context-xray

Inspect and manage the live agent context window with Context X-Ray. Use when context is getting large, the user asks what is in context, or stale tool results/files should be pinned, evicted, restored, or reported by an external host.

Quality

85%

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

75%Weight 40%Scale 1-5

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

A well-organized, information-dense instruction skill: it documents exact action semantics, safety rules, and a sequenced typical flow without padding. Its main gaps are the absence of a post-eviction verification step and a System Sections section detailed enough to warrant its own reference file.

Suggestions

Add a validation checkpoint to the Typical Flow, e.g. step 6: re-run `context-manifest-get` to confirm the expected token reclaim before reporting savings.

Trim changelog phrasing ("Manifests now cover...") and product-UI narration (Agent page Context tab, split meter) that do not change what the agent should do.

Move the provenance-label enumeration and governance-tier details from System Sections into a reference file (e.g. references/system-sections.md) and link to it from a two-line summary.

Dimension	Reasoning	Score
Conciseness	The body is dense and mostly lean — every action entry and rule carries non-obvious, domain-specific information — but changelog-style phrasing ("Manifests now cover the system-prompt half") and UI narration ("the Agent page Context tab renders the full breakdown... with a system-vs-conversation split meter") are minor instances that could be trimmed. Efficient with minor trims possible fits anchor 4, not 5 where every token earns its place.	4 / 5
Actionability	Concrete guidance throughout: exact action names with parameters ("Call `context-manifest-get` with the active `threadId`", "Sort segments by `tokenCount`") plus a when-to-use column per action. It falls short of anchor 5 because no actual invocation syntax or worked example is given, leaving minor gaps in executability.	4 / 5
Workflow Clarity	The Typical Flow is a clearly sequenced 5-step procedure with an undo path ("Offer `context-restore` if the user wants undo") and a protective pre-condition ("Never evict or summarize protected segments"). It is anchor 4 rather than 5 because there is no verification checkpoint (e.g. re-read the manifest to confirm token reclaim after evictions). The destructive/batch cap does not apply since eviction is explicitly reversible and never deletes history.	4 / 5
Progressive Disclosure	Sections (Actions, System Sections, Rules, Typical Flow) are well organized and there are no buried or nested references — nothing is referenced at all. However the ~61-line body exceeds the under-50-line simple-skill threshold, and the enumerated provenance labels plus governance-tier detail in System Sections would naturally belong in a separate reference file, giving minor organization gaps consistent with anchor 4.	4 / 5
	Total	16 / 20 Passed

Description

92%Weight 40%Scale 1-5

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

A strong description: it states concrete third-person capabilities and pairs them with explicit, naturally-phrased trigger conditions covering both the what and the when. The only weakness is moderate synonym coverage for how users describe context pressure.

Dimension	Reasoning	Score
Specificity	The description names six concrete operations in third person — "Inspect and manage the live agent context window", "pinned, evicted, restored, or reported" — covering every action the skill body exposes, matching the comprehensive-coverage anchor. It does not fit anchor 4 because there is no gap in the action coverage.	5 / 5
Completeness	"Inspect and manage the live agent context window" answers what the skill does, and "Use when context is getting large, the user asks what is in context, or stale tool results/files should be pinned, evicted, restored, or reported by an external host" gives concrete trigger phrases for when. Both what and when are explicitly and clearly stated, matching anchor 5.	5 / 5
Trigger Term Quality	Natural phrasings like "context is getting large", "what is in context", and "stale tool results/files" would match how users actually ask, but common synonyms (e.g. "context window is full", "free up tokens", "trim context") are absent. Good coverage with a few natural terms missing fits anchor 4 rather than 5.	4 / 5
Distinctiveness Conflict Risk	The niche (live agent context-window segment management with pin/evict/restore semantics) is distinct and no generic skill would collide with these triggers. Clear niche with minimal conflict risk matches anchor 5 rather than anchor 4, which requires some overlap with closely related skills.	5 / 5
	Total	19 / 20 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
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning

	Total	14 / 16 Passed

Repository: BuilderIO/agent-native
Path: .agents/skills/context-xray/SKILL.md
Commit: a941a2e

Reviewed: 9 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.