ax-cpp-agent-observability

Use when writing C++ code with `axllm` for agent tracing, centralized and multi-tenant usage accounting, action logs, runtime diagnostics, replay, and production debugging.

Quality

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

87%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is concise, actionable, and well-structured with one-level-deep references to package materials. Its main gap is the absence of explicit validation/checkpoint sequences for the observer lifecycle and package-regeneration workflow, which keeps workflow clarity below the top anchor.

Suggestions

Add an explicit ordered lifecycle for the usage observer — register at startup, enqueue synchronously, clear during teardown — with a validation checkpoint confirming the observer is unset after shutdown.

For the 'Treat AxIR as source of truth' guardrail, spell out the multi-step sequence: detect the docs/source disagreement, update the compiler, regenerate packages, then re-validate the package API docs — so the workflow has explicit checkpoints instead of a one-line instruction.

Note in the body which referenced files (e.g. API.md, examples/) ship inside the `axllm` package versus the repo, so the agent knows where to look and does not hunt for missing bundle files.

Dimension	Reasoning	Score
Conciseness	The body is lean and assumes Claude's competence — package facts, a core pattern, and a focused observer section with executable code — with no padding explaining what C++, tracing, or usage accounting are. Every section earns its tokens.	3 / 3
Actionability	It provides fully executable C++ (`axllm::agent(...)`, `axllm::set_usage_observer(...)`) plus concrete file pointers like `src/examples/cpp/generation/usage_observer.cpp`, named API symbols, and explicit runtime profiles — copy-paste ready guidance rather than pseudocode.	3 / 3
Workflow Clarity	The skill is primarily reference material, and while sections are well ordered, there is no explicit multi-step sequence with validation checkpoints for the riskier operations it touches (e.g., the observer registration/teardown lifecycle, regenerating packages after a compiler update). Not score 1 because structure and an implied lifecycle exist; not score 3 because there is no explicit validate/feedback loop.	2 / 3
Progressive Disclosure	It is a well-organized overview that points one level deep to real bundle paths (`API.md`, `axir-api.json`, `examples/`, `src/examples/cpp/generation/usage_observer.cpp`) rather than inlining them; no nested reference chains and easy navigation. No references/ scripts/ or assets/ bundles are present in this skill directory, but the referenced paths are the package's own files, signalled clearly.	3 / 3
	Total	11 / 12 Passed

Description

85%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, complete with an explicit trigger clause, and occupies a clear niche, with only slightly thin coverage of everyday user phrasings. It avoids vague fluff and over-claims while naming concrete capabilities.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capabilities — "agent tracing", "centralized and multi-tenant usage accounting", "action logs", "runtime diagnostics", "replay", and "production debugging" — each a specific action rather than vague domain language.	3 / 3
Completeness	It explicitly answers both what the skill does (the enumerated capabilities) and when to use it via the explicit "Use when..." clause, satisfying the rubric's anchor for clearly answering both.	3 / 3
Trigger Term Quality	It opens with a clear trigger ("Use when writing C++ code with `axllm`") and uses recognizable terms ("agent tracing", "usage accounting", "debugging"), but coverage of natural user phrasings is partial — a user would more often say "tracing", "logging", "usage/cost tracking", or "debug my agent" rather than the denser multi-tenant-accounting phrasing. Not score 3 because common variations (e.g., "cost tracking", "telemetry") are absent; not score 1 because genuine natural keywords are present.	2 / 3
Distinctiveness Conflict Risk	The C++ + `axllm` + agent-observability niche is tightly scoped with distinct triggers (usage observer, replay, production debugging), making it unlikely to fire for unrelated skills.	3 / 3
	Total	11 / 12 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 1 suspicious	Warning

	Total	14 / 16 Passed

Repository: ax-llm/ax
Path: packages/cpp/skills/ax-cpp-agent-observability/SKILL.md
Commit: 9087a01

Reviewed: 5 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.