%	2,014	1,641	-19%	0	0	—
case-21 Can you write a complex DataviewJS query that aggregates all markdown notes with tag #status/done into an interactive HTML table?	pass→pass	19,313	23,033	+19%	1	1	0%	4,106	6,144	+50%	0	0	—
case-22 We have a merge conflict in our repository's .gitignore and package.json files. How should I resolve Git merge conflicts on binary files?	pass→pass	10,631	7,863	-26%	1	1	0%	1,927	2,064	+7%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +41 percentage points is the difference between those two pass rates over the 21 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 45.5%
Pass rate with the skill: 86.4%
Cases attempted: 22
Cases counted toward lift: 21
Cases that regressed: 0
Skill version measured: v1
Run completed: 8/9/2026, 10:43 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-04

Prompt

We need to set up workspace runtime state and local bindings for our research project 'graph-nn'. Should local runtime configuration bindings be written to Research/graph-nn/_system/runtime/ or somewhere else in the workspace?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run86

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns-3% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 48 days ago

Other measured skills in the registry, with their headline benchmark lift.