Skill Cost
v0.1.0Track per-skill token usage and costs from OpenClaw session logs. Use when user asks about skill-level spending, which skill costs the most, or wants a per-s...
⭐ 1· 294·1 current·1 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description claim to compute per-skill token costs from OpenClaw session logs; the packaged scripts (bash wrapper + Python modules) implement exactly that behavior (discover agents, parse session JSONL, attribute tool calls to skills, compute costs). Required binaries (python3, bash) are appropriate and there are no unrelated credentials or surprising dependencies.
Instruction Scope
Runtime instructions explicitly require using the bash wrapper which runs the included Python scripts. The scripts read local session JSONL files (~/.openclaw/agents/*/sessions/) and scan installed skills' SKILL.md frontmatter to map tool names to skill names. This is consistent with the stated purpose, but note: it processes local assistant messages and tool-call contents (potentially sensitive conversation data) — the skill reads and analyzes that data locally.
Install Mechanism
No install spec is provided (instruction-only install), and code is packaged with the skill so nothing is downloaded from the network at install time. The wrapper and Python scripts are included in the skill bundle; installing the skill will place those files on disk, which is expected for this utility.
Credentials
The skill requests no environment variables, no credentials, and no config paths beyond reading standard OpenClaw directories. The scope of access (local session files and SKILL.md files under the user's home path) is proportional to the goal of attributing cost to skills.
Persistence & Privilege
The skill is not marked always:true and does not request elevated agent-wide privileges or modify other skills' configurations. It runs on-demand via the bash wrapper; autonomous invocation is the platform default and is not combined here with other alarming flags.
Assessment
This skill appears coherent and implements what it claims: it scans your local OpenClaw session logs and local SKILL.md files to attribute token usage to skills. Before installing, consider: (1) the tool will read and process assistant messages and tool-call contents from your session logs — which can include sensitive data — so only run it if you are comfortable having those logs parsed locally; (2) because the skill executes Python code from the installed files, review the packaged scripts (especially any remaining parts of cost_utils.py that parse message content) yourself or run in an isolated environment if you have concerns; (3) no network calls or credentials are requested by the code as provided, but if you fork or update the skill later, re-check for added network/credential usage.Like a lobster shell, security has layers — review code before you run it.
latestvk97fpwt5qamyq9jmr5wxt52ytn82hee4
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
💰 Clawdis
OSLinux · macOS · Windows
Any binpython3, bash