alignment-checkpoint

Create, review, and validate an alignment checkpoint. Use when a request is ambiguous, high-stakes, multi-step, or requires explicit approval before tool use.

Quality

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Passed

No known issues

Quality

Content

62%

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

The skill has a strong, well-sequenced workflow with explicit validation feedback loops, but its instructional steps are abstract and its deferred references do not map to the real bundle files.

Suggestions

Make the workflow steps more concrete by specifying what to inspect (e.g., the request text, constraints, and existing repo contracts) and how to emit outputs, or point to the contract.yaml schema for the exact output shape.

Link or name the actual bundle files (contract.yaml, evals.yaml, task-profile.json) from the Progressive Disclosure section so navigation reaches the real archived material, and reconcile the referenced path with the on-disk layout.

De-duplicate the overlapping Avoid, Constraints, and Anti-Patterns bullets and drop the verbatim repeat of the description to tighten conciseness toward level 3.

Dimension	Reasoning	Score
Conciseness	The body is terse and well-sectioned, but the frontmatter description is repeated verbatim and the Avoid, Constraints, and Anti-Patterns sections overlap conceptually (e.g., rewrites-without-approval and evidence-before-success appear multiple times), so it could be tightened past the level-2 anchor.	2 / 3
Actionability	As an instruction-only skill it reasonably avoids code, but steps like "Take the smallest action that advances the confirmed goal" and "Inspect 2-3 focused surfaces" are directional rather than concrete, leaving key details (which surfaces, which validator, how outputs are formatted) unspecified.	2 / 3
Workflow Clarity	The five-step Workflow is clearly sequenced with explicit validation checkpoints ("Stop at the first failed gate" in step 4) and a fix-then-rerun feedback loop (step 5 plus the Validation section's "do not proceed until it is fixed and rerun"), matching the level-3 anchor.	3 / 3
Progressive Disclosure	The body is organized as an overview with a dedicated Progressive Disclosure section, but the referenced archived path (`Infrastructure/references/deferred-skill-context/...`) does not correspond to the actual bundle files in `./references/` (contract.yaml, evals.yaml, task-profile.json), none of which are named or linked, so references are present but not clearly signaled.	2 / 3
	Total	9 / 12 Passed

Description

67%

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 cleanly covers both capability and an explicit Use-when trigger, but its actions and trigger terms are somewhat abstract and situational, limiting specificity and distinctiveness.

Suggestions

Replace the abstract object "an alignment checkpoint" with the concrete artifacts the skill produces (e.g., a goal/assumptions/options summary and an approval gate) to lift specificity toward level 3.

Add more natural noun-form trigger keywords users would actually say (e.g., "clarify a request", "get approval", "confirm a plan") alongside the situational descriptors.

Sharpen distinctiveness by naming the gating behavior ("hold tool use until the user approves a direction") so it is less likely to fire for generic planning skills.

Dimension	Reasoning	Score
Specificity	It lists three verbs ("Create, review, and validate") but applies them to a single abstract noun ("an alignment checkpoint"), so the actions are named yet not as concrete as distinct operations on distinct objects, matching the level-2 anchor rather than the multi-action level-3 example.	2 / 3
Completeness	It clearly states what ("Create, review, and validate an alignment checkpoint") and an explicit when clause ("Use when a request is ambiguous, high-stakes, multi-step, or requires explicit approval before tool use"), satisfying both halves with an explicit trigger.	3 / 3
Trigger Term Quality	Trigger terms like "ambiguous", "high-stakes", "multi-step", and "explicit approval" are situational descriptors a user might say, but they lack the varied concrete keyword coverage (e.g., multiple noun forms) seen in the level-3 anchor.	2 / 3
Distinctiveness Conflict Risk	"alignment checkpoint" is a recognizable niche, but the situational triggers (ambiguous, high-stakes, multi-step) are broad enough to overlap with other planning or clarification skills, so it is not yet a clearly conflict-free niche.	2 / 3
	Total	9 / 12 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: jscraik/Agent-Skills
Commit: 5a6027f

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.