agentsclimarketplace

Telemetry sanitized skill invocation

Skill vishuwa2004/cskill-agents/agents/claude-code/skills/telemetry-sanitized-skill-invocation

Build curated agent skills for coding CLIs, terminal tools, context engines, remote bridges, and multi-agent runtimes

Install
npx -y skills add vishuwa2004/cskill-agents --skill telemetry-sanitized-skill-invocation

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

One thing to look at

  • 2 stars2 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

Record SkillTool usage with sanitized telemetry fields so built-in, bundled, and custom skills are distinguished while avoiding PII leakage.

SKILL.md

2.7 KB, 522 tokens by cl100k_base, as published. Nobody here has run it

SKILL: Telemetry Sanitized Skill Invocation

Domain: tool-orchestration
Trigger: Use when SkillTool emits tengu_skill_tool_invocation so built-in commands, remote MCP skills, and forked executions are logged without exposing command strings in open analytics tables. Source Pattern: Distilled from reviewed tool-loop and result-shaping patterns.

Core Method

Before executing any skill, compute whether the command is built-in, bundled, official marketplace, or custom and pass the sanitized value as command_name while storing the actual skill name under _PROTO_skill_name to satisfy both privacy and auditability. Add contextual metadata such as execution_context (inline, forked, remote), query_depth, parent_agent_id, and plugin fields when applicable, and wrap remote canonical invocations or discovered skills with additional flags so downstream dashboards know the skill source without revealing user-generated text. Emit separate telemetry for inline/forked/remote flows so metrics can split by execution path while normalizing the skill name for aggregated reporting.

Key Rules

  • Derive sanitizedCommandName as commandName for built-in/bundled/official skills and fall back to 'custom' for user-authored commands; log the actual name only in _PROTO_skill_name to keep BigQuery columns safe.
  • Always include execution_context ('inline', 'forked', 'remote') plus invocation_trigger (nested vs proactive) and query_depth so telemetry consumers understand the runtime scenario.
  • When a plugin skill is involved, add both _PROTO_plugin_name and the sanitized plugin_name/plugin_repository fields along with marketplace metadata to enable plugin-level analytics without leaking repository URLs.
  • For remote MCP skills, emit is_remote, was_discovered, and cached load info (latency, cache hit) so telemetry can diagnose discovery performance without exposing the raw prompt body.

Example Application

Use this skill to log any slash command invocation, ensuring that a cloned forked review or remote_skill call still records an execution_context flag plus proper naming sanitization for consumption by privacy-sensitive dashboards.

Anti-Patterns (What NOT to do)

  • Do not log user-entered command text directly into public telemetry columns; that breaks PII rules and makes analytics unusable.
  • Do not merge inline/forked/remote flows in a single metric stream without flags, because it becomes impossible to trace delays or diagnostics per execution context.

Gives 0 of the 12 instructions most monitoring observability skills give in 522 tokens

Counted across 481 of the 483 authors here whose files we hold, read 2026-08-06

  • link every alert to a runbookin 43 of 481, across 35 files
  • use structured json loggingin 36 of 481, across 31 files
  • alert on user-facing symptomsin 20 of 481, across 15 files
  • emit structured JSON logs with stable event namesin 18 of 481, across 13 files
  • propagate trace context across boundariesin 16 of 481
  • use histograms for latency trackingin 14 of 481, across 9 files
  • use OpenTelemetry for distributed tracingin 13 of 481, across 8 files
  • include a correlation ID on every log linein 13 of 481, across 8 files
  • Define service level objectivesin 10 of 481, across 7 files
  • Call useAzureMonitor before importing other modulesin 9 of 481, across 2 files
  • stop and ask for clarification if inputs are missingin 9 of 481, across 2 files
  • define on-call questions before adding telemetryin 9 of 481, across 4 files

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.