doc-search
v1.0.0Search a local document library (Obsidian vault, wiki, notes, etc.) using BM25 inverted index + LLM query expansion + grep for precise location. Use when the...
⭐ 0· 100·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (local document search using BM25 + LLM expansion + grep) align with the included scripts which build a local inverted index and perform BM25 ranking and JSON output. Required binaries/credentials/config are minimal/none, which is appropriate for a pure-local search tool.
Instruction Scope
The SKILL.md instructions are generally scoped to indexing and searching the user-provided docs_dir, and to running local grep for line-level hits. However there are inconsistencies: SKILL.md/README sometimes refer to storing the index at <docs_dir>/.doc-search/index.json while the actual scripts use <docs_dir>/.cache/index.json. SKILL.md instructs the agent to perform LLM query expansion (which is reasonable for an agent-based skill), but the included search script expects an already-expanded query string — the expansion step is not implemented in the Python code. These inconsistencies could cause the workflow to fail unless the agent or installer follows the correct conventions.
Install Mechanism
No install spec (instruction-only) and included Python scripts are plain text; no network downloads, no package installs, and no archive extraction. Execution is local via python3. This is low-risk from an install-mechanism perspective.
Credentials
The skill requests no environment variables, credentials, or unrelated config paths. It does read user files under the provided docs_dir and writes an index file into the docs directory (per code: <docs_dir>/.cache/index.json). That filesystem access is expected and proportionate to the stated purpose.
Persistence & Privilege
Flags show always:false and normal autonomous invocation allowed. The skill writes an index file next to the documents (per code) and does not request elevated privileges or attempt to modify other skills or system-wide settings.
Assessment
This skill is coherent for local document search and appears to run entirely on your machine (no network calls or secrets required). Before installing or running it: 1) verify where the index will be written — the code writes to <docs_dir>/.cache/index.json but README/SKILL.md also mention .doc-search (fix or be explicit about the path you want); 2) confirm the platform places the scripts at the SKILL.md path (~/.claude/skills/doc-search/scripts/) or adjust the commands to where you store the files; 3) do not point the skill at sensitive system directories — it will read all files under the docs_dir and create an index file there; 4) be aware the SKILL.md expects the agent/LLM to expand the query (the Python search script expects an already-expanded string), so ensure the agent performs expansion or pass expanded queries yourself; 5) inspect the scripts locally (they are plain Python) and run them manually once to confirm behavior before giving the agent permission to invoke them automatically.Like a lobster shell, security has layers — review code before you run it.
latestvk9722ekeswsykyx1krke67sb6h839rf0
License
MIT-0
Free to use, modify, and redistribute. No attribution required.