Token manager

Skill jquigley3/session-keeper/skills/token-manager

Host-side secret proxy and onboarding flow for sandboxed Claude sessions

Install

npx -y skills add jquigley3/session-keeper --skill token-manager

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Manage tokens and secrets for Claude Code sessions. Use this skill whenever the user mentions an API key, token, secret, password, credential, or wants to store/retrieve/audit sensitive values. Triggers on phrases like "save my token", "add this API key", "where did I use that token", "find secrets in my sessions", "audit my credentials", "store this key", "what tokens do I have", or any time the user pastes or references a sensitive value. Also use when the user wants to know if a secret has leaked into chat history.

SKILL.md

4.7 KB, as published. Nobody here has run it

Token Manager

Helps you store, reference, and audit tokens and secrets across Claude Code sessions.

The convention

All tokens, API keys, and secrets live in ~/.claude/.env in NAME=value format:

ANTHROPIC_API_KEY=sk-ant-...
GITHUB_TOKEN=ghp_...
MY_GITEA_TOKEN=abc123...
OPENAI_KEY=sk-...

Even when an env var isn't technically required (e.g., a one-time token you'll use in a command), add it with a descriptive name. The name is a record; the value is a reference. Future-you will thank current-you.

When a user shares a token value or asks you to use one:

Ask if they'd like it added to ~/.claude/.env
Suggest a clear SCREAMING_SNAKE_CASE name
Reference it by name going forward (e.g., $GITHUB_TOKEN in commands)

When a token is already in .env: Reference it by name, never paste the value.

Adding a token

If the user asks to save or add a token:

# Append to ~/.claude/.env (create if it doesn't exist)
echo 'MY_TOKEN_NAME=the-value' >> ~/.claude/.env

Or open the file for them to edit:

# Show current .env contents (masked)
python3 <skill-dir>/scripts/sk-tokens.py --simple --json | python3 -c "
import json,sys; d=json.load(sys.stdin)
print('\n'.join(set(h['env_name'] for h in d)))"

Auditing session history

The bundled script scripts/sk-tokens.py can audit session history in two modes.

Simple mode — find known tokens in chat

Searches session history for token values and names already in ~/.claude/.env:

python3 <skill-dir>/scripts/sk-tokens.py --simple

Filter to a specific token:

python3 <skill-dir>/scripts/sk-tokens.py --simple --name GITHUB_TOKEN

Complex mode — find uncatalogued secrets

Scans session chat for values that look like tokens but are NOT in ~/.claude/.env:

python3 <skill-dir>/scripts/sk-tokens.py --complex

Limit to recent sessions:

python3 <skill-dir>/scripts/sk-tokens.py --complex --days 30

Agent-injectable patterns

Pass --pattern to add custom regex. The pattern can use a capture group (...) to extract just the secret value:

# Find Gitea tokens (40-char hex) in chat
python3 <skill-dir>/scripts/sk-tokens.py --complex \
  --pattern '[0-9a-f]{40}'

# Find values after "Authorization: token "
python3 <skill-dir>/scripts/sk-tokens.py --complex \
  --pattern '(?i)Authorization:\s+token\s+([^\s]+)'

JSON output (for agents)

python3 <skill-dir>/scripts/sk-tokens.py --complex --json

Returns an array of hit objects with: session, project, line, role, label, matched (masked), context.

Scrubbing tokens from session history

Once you've found tokens in session history, you can scrub them in place. The script rewrites JSONL files atomically, replacing token values with a masked form. Three modes:

Mode	Example output	Use when
1 (default)	`ghp_xxx2345`	You want to know which token it was
2	`ghp_xx`	You want to keep the type identifier only
3	`xxx`	Full redaction

# Dry run first — see what would change
python3 <skill-dir>/scripts/sk-tokens.py --scrub --dry-run

# Scrub with mode 1 (keep prefix + last 4)
python3 <skill-dir>/scripts/sk-tokens.py --scrub --scrub-mode 1

# Scrub a specific known token (simple mode) with full redaction
python3 <skill-dir>/scripts/sk-tokens.py --scrub --simple --name GITHUB_TOKEN --scrub-mode 3

# Scrub and also clean the value out of .env
python3 <skill-dir>/scripts/sk-tokens.py --scrub --scrub-mode 2 --remove-from-env

--remove-from-env replaces the value in .env with [SCRUBBED] so you retain the variable name as a record.

Docker sandboxes

python3 <skill-dir>/scripts/sk-tokens.py --complex --sandbox

Also crawls running Docker sandbox containers.

.env format

# Comments are supported
NAME=value
NAME="value with spaces"
NAME='single quoted'

Empty values and lines without = are ignored. Values are stripped of surrounding quotes.

<skill-dir> is the directory containing this SKILL.md — typically ~/.claude/skills/token-manager/.