AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Recall

skill-kalawcreator919-claude-obsidian-recall · by kalawcreator919

Search vault by keyword with relevance scoring. Supports temporal filters, timeline mode, and active-topics mode.

No reviews yet
0 installs
18 views
0.0% view→install

Install

$ agentstack add skill-kalawcreator919-claude-obsidian-recall

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-kalawcreator919-claude-obsidian-recall)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
4mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Recall? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Recall — Obsidian Note Search

Quickly find relevant notes in the vault by keyword, scored and ranked.

> Configure your vault path in setup.sh or set the VAULT_PATH environment variable.

Search and display only — never move or modify any notes.

Vault: $VAULT_PATH

Supported Modes

| Mode | Syntax | Description | |------|--------|-------------| | Basic search | /recall keyword | Standard keyword search with scoring | | Temporal filter | /recall keyword --since 2w | Filter by time (e.g., --since 1m, --month 2026-03, --between 2026-01 2026-03) | | Timeline | /recall keyword --timeline | ASCII activity chart showing note frequency over time | | Active topics | /recall --active-topics | Top tags across the vault, ranked by frequency |

Process

Step 1 — Understand Search Intent

Analyze the user's keyword and expand into 2-5 related search terms.

Examples:

  • /recall deployment → "deployment", "deploy", "CI/CD", "shipping", "release"
  • /recall database → "database", "DB", "SQL", "Prisma", "schema"

Temporal parsing:

  • --since 2w → filter notes modified within the last 2 weeks
  • --since 1m → last month
  • --month 2026-03 → notes from March 2026
  • --between 2026-01 2026-03 → notes from January to March 2026

Step 2 — Full Vault Search

Search the entire vault at once, not folder by folder. Use 3 methods in parallel for each search term:

  1. Filename — Glob tool:

`` Glob: pattern = "**/*keyword*", path = "$VAULT_PATH" ``

  1. Tags — Grep tool:

`` Grep: pattern = "tags:.*keyword", path = "$VAULT_PATH", glob = "**/*.md", -i = true ``

  1. Content — Grep tool (fetch context directly):

`` Grep: pattern = "keyword", path = "$VAULT_PATH", glob = "**/*.md", -i = true, output_mode = "content", -C = 2, head_limit = 50 ``

Run all search terms in parallel. If results exceed 50 notes, take only the top 10 matches per method.

Quality field: For files in the results, check for a quality: frontmatter field. If present, note the value (high/medium/low) for Step 3 scoring.

Temporal filter: If --since, --month, or --between is specified, exclude notes whose date: frontmatter falls outside the range.

Step 2.5 — Semantic Search (Optional)

If a semantic_search tool is available (e.g., vault-semantic-search MCP server):

  1. Call semantic_search(query=user's search terms, top_k=10)
  2. Add results to the Step 3 candidate pool, tagged as "semantic match"
  3. Semantic search finds notes using different wording but the same concept

If the semantic_search tool is not available, skip this step silently.

Step 3 — Score and Rank

Deduplicate all results, then score:

| Condition | Score | |-----------|-------| | Filename match | +3 | | Tags match | +2 | | Content match (+1 per distinct search term, max +3) | +1 to +3 | | Located in 30 - Notes/ | +2 bonus | | Located in 10 - Projects/ | +1 bonus | | Recently modified (within 30 days) | +1 bonus | | quality: high | +3 bonus | | quality: medium | +1 bonus | | quality: low | +0 | | semanticsearch score > 0.7 | +4 bonus | | semanticsearch score 0.5-0.7 | +2 bonus |

Take the Top 10. Break ties by folder priority: 30 > 10 > 20 > 00 > 40 > 99.

Step 4 — Display Results

For each Top 10 note, distill a one-sentence key insight from the Step 2 Grep content results (no need to Read unless context is insufficient).

Standard mode:

/recall [keyword] results

Found N related notes:

Most Relevant
1. [[Note Name]] (30 - Notes) — X pts
   > [One-sentence key insight]

2. [[Note Name]] (10 - Projects/XXX) — X pts
   > [One-sentence summary]

3. [[Note Name]] (20 - Areas/YYY) — X pts
   > [One-sentence summary]

Other Related
4-10. ...

Related Note Network
[[Note A]]  [[Note B]] (based on wikilinks)

Suggestion: [specific actionable suggestion]

Timeline mode (--timeline):

/recall [keyword] — Timeline

Activity for "keyword" (past 6 months):

2026-04  ████████  8 notes
2026-03  ████       4 notes
2026-02  ██         2 notes
2026-01               0 notes
2025-12  █           1 note
2025-11  ███         3 notes

Peak: 2026-04 (8 notes)
Total: 18 notes mentioning "keyword"

Active topics mode (--active-topics):

/recall --active-topics

Top 10 Active Topics (past 30 days):

1. deployment (12 mentions across 8 notes)
2. testing (9 mentions across 6 notes)
3. design-system (7 mentions across 5 notes)
...

Step 5 — Follow-up Actions

Available actions:
- Enter a note number to view full content
- /recall [another keyword] to search a different topic
- /daily-review to process related Inbox notes

When the user enters a number, Read and display the corresponding note in full.

Edge Cases

  • 0 results: Suggest 2-3 similar keywords to retry
  • Non-ASCII search terms: Grep supports them natively
  • Missing folders: Skip silently, do not error

Guidelines

  • Read-only — Never modify, move, or delete notes
  • Keep technical terms in English
  • Case-insensitive search
  • Wikilinks use [[Note Name]] without the .md extension

Related Skills

  • [[daily-review]] — Process notes found in search results
  • [[weekly-review]] — Surfaces resurfacing opportunities for older notes
  • [[vault-lint]] — Find orphan notes that recall might miss

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.