From skillry-runtime-and-local-app
Use when you need to collect concise logs, environment facts, versions, and failure evidence without exposing secrets.
How this skill is triggered — by the user, by Claude, or both
Slash command
/skillry-runtime-and-local-app:11-log-and-diagnostics-bundleThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill assembles a compact, shareable diagnostics snapshot that contains everything needed to reproduce and understand a failure: relevant log lines, runtime versions, env var shape (keys but not values), reproduction steps, and error stack traces — while systematically redacting secrets, tokens, and credentials. The output can be safely pasted into a GitHub issue, shared with a teammate, o...
This skill assembles a compact, shareable diagnostics snapshot that contains everything needed to reproduce and understand a failure: relevant log lines, runtime versions, env var shape (keys but not values), reproduction steps, and error stack traces — while systematically redacting secrets, tokens, and credentials. The output can be safely pasted into a GitHub issue, shared with a teammate, or attached to a support ticket.
tail -f or your log aggregator directly.node --version; npm --version; npx --version # Node ecosystem
python --version; pip --version # Python
go version # Go
java -version; mvn --version # Java/Maven
docker --version; docker compose version # Containers
uname -a # OS / kernel
Record the output verbatim — version mismatches are the most common cause of "works on my machine."
env | grep -E "NODE|PYTHON|PATH|PORT|DATABASE|REDIS|MONGO|API|SECRET|TOKEN|KEY|URL|HOST|ENV|APP|SERVICE" | sed 's/=.*/=<REDACTED>/'
This shows which keys are set and which are missing without exposing any values. For .env files: grep -E "^[A-Z_]+" .env | sed 's/=.*/=<REDACTED>/'.
Extract the relevant log window. From a log file: grep -n "ERROR\|FATAL\|panic\|Exception\|Traceback" app.log | tail -50. From a running process: capture stderr for 30 seconds of reproduction: command 2>&1 | tee /tmp/diag-$(date +%s).log. Include 20 lines before and after each error line for context: grep -n "ERROR" app.log | head -5 | awk -F: '{print $1}' | xargs -I{} sed -n "$(({})-20),$(({}+20))p" app.log.
Capture dependency installation state. Node: cat package.json | jq '{name,version,dependencies,devDependencies}' and npm ls --depth=0 2>&1 | head -30. Python: pip list --format=columns | head -30. Go: cat go.mod. This reveals lock file drift and unexpected package versions.
Record the exact reproduction steps. Write out in numbered order: the command run, any preconditions (specific env vars set, specific file present), and the expected vs. actual behavior. Be precise: "I ran npm run dev from the repo root with NODE_ENV=development" is more useful than "I started the app."
Capture process and resource context at failure time. If the app crashes:
ps aux | grep "node\|python\|java" # running processes
df -h / # disk space
free -m # memory (Linux)
ulimit -a # resource limits
On macOS: vm_stat | head -10 instead of free.
Collect relevant config file snippets (redacted). If the failure relates to a config file (e.g., nginx.conf, docker-compose.yml, k8s/deployment.yaml), include the relevant section with any secret values replaced by <REDACTED>. Never include TLS private keys, API tokens, or database passwords.
Assemble the bundle into a structured document. Structure:
## Environment
## Reproduction steps
## Actual behavior (log excerpt)
## Expected behavior
## Dependency versions
## Env var keys present
## Config (redacted)
grep -iE "(password|secret|token|apikey|private_key|bearer)\s*[:=]\s*[^\s<]" bundle.txt
Any match must be replaced with <REDACTED> before sharing.
<REDACTED>npm ls, pip list, go.mod).env files, private keys, and bearer tokens are NOT included in the bundle.env file: the most dangerous mistake — includes all secrets. Always extract only the key names.{email: '[email protected]', password: '...'} after clearing cookies" is.A structured diagnostics bundle containing:
SECRET, TOKEN, PASSWORD, KEY, PRIVATE, or CREDENTIAL — replace with <REDACTED>.npx claudepluginhub fluxonlab/skillry --plugin skillry-runtime-and-local-appCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.