skills/ss/SKILL.md
Load screenshot images or other files from Dropbox screenshots directory. Use when user invokes /ss directly. NEVER manually list/read files from the screenshots dir — this skill handles Dropbox sync delays, freshness checks, retry logic that manual reads miss. Supports: /ss 2 (latest 2), /ss latest3, /ss filename.png (exact/substring), /ss full-path. Also non-image files (e.g., /ss pattern-4-variations.html).
npx skillsauth add takazudo/claude-resources ssInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Load screenshot images or other files from the Dropbox screenshots directory and present them in the conversation. Also supports non-image files (HTML, text, etc.) that the user shares via this directory.
Use $DROPBOX_SCREENSHOTS_DIR env var (set in .zshrc for both macOS and WSL2).
The following epoch was captured at the moment this command was invoked:
Invocation epoch: !date +%s
Use this as a cutoff for "Latest N" mode — only consider image files whose modification time is <= this epoch. This prevents picking up screenshots that appeared after the user typed /ss.
$ARGUMENTS determines which files to load:
| Pattern | Meaning | Example |
| --- | --- | --- |
| Bare number (2, 3) | Latest N images at invocation time | /ss 2 |
| latestN | Latest N images at invocation time | /ss latest3 |
| Full path starting with / | Exact file | /ss '/Users/.../file.png' |
| Non-image filename (.html, .txt, etc.) | Read a non-image file from screenshots dir | /ss pattern-4-variations.html |
| Other string (exact) | Exact filename in screenshots dir | /ss Screenshot 2026-03-14 at 3.27.17.png |
| Other string (partial) | Substring search in filenames | /ss foo-bar-moo.png |
| (empty) | Latest 1 image | /ss |
Default: When $ARGUMENTS is empty or blank, treat it as 1 (load the latest single screenshot). This is the most common use case.
Non-image files: When the argument has a non-image extension (e.g., .html, .txt, .json, .css, .js, .svg, .md), look for that file in $DROPBOX_SCREENSHOTS_DIR and read it as a text file using the Read tool. This is for when the user shares files (not just screenshots) via the Dropbox screenshots directory.
$DROPBOX_SCREENSHOTS_DIR syncs to both macOS (BSD stat) and WSL2/Linux (GNU stat), and their flags for "modification time as epoch seconds" differ (stat -f %m vs stat -c %Y). Naively falling back with stat -f %m "$f" 2>/dev/null || stat -c %Y "$f" is not safe: GNU stat -f means "show filesystem status" (not "format"), so it still exits non-zero but leaks a garbage filesystem-info block to stdout first, corrupting the captured value. Define this function once at the top of any bash snippet below that needs a file's mtime — each Bash tool call runs in a fresh shell, so redefine it per invocation:
mtime_of() {
local m
m=$(stat -f %m "$1" 2>/dev/null)
if [ $? -ne 0 ] || [ -z "$m" ]; then
m=$(stat -c %Y "$1" 2>/dev/null)
fi
echo "$m"
}
All stat calls below use this mtime_of helper instead of calling stat directly.
List all image files (png, jpg, jpeg, gif, webp, tiff) in the screenshots directory, filter to only those with modification time <= the invocation epoch, sort by modification time descending, take the first N.
CUTOFF=<invocation_epoch>
ls -t "$DROPBOX_SCREENSHOTS_DIR"/*.{png,jpg,jpeg,gif,webp,tiff} 2>/dev/null | while IFS= read -r f; do
[ "$(mtime_of "$f")" -le "$CUTOFF" ] && echo "$f"
done | head -n $N
This ensures that even if new screenshots appear while Claude is processing, only files that existed at invocation time are selected.
When the argument has a non-image extension (not .png, .jpg, .jpeg, .gif, .webp, .tiff):
$DROPBOX_SCREENSHOTS_DIR/<filename>$DROPBOX_SCREENSHOTS_DIR/<filename>foo-bar-moo.png → foo-bar-moo)screenshot ↔ screenshots, image ↔ images)# Substring search example
SEARCH="foo-bar-moo"
find "$DROPBOX_SCREENSHOTS_DIR" -maxdepth 1 -type f \( -iname "*.png" -o -iname "*.jpg" -o -iname "*.jpeg" -o -iname "*.gif" -o -iname "*.webp" -o -iname "*.tiff" \) -iname "*${SEARCH}*" -print0 | xargs -0 ls -t 2>/dev/null | head -n 1
Use the path as-is.
The user typically takes a screenshot and immediately runs /ss. Dropbox sync introduces a delay of a few seconds (sometimes longer), so the file may not exist locally yet. This section handles that proactively.
For "Latest N" mode (bare number, latestN, or empty argument), always sleep 3 seconds before the first file lookup. This gives Dropbox a moment to sync the new screenshot. This simple delay avoids the most common case where old files are returned because the new one hasn't arrived yet.
# Always do this before the first file listing in "Latest N" mode
sleep 3
After listing files, check whether the results are fresh enough relative to the invocation epoch. A screenshot the user just took should have a modification time within ~60 seconds of the invocation epoch.
Freshness rule: At least one of the N files must have a modification time within 120 seconds before the invocation epoch. If ALL files are older than 120 seconds before the invocation epoch, the user's new screenshot likely hasn't synced yet — retry.
CUTOFF=<invocation_epoch>
FRESHNESS_THRESHOLD=$((CUTOFF - 120))
# After getting the N files, check the newest one:
NEWEST_MTIME=$(mtime_of "$NEWEST_FILE")
if [ "$NEWEST_MTIME" -lt "$FRESHNESS_THRESHOLD" ]; then
# All results are stale — the new screenshot hasn't synced yet, retry
fi
If the file is not found (specific filename mode) or the freshness check fails (Latest N mode), poll with retries:
[ -f "$TARGET" ])# Specific filename mode
for i in $(seq 1 24); do
[ -f "$TARGET" ] && break
sleep 5
done
# Latest N mode (after initial 3-second delay already elapsed)
for i in $(seq 1 24); do
# Re-run the file listing and freshness check
FILES=$(ls -t "$DROPBOX_SCREENSHOTS_DIR"/*.{png,jpg,jpeg,gif,webp,tiff} 2>/dev/null | while IFS= read -r f; do
[ "$(mtime_of "$f")" -le "$CUTOFF" ] && echo "$f"
done | head -n $N)
NEWEST=$(echo "$FILES" | head -n 1)
[ -n "$NEWEST" ] && [ "$(mtime_of "$NEWEST")" -ge "$FRESHNESS_THRESHOLD" ] && break
sleep 5
done
Use the Read tool to read each file. The Read tool supports reading image files (PNG, JPG, etc.) visually, and text files as content.
After reading, briefly acknowledge which file(s) were loaded (filename only, not full path) and ask what the user would like to do with them.
After presenting files, you MUST treat them as the user's intended files. The user just took these screenshots and ran /ss — they expect you to work with whatever was loaded.
NEVER do this:
/ss wasn't invokedThe user's screenshots ARE the context. Even if the content seems unrelated to the prior conversation, the user may be introducing new context, switching topics, or showing something you don't yet understand.
Only if the files are obviously stale (e.g., timestamps are many hours old, content is clearly from a different session), follow this two-step recovery:
Step 1: Wait and recheck (Dropbox sync delay)
The user's actual screenshot may still be syncing. Wait 2 minutes, then re-list the directory for newer files:
# Wait for potential Dropbox sync delay
sleep 120
# Re-check for newly synced files (allow mtime up to 180s after invocation epoch)
LATE_CUTOFF=$((CUTOFF + 180))
ls -t "$DROPBOX_SCREENSHOTS_DIR"/*.{png,jpg,jpeg,gif,webp,tiff} 2>/dev/null | while IFS= read -r f; do
MTIME=$(mtime_of "$f")
[ "$MTIME" -le "$LATE_CUTOFF" ] && [ "$MTIME" -gt "$CUTOFF" ] && echo "[NEW] $f"
done | head -n 3
If new files appeared, read and present them — these are likely the intended screenshots.
Step 2: Ask the user
If no new files appeared after waiting, present what you have and ask the user to confirm:
"I loaded [filenames]. These screenshots appear to be from [timestamp]. Are these the ones you meant, or should I wait longer for a newer screenshot to sync?"
Never skip both steps. If you suspect the files are wrong, you MUST either wait-and-recheck OR ask the user. Silently ignoring the user's /ss invocation is never acceptable.
tools
Acceptance gate for a branch produced by an OpenAI Codex CLI run — usually Codex implementing a /big-plan epic that was handed off to it. Codex reports the work 'done' (or the user flags it WIP with corrections); this skill confirms the branch actually fulfils the original spec, fixes what falls short, and routes larger discoveries into GitHub issues. Use when: (1) User says '/finalize-codex-work', 'finalize codex work', 'confirm the codex work', 'check the codex branch', or 'codex said it's done', (2) A branch is the result of a Codex CLI session and needs verification against its spec issue/PR, (3) After assigning a /big-plan epic to Codex CLI. Pass -m/--merge to run /pr-complete -c at the end.
tools
Read a Figma design node directly from a share URL via the Figma REST API — no Dev Mode subscription, no MCP, no desktop app. Renders the node to PNG and dumps its full style/layout JSON so the design can be described, compared, or implemented. Use whenever the user gives a Figma design URL (figma.com/design/... or /file/...) and wants to see, read, inspect, reference, or implement that node — including `/fig-url-refer <url>`. This is the URL-based counterpart to `/figrefer` (which needs a Dev-plan desktop MCP); prefer this one when the input is a URL rather than a live desktop selection.
tools
Sync the user's Claude Code workflow skills into the OpenAI Codex CLI settings repo ($HOME/.codex) as Codex-native ports, fix the Codex .gitignore for new local state, then commit and push. Use when: (1) user says '/dev-codex-sync-settings-from-claude', 'sync codex settings', 'sync claude skills to codex', 'port skills to codex', or 'update codex from claude'; (2) after updating ~/.claude workflow skills (big-plan, x, x-as-pr, x-wt-teams) and Codex should catch up; (3) the $HOME/.codex repo has drifted behind $HOME/.claude. The ports are condensed Codex-native REWRITES, never file copies.
development
Analyze a video file (mov, mp4, webm, etc.) or a YouTube video by extracting still frames with ffmpeg and reading them chronologically with vision — Claude cannot ingest video files directly. Use whenever the user provides a video file path or YouTube URL and wants to know what happens in it: "read this video", "watch this video", "check this recording", "what happens in this .mov/.mp4", analyzing a screen recording of a UI bug, or verifying UI behavior captured in a video, even if they don't name this skill.