Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Upstash Redis Key-Value Store · mkurman | DecimalAI

mkurman/upstash-redis-kv

SkillScore55Maintained

Read and write to Upstash Redis-compatible key-value store via REST API. Use when there is a need to save or retrieve key-value data, use Redis features (caching, counters, lists, sets, hashes, sorted sets, etc.) for the current interaction, or when the user explicitly asks to use Upstash or Redis.

Databy mkurman

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Data Engineer pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/mkurman-upstash-redis-kv/SKILL.md

decimalai skills pull mkurman-upstash-redis-kv --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/mkurman-upstash-redis-kv/SKILL.md https://app.decimal.ai/s/mkurman-upstash-redis-kv/SKILL.md

Raw file: /s/mkurman-upstash-redis-kv/SKILL.md ↗ · machine-readable summary at /s/mkurman-upstash-redis-kv.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 8d ago

—+373% tokens0% turns55% of whole cases pass with skill (12/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-06	✗→✓	▲ Improved	452%	0%
case-02	✗→✓	▲ Improved	874%	0%
case-04	✗→✓	▲ Improved	265%	0%
case-05	✗→✓	▲ Improved	690%	0%
case-08	✗→✓	▲ Improved	366%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

mkurman / zorai

skills/productivity/agent-skills/upstash-redis-kv/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1

Skill contents

SKILL.md

Upstash Redis Key-Value Store

Interact with Upstash's Redis-compatible key-value store using the REST interface.

Script Location

bash
bun run scripts/upstash-client.ts <command> [args...]

IMPORTANT: Always run with bun run, not directly.

Configuration

Environment Variables

The script uses these environment variables by default:

UPSTASH_REDIS_REST_URL - The Upstash REST API URL
UPSTASH_REDIS_REST_TOKEN - The Upstash REST API token

Overriding Credentials

If the user provides credentials from another source (conversation context, a file, etc.), use the --url and --token flags to override environment variables:

bash
bun run scripts/upstash-client.ts --url "https://..." --token "AX..." GET mykey

Priority: Command-line flags > Environment variables

Command Reference

String Commands

bash
# Get/Set
GET <key>
SET <key> <value> [--ex seconds] [--px ms] [--nx] [--xx] [--keepttl] [--get]
SETNX <key> <value>                    # Set if not exists
SETEX <key> <seconds> <value>          # Set with expiration

# Multiple keys (key/value pairs)
MGET <key1> [key2...]
MSET <key1> <val1> [key2 val2...]
MSETNX <key1> <val1> [key2 val2...]    # Set all if none exist

# Counters
INCR <key>
INCRBY <key> <increment>
INCRBYFLOAT <key> <increment>
DECR <key>
DECRBY <key> <decrement>

# String manipulation
APPEND <key> <value>
STRLEN <key>
GETRANGE <key> <start> <end>
SETRANGE <key> <offset> <value>

Hash Commands

Hashes store field-value pairs. Pass fields and values as alternating arguments:

bash
# Set hash fields (field/value pairs)
HSET <key> <field1> <val1> [field2 val2...]
HSETNX <key> <field> <value>           # Set field if not exists

# Get hash fields
HGET <key> <field>
HMGET <key> <field1> [field2...]
HGETALL <key>

# Hash operations
HDEL <key> <field1> [field2...]
HEXISTS <key> <field>
HKEYS <key>
HVALS <key>
HLEN <key>
HINCRBY <key> <field> <increment>
HINCRBYFLOAT <key> <field> <increment>
HSCAN <key> <cursor> [MATCH pattern] [COUNT count]

Examples:

bash
# Store user data
bun run scripts/upstash-client.ts HSET user:1 name "John" email "john@example.com" age 30

# Get single field
bun run scripts/upstash-client.ts HGET user:1 name

# Get all fields
bun run scripts/upstash-client.ts HGETALL user:1

# Increment numeric field
bun run scripts/upstash-client.ts HINCRBY user:1 age 1

List Commands

Lists are ordered collections. Values are pushed/popped from left (head) or right (tail):

bash
# Push elements
LPUSH <key> <val1> [val2...]           # Push to head
RPUSH <key> <val1> [val2...]           # Push to tail
LPUSHX <key> <val1> [val2...]          # Push if list exists
RPUSHX <key> <val1> [val2...]

# Pop elements
LPOP <key> [count]
RPOP <key> [count]

# Access elements
LRANGE <key> <start> <stop>            # Get range (0 = first, -1 = last)
LLEN <key>
LINDEX <key> <index>

# Modify
LSET <key> <index> <value>
LREM <key> <count> <value>             # Remove count occurrences
LTRIM <key> <start> <stop>             # Keep only range
LINSERT <key> <BEFORE|AFTER> <pivot> <value>
LPOS <key> <value>
LMOVE <src> <dst> <LEFT|RIGHT> <LEFT|RIGHT>

Examples:

bash
# Create a task queue
bun run scripts/upstash-client.ts RPUSH tasks "task1" "task2" "task3"

# Get all tasks
bun run scripts/upstash-client.ts LRANGE tasks 0 -1

# Pop task from front (FIFO queue)
bun run scripts/upstash-client.ts LPOP tasks

# Pop task from back (LIFO stack)
bun run scripts/upstash-client.ts RPOP tasks

Set Commands

Sets store unique, unordered members:

bash
# Add/remove members
SADD <key> <member1> [member2...]
SREM <key> <member1> [member2...]

# Query
SMEMBERS <key>
SISMEMBER <key> <member>
SMISMEMBER <key> <member1> [member2...]
SCARD <key>

# Random access
SPOP <key> [count]
SRANDMEMBER <key> [count]

# Set operations
SINTER <key1> [key2...]
SINTERSTORE <dest> <key1> [key2...]
SUNION <key1> [key2...]
SUNIONSTORE <dest> <key1> [key2...]
SDIFF <key1> [key2...]
SDIFFSTORE <dest> <key1> [key2...]
SMOVE <src> <dst> <member>
SSCAN <key> <cursor> [MATCH pattern] [COUNT count]

Examples:

bash
# Add tags
bun run scripts/upstash-client.ts SADD article:1:tags "javascript" "redis" "nodejs"

# Check membership
bun run scripts/upstash-client.ts SISMEMBER article:1:tags "javascript"

# Get all members
bun run scripts/upstash-client.ts SMEMBERS article:1:tags

# Find common tags between articles
bun run scripts/upstash-client.ts SINTER article:1:tags article:2:tags

Sorted Set Commands

Sorted sets store members with scores for ranking:

bash
# Add members (score/member pairs)
ZADD <key> <score1> <member1> [score2 member2...] [--nx] [--xx] [--gt] [--lt] [--ch]

# Remove
ZREM <key> <member1> [member2...]
ZREMRANGEBYRANK <key> <start> <stop>
ZREMRANGEBYSCORE <key> <min> <max>

# Scores and ranks
ZSCORE <key> <member>
ZMSCORE <key> <member1> [member2...]
ZRANK <key> <member>                   # Rank (low to high)
ZREVRANK <key> <member>                # Rank (high to low)
ZINCRBY <key> <increment> <member>

# Range queries
ZRANGE <key> <start> <stop> [--withscores] [--rev] [--byscore] [--bylex]
ZRANGEBYSCORE <key> <min> <max> [--withscores] [--limit off,count]
ZREVRANGE <key> <start> <stop> [--withscores]
ZREVRANGEBYSCORE <key> <max> <min> [--withscores] [--limit off,count]

# Counting
ZCARD <key>
ZCOUNT <key> <min> <max>

# Pop
ZPOPMIN <key> [count]
ZPOPMAX <key> [count]

# Set operations
ZINTERSTORE <dest> <numkeys> <key1> [key2...]
ZUNIONSTORE <dest> <numkeys> <key1> [key2...]
ZSCAN <key> <cursor> [MATCH pattern] [COUNT count]

Examples:

bash
# Create leaderboard (score member pairs)
bun run scripts/upstash-client.ts ZADD leaderboard 1000 "player1" 1500 "player2" 1200 "player3"

# Get top 3 with scores (highest first)
bun run scripts/upstash-client.ts ZRANGE leaderboard 0 2 --rev --withscores

# Get player's rank
bun run scripts/upstash-client.ts ZREVRANK leaderboard "player2"

# Increment player's score
bun run scripts/upstash-client.ts ZINCRBY leaderboard 100 "player1"

# Get players with scores between 1000 and 1500
bun run scripts/upstash-client.ts ZRANGEBYSCORE leaderboard 1000 1500 --withscores

Key Commands

bash
# Delete
DEL <key1> [key2...]
UNLINK <key1> [key2...]                # Async delete

# Existence/Type
EXISTS <key1> [key2...]
TYPE <key>

# Expiration
EXPIRE <key> <seconds>
EXPIREAT <key> <timestamp>
PEXPIRE <key> <milliseconds>
PEXPIREAT <key> <timestamp>
TTL <key>
PTTL <key>
PERSIST <key>                          # Remove expiration

# Rename
RENAME <key> <newkey>
RENAMENX <key> <newkey>

# Search
KEYS <pattern>                         # Use with caution in production
SCAN <cursor> [MATCH pattern] [COUNT count]

# Other
COPY <src> <dst>
DUMP <key>
TOUCH <key1> [key2...]
RANDOMKEY
OBJECT ENCODING|FREQ|IDLETIME|REFCOUNT <key>

Examples:

bash
# Set key with 1 hour expiration
bun run scripts/upstash-client.ts SET session:abc "data"
bun run scripts/upstash-client.ts EXPIRE session:abc 3600

# Or in one command
bun run scripts/upstash-client.ts SET session:abc "data" --ex 3600

# Check TTL
bun run scripts/upstash-client.ts TTL session:abc

# Scan keys matching pattern
bun run scripts/upstash-client.ts SCAN 0 MATCH "user:*" COUNT 100

Server Commands

bash
PING [message]
ECHO <message>
DBSIZE
TIME
INFO [section]
FLUSHDB                                # Delete all keys in current DB (DANGEROUS)
FLUSHALL                               # Delete all keys in all DBs (DANGEROUS)

Command Options

SET Options

bash
--ex <seconds>     # Expire in seconds
--px <ms>          # Expire in milliseconds
--exat <ts>        # Expire at Unix timestamp (seconds)
--pxat <ts>        # Expire at Unix timestamp (ms)
--nx               # Only set if key does not exist
--xx               # Only set if key exists
--keepttl          # Retain existing TTL
--get              # Return old value

ZADD Options

bash
--nx               # Only add new members
--xx               # Only update existing members
--gt               # Only update if new score > current
--lt               # Only update if new score < current
--ch               # Return number of changed elements

ZRANGE Options

bash
--withscores       # Include scores in output
--byscore          # Range by score instead of rank
--bylex            # Range by lexicographical order
--rev              # Reverse order
--limit off,count  # Limit results (e.g., --limit 0,10)

Output Format

String values: Printed directly
null/nil: Prints (nil)
Objects/arrays: Pretty-printed as JSON

Confirmation Behaviour

Default: Ask for Confirmation

Before executing any destructive operation (write, modify, or delete), you MUST ask the user for confirmation. This includes:

Write operations:

SET, SETNX, SETEX, PSETEX, MSET, MSETNX
HSET, HSETNX
LPUSH, RPUSH, LPUSHX, RPUSHX, LSET, LINSERT
SADD
ZADD

Modify operations:

INCR, INCRBY, INCRBYFLOAT, DECR, DECRBY
APPEND, SETRANGE
HINCRBY, HINCRBYFLOAT
LREM, LTRIM, LMOVE
ZINCRBY

Delete operations:

DEL, UNLINK
HDEL
LPOP, RPOP
SREM, SPOP, SMOVE
ZREM, ZREMRANGEBYRANK, ZREMRANGEBYSCORE, ZPOPMIN, ZPOPMAX
FLUSHDB, FLUSHALL (extra caution)

TTL/Rename operations:

EXPIRE, EXPIREAT, PEXPIRE, PEXPIREAT, PERSIST
RENAME, RENAMENX

Example confirmation prompt: > "I'm about to HSET user:1 with fields {name: "John", email: "john@example.com"}. Proceed?"

YOLO Mode: Skip Confirmation

If the user indicates they do not want to be asked for confirmation, respect this for all subsequent operations. Indicators include:

"YOLO mode"
"Don't ask for confirmation"
"You're free to make changes without asking"
"Just do it"
"No need to confirm"
"Auto-approve" or "auto-confirm"
Any similar phrasing indicating blanket approval

Once YOLO mode is activated, proceed with destructive operations without asking, but still inform the user what was done.

Example: > Set user:1 with {name: "John", email: "john@example.com"} - done.

Error Handling

If credentials are missing or invalid, the script will exit with an error message. Ensure the user has configured either:

Environment variables (UPSTASH_REDIS_REST_URL, UPSTASH_REDIS_REST_TOKEN)
Or provides credentials via --url and --token flags

When to Use This Skill

User explicitly asks to store or retrieve data from Upstash/Redis
Need to persist data across conversations or sessions
Implementing caching for expensive operations
Maintaining counters, rate limits, or statistics
Storing user preferences or session data
Building leaderboards or rankings (sorted sets)
Managing queues or task lists (lists)
Tagging or categorisation (sets)
Storing structured objects (hashes)
Any scenario requiring fast key-value storage

Want to see if mkurman/upstash-redis-kv would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/21/2026, 22:41 UTC · 22 cases

Measured on 16 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+41%

55% vs 14%

Agent turns

1.0 vs 1.0

Tokens

+373%

3,908.31 vs 825.56

fail → pass · 9pass → pass · 3fail → fail · 10

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-21 How do I send a JSON payload to an HTTP target endpoint using Upstash QStash via curl?	pass→pass	9,006	6,436	-29%	1	1	0%	1,730	4,636	+168%	0	0	—
case-06 Fetch the top 10 highest scoring items along with their scores from the sorted set key 'weekly_scores'.	fail→pass	3,801	4,986	+31%	1	1	0%	676	3,734	+452%	0	0	—
case-22 How do I query an Upstash Vector index using raw HTTP POST with a vector array [0.1, 0.2, 0.3] and topK of 5?	pass→pass	8,078	6,709	-17%	1	1	0%	1,517	4,707	+210%	0	0	—
case-15 Push the item 'event_start' to the front (head) of the list key 'audit_log'. You have blanket approval to apply changes.	fail→fail	2,824	4,342	+54%	1	1	0%	394	3,617	+818%	0	0	—
case-01 Here are my Upstash API details: URL is `https://my-db.upstash.io` and token is `AX123456`. Can you fetch all the attributes stored under the hash key `user:1002` and print the resulting data?	fail→fail	6,628	8,227	+24%	1	1	0%	1,290	3,780	+193%	0	0	—
case-02 I need to update our high scores database. Please add player 'Alex' with score 4500 to the 'game_leaderboard' key. No need to confirm any write actions, just go ahead and make the change, then let me know once it is complete.	fail→pass	2,483	5,109	+106%	1	1	0%	377	3,671	+874%	0	0	—
case-03 Can you retrieve and remove the top item from our list key 'job_queue' in Redis? Show me what task was extracted.	fail→fail	6,326	7,464	+18%	1	1	0%	1,165	3,870	+232%	0	0	—
case-04 I want to cache a user session token 'sess_88321' with value 'active_user' for 600 seconds, but only if the session token key does not already exist. How should this command be invoked using our client script?	fail→pass	6,243	3,645	-42%	1	1	0%	1,132	4,132	+265%	0	0	—
case-05 We want to retrieve the first 5 elements from the list key 'activity_stream' without modifying or deleting anything. What command should be executed?	fail→pass	3,002	3,224	+7%	1	1	0%	497	3,925	+690%	0	0	—
case-07 Increment the 'visit_count' field in the hash key 'analytics:page:home' by 1.	fail→fail	3,773	2,335	-38%	1	1	0%	612	3,754	+513%	0	0	—
case-08 Check whether the user ID 'usr_99' is present in the set key 'beta_testers'.	fail→pass	4,451	2,324	-48%	1	1	0%	810	3,775	+366%	0	0	—
case-09 Remove the key 'temp_cache:101' from the database.	fail→fail	4,389	2,234	-49%	1	1	0%	710	3,741	+427%	0	0	—
case-10 Auto-approve mode is on. Add tags 'sports', 'news', and 'trending' to the set key 'article:50:tags'.	fail→fail	1,753	4,794	+173%	1	1	0%	231	3,628	+1471%	0	0	—
case-20 How do I configure the save parameter in redis.conf for snapshotting a self-hosted Redis instance to disk every 60 seconds if at least 1000 keys changed?	pass→pass	7,977	4,819	-40%	1	1	0%	1,346	4,267	+217%	0	0	—
case-11 Find all keys matching the pattern 'session:*' starting at cursor 0 with a count limit of 50.	fail→pass	3,111	1,638	-47%	1	1	0%	486	3,650	+651%	0	0	—
case-12 I'm giving you full permission to make changes without asking for confirmation. Store the following attributes for key 'product:20': title as 'Wireless Mouse', price as 29, stock as 150.	fail→fail	2,296	4,998	+118%	1	1	0%	361	3,684	+920%	0	0	—
case-13 Determine the 0-based rank of player 'Sam' in the sorted set 'leaderboard' ordered from highest score to lowest score.	fail→fail	2,762	5,854	+112%	1	1	0%	470	3,726	+693%	0	0	—
case-14 Check how many seconds remain before key 'auth:token:abc' expires.	fail→pass	2,316	4,326	+87%	1	1	0%	377	3,635	+864%	0	0	—
case-16 Set key 'report:daily' to expire in 3600 seconds.	fail→pass	3,159	2,144	-32%	1	1	0%	460	3,736	+712%	0	0	—
case-17 Retrieve the values stored in keys 'config:feature_a', 'config:feature_b', and 'config:feature_c' at once.	fail→fail	2,491	4,975	+100%	1	1	0%	332	3,733	+1024%	0	0	—
case-18 Clear all keys from the current database.	fail→fail	4,761	1,683	-65%	1	1	0%	760	3,624	+377%	0	0	—
case-19 Just do it without asking for confirmation: remove all entries from sorted set 'events_timeline' with scores between 0 and 1000.	fail→pass	3,291	1,406	-57%	1	1	0%	554	3,676	+564%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 16 counted toward the lift figure. The other 6 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +41 percentage points is the difference between those two pass rates over the 16 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 13.6%
Pass rate with the skill: 54.6%
Cases attempted: 22
Cases counted toward lift: 16
Cases that regressed: 4
Skill version measured: v1
Run completed: 8/21/2026, 22:41 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-06

Prompt

Fetch the top 10 highest scoring items along with their scores from the sorted set key 'weekly_scores'.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run55

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+196% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 7 days ago

Other measured skills in the registry, with their headline benchmark lift.

Ready to supercharge your AI agents?

mkurman/upstash-redis-kv

SkillSafety

Source

Upstash Redis Key-Value Store

Script Location

Configuration

Environment Variables

Overriding Credentials

Command Reference

String Commands

Hash Commands

List Commands

Set Commands

Sorted Set Commands

Key Commands

Server Commands

Command Options

SET Options

ZADD Options

ZRANGE Options

Output Format

Confirmation Behaviour

Default: Ask for Confirmation

YOLO Mode: Skip Confirmation

Error Handling

When to Use This Skill

How this skill was measured

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-06

Related skills