This is a solid skill description that clearly communicates its purpose and provides excellent trigger coverage. The 'Use when' and 'Triggers include' clauses make it very easy for Claude to know when to select this skill. The only minor weakness is that the capability description could be slightly more detailed about what the saved output includes beyond 'conversation summary'.

Dimension	Reasoning	Score
Specificity	Names the domain (saving session context) and one specific action (save session to CONTEXT-llm.md with conversation summary), but doesn't list multiple concrete actions beyond that single capability.	2 / 3
Completeness	Clearly answers both 'what' (save session to CONTEXT-llm.md with conversation summary) and 'when' (explicit 'Use when' clause plus a 'Triggers include' list with specific phrases).	3 / 3
Trigger Term Quality	Excellent coverage of natural trigger terms users would say: 'save context', 'save session', 'checkpoint', 'save my progress', 'saving work', 'preserving session state'. These are highly natural phrases a user would use.	3 / 3
Distinctiveness Conflict Risk	Highly distinctive with a specific file target (CONTEXT-llm.md), a clear niche (session/context saving), and distinct trigger terms that are unlikely to conflict with other skills.	3 / 3
	Total	11 / 12 Passed

Implementation

47%

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 well-structured multi-phase workflow with good error handling (AskUserQuestion guard) and clear sequencing, but is severely undermined by verbosity. The Philosophy, When to Use/Not to Use, and Anti-Patterns sections add significant token cost with minimal actionable value for Claude. The core output template is deferred to reference.md, which weakens actionability in the main file.

Suggestions

Remove or condense the 'Philosophy', 'When to Use', 'When Not to Use', and 'Anti-Patterns' sections to at most 5-10 bullet points total — Claude doesn't need philosophical rationale or extensive 'why' explanations for each rule.

Inline a minimal example of the CONTEXT file output template (even abbreviated) so the skill is self-contained for the most common case, rather than requiring a reference.md lookup.

Consolidate the three usage examples into one compact example showing the most common invocation and expected output files.

Remove the redundant 'why' explanations after each anti-pattern — a single-line anti-pattern statement is sufficient for Claude.

Dimension	Reasoning	Score
Conciseness	The skill is excessively verbose at ~150+ lines. The 'Philosophy', 'When to Use', 'When Not to Use', and 'Anti-Patterns' sections are largely unnecessary — Claude doesn't need philosophical justifications or extensive 'why' explanations for each anti-pattern. The usage examples repeat information already clear from the workflow. Much of this content could be cut by 50%+ without losing actionable guidance.	1 / 3
Actionability	The workflow phases provide concrete commands (e.g., `rtk ls`, `mkdir -p`, `mv`) and specific file paths, but the core synthesis step (Phase 2) is abstract — it describes what to extract but doesn't show the actual template or output format inline. The skill defers the critical CONTEXT file template to reference.md, leaving the main skill without a concrete example of the output artifact.	2 / 3
Workflow Clarity	The three-phase workflow (Gather → Analyze → Write) is clearly sequenced with explicit parallel execution instructions, stream resolution logic, and a conditional auto-archive step (Phase 3b). The AskUserQuestion guard provides a clear error recovery feedback loop. Validation is addressed through the reference.md quality self-check and the INDEX.md upsert step.	3 / 3
Progressive Disclosure	The skill correctly references `reference.md` for the template and detailed rules, which is good one-level-deep disclosure. However, the main file itself contains too much inline content (Philosophy, When to Use, When Not to Use, Anti-Patterns, Usage Examples) that bloats the overview and undermines the purpose of progressive disclosure — much of this belongs in a reference or could be drastically condensed.	2 / 3
	Total	8 / 12 Passed

Validation

81%

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 — 9 / 11 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	9 / 11 Passed

Reviewed

5 days ago

Table of Contents

Discovery Implementation Validation