Credential-safe process probing: vetted helpers for env vars / argv / fds / network plus a PreToolUse hook blocking raw /proc/<pid>/{environ,cmdline} reads.
A personal Claude Code plugin marketplace.
Add this marketplace once:
/plugin marketplace add StefanKarpinski/karpinski-claude-plugins
Then install whichever plugins you want:
/plugin install process-probe@karpinski-claude-plugins
To pin a specific tag or branch:
/plugin marketplace add StefanKarpinski/karpinski-claude-plugins#v0.1.0
process-probeCredential-safe helpers for inspecting a running process — environment variables, command-line argv, file descriptors, network connections — paired with a PreToolUse hook that blocks direct reads of /proc/<pid>/environ and /proc/<pid>/cmdline.
Why it exists: ad-hoc redaction (e.g., a one-line sed filter over /proc/<pid>/environ) is one regex bug away from emitting credentials in clear into the conversation transcript. This plugin replaces the unsafe path with a vetted command (whitelist-style redaction by variable name and by value shape) and a tool-call hook that blocks the unsafe path entirely.
After install, the plugin puts a single command on your PATH: process-probe, with six subcommands. All output is JSON or JSONL — one line per record for list-shaped commands.
process-probe --help # list subcommands
process-probe env-keys <pid> # JSONL: {"name": "..."} per env var
process-probe env-values <pid> NAME... # JSONL: per-name {"value"} / {"redacted"} / {"unset"}
process-probe env-values <pid> --unsafe-show NAME # per-name override (auditable)
process-probe cmdline <pid> # JSONL: per-argv {"index", "value"?, "redacted"?}
process-probe info <pid> # one JSON object: ps summary
process-probe fds <pid> # JSONL: open file descriptors
process-probe network <pid> # JSONL: TCP/UDP connections
The PreToolUse hook fires automatically across Bash, Read, Edit, Write, and NotebookEdit — any tool call referencing /proc/<real-pid>/environ or /proc/<real-pid>/cmdline is blocked with a pointer back to process-probe.
See plugins/process-probe/skills/process-probe/SKILL.md for the full reference: usage examples, override semantics, the secret-detection heuristics (both name-keyword and value-shape axes), the hook's coverage matrix, and known bypass surfaces.
The secret-detection heuristics follow the same two-axis shape as detect-secrets, gitleaks, and truffleHog: keyword match on the variable name, plus known-credential-format and entropy detection on the value.
.claude-plugin/
marketplace.json # marketplace manifest — lists the plugins
plugins/
process-probe/
.claude-plugin/
plugin.json # plugin manifest
bin/
process-probe # the only thing on PATH; subcommand dispatcher
libexec/ # subcommand implementations (not on PATH)
env-keys, env-values, cmdline, info, fds, network
lib/ # internal Python modules
_secret_heuristics.py, _test_heuristics.py
hooks/
hooks.json # PreToolUse hook wiring
proc-probe-guard.sh # the hook script
skills/process-probe/
SKILL.md # skill description + reference docs
Why a single process-probe entry point with subcommands rather than separate executables: most of the subcommand names (info, network, fds, …) are generic enough to collide with system tools (info shadows /usr/bin/info on most Linuxes), and PATH ordering puts plugin bin/ after /usr/bin, so a bare info <pid> would launch GNU info trying to read manual <pid>. One namespaced entry point sidesteps this and matches the conventional shape of git / kubectl / docker / gh.
Adding a new plugin to this marketplace: drop a directory under plugins/, give it .claude-plugin/plugin.json, and append an entry to .claude-plugin/marketplace.json.
MIT. See LICENSE.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub stefankarpinski/karpinski-claude-plugins --plugin process-probeUltra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Memory compression system for Claude Code - persist context across sessions
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills.