shmakk 1.1.0 → 1.2.0

package/skills/documents.md

@@ -0,0 +1,120 @@
+---
+name: documents
+version: 1
+category: files
+---
+
+# Document Analysis & Processing
+
+Read, summarize, extract, and process PDF, Word, and other document formats.
+
+## When to use this skill
+
+- User wants to read or summarize a PDF, Word doc, or other document
+- User wants to extract specific information from a document
+- User wants to compare two versions of a document
+- User mentions "PDF", "document", "report", "paper", "contract", "file"
+
+## Prerequisites
+
+Check available tools:
+```
+which pdftotext pdfinfo pandoc tesseract libreoffice 2>/dev/null
+```
+
+- `pdftotext` (poppler-utils) — extract text from PDF
+- `pandoc` — convert between document formats
+- `tesseract` — OCR for scanned PDFs/images
+- `libreoffice` — convert Word/Excel files to text/PDF
+
+If `pdftotext` is missing: `sudo apt install poppler-utils` or `brew install poppler`
+
+## Procedure
+
+### Read a PDF
+
+```
+pdftotext document.pdf -           # output to stdout
+pdftotext -f 1 -l 5 document.pdf - # pages 1-5 only
+pdfinfo document.pdf               # metadata: pages, title, author, size
+```
+
+For scanned PDFs (no selectable text), use OCR:
+```
+tesseract document.pdf output_base txt
+```
+
+### Read a Word document
+
+Convert to text first:
+```
+pandoc document.docx -t plain
+```
+
+Or to markdown (preserves structure):
+```
+pandoc document.docx -t markdown
+```
+
+### Read spreadsheets
+
+```
+pandoc spreadsheet.xlsx -t plain
+```
+
+Or use `libreoffice --headless --convert-to csv file.xlsx`
+
+### Summarize a long document
+
+1. Extract the full text
+2. If very long (>10k words), extract: title, table of contents, introduction, conclusion/executive summary, headings
+3. Build the summary from these key sections plus any explicitly requested detail
+
+### Compare two document versions
+
+```
+diff <(pdftotext v1.pdf -) <(pdftotext v2.pdf -)
+```
+
+For Word docs: convert both to markdown first, then diff.
+
+### Find specific information in a document
+
+After extracting text, search within it:
+```
+pdftotext document.pdf - | grep -i "search term" -A 3 -B 3
+```
+
+## Output format
+
+**For summaries:**
+```
+## Document Summary: [filename]
+
+**Type:** PDF, 24 pages
+**Author:** John Smith, 2024-01-15
+
+**Overview:** [1-2 sentence description of what the document is]
+
+**Key Points:**
+- Point 1
+- Point 2
+- Point 3
+
+**Notable Details:** [anything specific the user should know]
+```
+
+**For extraction:** present the extracted information cleanly, citing the page number where found.
+
+## Pitfalls
+
+- Scanned PDFs require OCR — `pdftotext` returns empty or garbled output; detect this and switch to `tesseract`
+- Password-protected PDFs: `pdftotext` will fail; inform user and ask them to provide an unlocked version
+- Large documents: extract just the needed sections rather than the entire text to avoid context overload
+- Tables in PDFs often don't extract cleanly — warn user if table data may be malformatted
+- Legal documents: extract verbatim quotes for important clauses rather than paraphrasing
+
+## Verification
+
+After extraction: check that the text makes sense (not garbled) before proceeding with analysis.
+If extracting page ranges, verify page count with `pdfinfo` first.

package/skills/email.md

@@ -0,0 +1,113 @@
+---
+name: email
+version: 1
+category: productivity
+---
+
+# Email Handling
+
+Read, search, compose, and send email via CLI email clients.
+
+## When to use this skill
+
+- User wants to check their inbox or find a specific email
+- User wants to compose or reply to an email
+- User asks about an email thread or conversation
+- User mentions "email", "inbox", "message", "send", "reply", "mail"
+
+## Prerequisites
+
+Check what's available:
+```
+which himalaya mutt neomutt aerc msmtp 2>/dev/null
+```
+
+**himalaya** is the recommended CLI email client — lightweight, fast, works with IMAP/SMTP.
+**mutt/neomutt** — powerful but requires more configuration.
+**msmtp** — for sending only, pairs with any client.
+
+If none installed, suggest: `himalaya` (`cargo install himalaya` or package manager).
+
+For sending only, `msmtp` with `~/.msmtprc` configured is the simplest path.
+
+## Procedure
+
+### Check inbox (himalaya)
+
+```
+himalaya list                    # list messages in inbox
+himalaya list --folder Sent      # list sent folder
+himalaya search "from:alex"      # search by sender
+himalaya search "subject:invoice" "since:1 week"
+```
+
+### Read a message
+
+```
+himalaya read <id>               # read message by ID
+himalaya attachment download <id> <filename>
+```
+
+### Compose and send
+
+```
+himalaya write                   # interactive compose
+```
+
+Or pipe directly:
+```
+himalaya send <<EOF
+From: user@example.com
+To: recipient@example.com
+Subject: Meeting tomorrow
+
+Hi,
+
+Can we meet tomorrow at 2pm?
+
+Best,
+Marcus
+EOF
+```
+
+### Reply to a message
+
+```
+himalaya reply <id>
+himalaya reply-all <id>
+```
+
+### With mutt/neomutt
+
+For reading: launch `neomutt` interactively.
+For sending: `echo "body" | neomutt -s "Subject" -a attachment.pdf -- recipient@example.com`
+
+### With msmtp (send only)
+
+```
+msmtp recipient@example.com < message.txt
+```
+
+## Output format
+
+When summarizing inbox:
+```
+INBOX — 3 unread
+
+ 1. [unread] Alex Chen — "PR review feedback" (2h ago)
+ 2. [unread] GitHub — "Build failed: main" (4h ago)
+ 3.          Client A — "Invoice received" (yesterday)
+```
+
+When summarizing a thread, extract the key points and action items.
+
+## Pitfalls
+
+- himalaya requires an account configured in `~/.config/himalaya/config.toml` — if not set up, prompt user to configure it first
+- IMAP credentials should never be shown or logged; reference config paths only
+- Large attachments: warn user before downloading
+- Do NOT send emails without explicit user confirmation of the recipient, subject, and content
+
+## Verification
+
+After sending: confirm with `himalaya list --folder Sent` that the message appears.

package/skills/expenses.md

@@ -0,0 +1,140 @@
+---
+name: expenses
+version: 1
+category: business
+---
+
+# Expense Tracking & Analysis
+
+Track, categorize, and analyze business or personal expenses for reporting and tax purposes.
+
+## When to use this skill
+
+- User wants to track or log an expense
+- User wants to generate an expense report
+- User wants to see spending by category over a period
+- User mentions "expense", "receipt", "reimbursement", "claim", "spend"
+
+## Procedure
+
+### Expense log format
+
+Maintain a simple CSV at `~/expenses/expenses.csv`:
+```
+date,amount,currency,category,vendor,description,receipt,reimbursable
+2024-01-15,42.50,EUR,Meals,Restaurang Oaxen,Client lunch with Alex,receipt-001.jpg,yes
+2024-01-16,29.90,EUR,Transport,SL,Monthly transit pass,,no
+2024-01-17,120.00,EUR,Software,Adobe,Creative Cloud annual,,yes
+```
+
+### Log a new expense
+
+Append to the CSV:
+```
+echo "2024-01-15,42.50,EUR,Meals,Restaurang Oaxen,Client lunch,,yes" >> ~/expenses/expenses.csv
+```
+
+Or write to file with read_file + edit_file to verify format is correct.
+
+### Generate expense report
+
+**Summary by category:**
+```python
+import csv
+from collections import defaultdict
+
+with open(os.path.expanduser('~/expenses/expenses.csv')) as f:
+    rows = list(csv.DictReader(f))
+
+# Filter by date range
+from_date = "2024-01-01"
+to_date = "2024-01-31"
+rows = [r for r in rows if from_date <= r['date'] <= to_date]
+
+# Sum by category
+by_category = defaultdict(float)
+reimbursable_total = 0.0
+for row in rows:
+    by_category[row['category']] += float(row['amount'])
+    if row.get('reimbursable', '').lower() == 'yes':
+        reimbursable_total += float(row['amount'])
+
+total = sum(by_category.values())
+
+print(f"EXPENSE REPORT: {from_date} to {to_date}\n")
+print(f"{'Category':<20} {'Amount':>10}")
+print("-" * 32)
+for cat, amount in sorted(by_category.items(), key=lambda x: -x[1]):
+    print(f"{cat:<20} {amount:>9.2f} EUR")
+print("-" * 32)
+print(f"{'Total':<20} {total:>9.2f} EUR")
+print(f"{'Reimbursable':<20} {reimbursable_total:>9.2f} EUR")
+```
+
+### Generate reimbursement report
+
+Filter to reimbursable expenses only and produce a report for submission:
+
+```python
+reimbursable = [r for r in rows if r.get('reimbursable', '').lower() == 'yes']
+```
+
+Format as a table for submission to employer/client.
+
+### Organize receipts
+
+Store receipt images/PDFs in `~/expenses/receipts/YYYY-MM/`:
+```
+mkdir -p ~/expenses/receipts/2024-01/
+# Move receipt file into the directory, name matching expense
+```
+
+### Tax-deductible categories
+
+For self-employed / freelance (Swedish context):
+- Software & subscriptions (100% deductible for business use)
+- Client meals (75% deductible in Sweden, requires documentation of who was present)
+- Home office (proportional — discuss with accountant)
+- Transport to client sites (deductible)
+- Professional development (courses, books) (deductible)
+- Equipment (may need to be depreciated over multiple years)
+
+Flag transactions by deductibility for easier tax filing.
+
+## Output format
+
+```
+EXPENSE REPORT — January 2024
+
+Category          Count    Amount
+──────────────────────────────────
+Software            3     €149.90
+Meals               4     €187.50
+Transport           5     €118.00
+Equipment           1     €249.00
+──────────────────────────────────
+Total              13     €704.40
+
+Reimbursable:         €436.90  ← submit to client
+Non-reimbursable:     €267.50
+
+REIMBURSABLE EXPENSES (for submission)
+Date        Amount    Vendor              Description
+──────────────────────────────────────────────────────
+2024-01-15  €42.50   Restaurang Oaxen    Client lunch (Alex Chen)
+2024-01-18  €249.00  Apple Store         External monitor for project
+2024-01-22  €89.00   Adobe               CC subscription (billable)
+2024-01-28  €56.40   SAS                 Flight to client (Stockholm–Malmö)
+```
+
+## Pitfalls
+
+- Always note who was present at client meals (required for meal deduction)
+- Keep digital copies of all receipts — original paper receipts must be kept for 7 years in Sweden
+- Currency: record original currency AND amount (don't convert on the fly — use bank statement rate for tax purposes)
+- Separate business and personal expenses — don't mix in the same account if avoidable
+
+## Verification
+
+After generating a report: sum the per-category totals manually and verify they match the total row.
+Cross-check reimbursable total against individual reimbursable items.

package/skills/file-ops.md

@@ -0,0 +1,149 @@
+---
+name: file-ops
+version: 1
+category: system
+---
+
+# File Operations & Organization
+
+Bulk rename, reorganize, search, deduplicate, and manage files across directories.
+
+## When to use this skill
+
+- User wants to rename many files following a pattern
+- User wants to organize files by type, date, or category
+- User wants to find duplicate files
+- User wants to find files matching certain criteria
+- User mentions "rename files", "organize", "move files", "find duplicates", "bulk"
+
+## Procedure
+
+### Find files by criteria
+
+```
+find . -name "*.jpg" -type f                    # by extension
+find . -name "*.log" -mtime +30 -type f         # older than 30 days
+find . -size +50M -type f                        # larger than 50MB
+find . -empty -type f                            # empty files
+find . -newer reference.txt -type f              # modified after a file
+find ~/Downloads -mtime +90 -type f              # old downloads
+```
+
+### Bulk rename files
+
+**Preview first (ALWAYS preview before executing):**
+
+Pattern-based rename with `rename` (Perl rename, common on Linux):
+```
+rename -n 's/old/new/' *.txt          # -n is dry run, shows what would change
+rename 's/old/new/' *.txt             # actual rename
+rename 's/ /_/g' *                    # replace spaces with underscores
+rename 's/^/prefix_/' *.jpg           # add prefix
+rename 's/\.jpeg$/.jpg/' *.jpeg       # normalize extension
+```
+
+With `mv` and a loop (when rename isn't available):
+```
+for f in *.txt; do mv -n "$f" "${f/old/new}"; done   # -n = no-clobber
+```
+
+Date-based rename (add date prefix):
+```
+for f in *.jpg; do
+  date=$(stat -c %y "$f" | cut -d' ' -f1)
+  mv -n "$f" "${date}_${f}"
+done
+```
+
+### Organize files by type
+
+```
+mkdir -p organized/{images,documents,videos,audio,archives,other}
+
+for f in *; do
+  case "${f##*.}" in
+    jpg|jpeg|png|gif|webp|svg) mv "$f" organized/images/ ;;
+    pdf|doc|docx|odt|txt|md)  mv "$f" organized/documents/ ;;
+    mp4|mkv|avi|mov|webm)     mv "$f" organized/videos/ ;;
+    mp3|flac|wav|ogg|m4a)     mv "$f" organized/audio/ ;;
+    zip|tar|gz|bz2|7z|rar)   mv "$f" organized/archives/ ;;
+    *) [ -f "$f" ] && mv "$f" organized/other/ ;;
+  esac
+done
+```
+
+### Organize by date (year/month)
+
+```
+for f in *.jpg; do
+  year=$(stat -c %y "$f" | cut -d'-' -f1)
+  month=$(stat -c %y "$f" | cut -d'-' -f2)
+  mkdir -p "$year/$month"
+  mv -n "$f" "$year/$month/"
+done
+```
+
+### Find and handle duplicates
+
+Using `fdupes` (most reliable):
+```
+fdupes -r .                    # find duplicates recursively
+fdupes -r -d .                 # interactively delete duplicates (keeps first)
+fdupes -r -f . | tail -n +2   # list all but the first of each group
+```
+
+Without fdupes (using md5sum):
+```
+find . -type f -exec md5sum {} \; | sort | awk 'seen[$1]++ {print $2}' > duplicates.txt
+cat duplicates.txt              # review before deleting
+```
+
+### Clean up specific file types
+
+Find and remove old log files:
+```
+find . -name "*.log" -mtime +30 -exec ls -lh {} \;    # preview
+find . -name "*.log" -mtime +30 -delete                # delete (careful!)
+```
+
+Find and compress old files:
+```
+find ~/Documents -mtime +365 -type f -exec gzip {} \;
+```
+
+### Sync or mirror directories
+
+```
+rsync -av --dry-run source/ destination/    # preview
+rsync -av source/ destination/              # execute
+rsync -av --delete source/ destination/     # mirror (deletes files not in source!)
+```
+
+## Output format
+
+When previewing bulk operations, always show the full list of changes before asking the user to confirm:
+
+```
+PREVIEW: Rename 23 files (replace spaces with underscores)
+
+  "my photo 001.jpg"     →  "my_photo_001.jpg"
+  "my photo 002.jpg"     →  "my_photo_002.jpg"
+  "meeting notes.docx"   →  "meeting_notes.docx"
+  ... (20 more)
+
+Proceed? This cannot be easily undone.
+```
+
+## Pitfalls
+
+- **ALWAYS show a preview before executing bulk operations** — mistakes are hard to undo
+- Use `-n` (no-clobber) with `mv` to prevent overwriting existing files
+- `find -delete` is permanent — no trash. Use `ls` or move to a temp dir first
+- `rsync --delete` will delete files at destination that don't exist at source — confirm this is intended
+- Beware of files with spaces or special characters in names — always quote `"$f"` in loops
+- Check available disk space before large copy/move operations: `df -h`
+
+## Verification
+
+After bulk rename: spot-check 3-5 files to confirm names are correct.
+After organize: verify files landed in the right directories with `ls` on each output directory.