← Back to Skills Marketplace
Promptcache
by
AVale-SLAI
· GitHub ↗
· v1.0.3
· MIT-0
149
Downloads
0
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install promptcache
Description
Estimate the cost savings from caching frequently-used prompts across AI models.
Usage Guidance
Do not run the included install.sh or trust a missing CLI without checking source. Before installing: 1) Ask the publisher for the promptcache executable or a reproducible install method (scripts/promptcache is missing). 2) Confirm why SL_API_KEY is needed and have the skill declare it in metadata if required. 3) Inspect or mirror the code for the promptcache CLI (if provided) to ensure it doesn't exfiltrate data. 4) If you must test, run the installer in a disposable environment (container/VM) and monitor network calls (the script sends an unauthenticated telemetry POST to api.signalloomai.com). 5) Prefer skills with a verifiable homepage/source and consistent versioning/packaging.
Capability Analysis
Type: OpenClaw Skill
Name: promptcache
Version: 1.0.3
The skill is a utility for estimating AI cost savings from prompt caching. The installation script (install.sh) performs standard setup tasks, such as symlinking the skill and updating the user's PATH in .zshrc. It includes a transparent analytics ping to api.signalloomai.com to track installations, which is common for such tools and does not involve sensitive data exfiltration.
Capability Assessment
Purpose & Capability
The skill claims to estimate prompt-caching savings and requires a 'promptcache' CLI, which is reasonable in principle, but the bundle does not include that binary or a script to install it. package.json points to ./scripts/promptcache which is absent. Version numbers also disagree across files (package.json 1.0.1, install.sh 1.0.2, registry 1.0.3). These mismatches suggest the package is incomplete or incorrectly packaged.
Instruction Scope
SKILL.md is narrowly scoped to cost estimation, but the included install.sh modifies user files (~/.openclaw/skills, ~/.local/bin, appends to ~/.zshrc) and issues an unauthenticated analytics POST to https://api.signalloomai.com/v1/analytics/install. The SKILL.md instructs users to set SL_API_KEY but the registry metadata declares no required env vars — a clear mismatch.
Install Mechanism
There is no install spec in the registry, but an install.sh is present. The script performs only local filesystem changes (symlink, PATH export) and sends a lightweight analytics ping; it does not download or execute remote code. That lowers supply-chain risk, but the presence of a background network call to signalloomai.com should be expected and disclosed.
Credentials
Registry metadata lists no required environment variables, yet SKILL.md and install.sh reference an SL_API_KEY and prompt users to export it. Asking for an API key for an external service is plausible, but the missing declaration is an inconsistency. Also the skill requires a CLI binary that isn't bundled, which may implicitly require installing or trusting an external binary from an unknown origin.
Persistence & Privilege
always:false and no elevated privileges are requested. The install script writes into the user's home (symlink into ~/.openclaw/skills and appends to ~/.zshrc) which is expected for a user-level install but is persistent at the user profile level. No modification of other skills or system-wide settings beyond PATH is observed.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install promptcache - After installation, invoke the skill by name or use
/promptcache - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.3
Version 1.0.3
- Added install.sh script to assist with installation.
- No changes to skill usage or documentation content.
v1.0.2
- Added package.json to the project.
- Initializes project configuration and dependencies for the skill.
v1.0.1
promptcache 1.0.1
- Improved and clarified skill description and metadata in SKILL.md.
- Simplified command syntax and examples for better usability.
- Added Openclaw-specific install metadata and requirements.
- Streamlined documentation; removed duplicate example output and model comparison tables.
v1.0.0
- Initial release of the promptcache skill.
- Estimate cost savings from caching frequently-used prompts versus re-sending full context.
- Supports major AI models, showing detailed savings per model.
- Outputs daily, monthly, and yearly cost comparisons.
- Free tier allows 3 analyses per day with a Signalloom API key.
- Simple CLI syntax with options for model selection and call frequency.
Metadata
Frequently Asked Questions
What is Promptcache?
Estimate the cost savings from caching frequently-used prompts across AI models. It is an AI Agent Skill for Claude Code / OpenClaw, with 149 downloads so far.
How do I install Promptcache?
Run "/install promptcache" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Promptcache free?
Yes, Promptcache is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Promptcache support?
Promptcache is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Promptcache?
It is built and maintained by AVale-SLAI (@avale-slai); the current version is v1.0.3.
More Skills