session-management

Context preservation, tiered summarization, resumability

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 body is highly actionable and clearly sequenced with concrete templates, triggers, and commands, but it is far too long for a single file: it repeats its core decision logic several times and inlines every template instead of offloading them to reference files. It would benefit substantially from splitting templates into the references/ bundle and de-duplicating the checkpoint logic.

Suggestions

Move the full file templates (current-state.md, decisions.md, code-landmarks.md, archive) into separate files under references/ and replace them with one-line pointers, fixing the monolithic structure.

Remove the redundant decision logic — keep only one of the ASCII Decision Heuristic, the Quick Reference decision tree, or the CLAUDE.md self-check questions.

Tighten or trim the multi-example decisions.md and code-landmarks.md walkthroughs to a single compact example each to recover token budget.

Dimension	Reasoning	Score
Conciseness	The ~525-line body is verbose and padded: the checkpoint decision logic is repeated three times (the ASCII 'Decision Heuristic' box, the 'Quick Reference' decision tree, and the CLAUDE.md self-check questions) and full templates for current-state.md, decisions.md, code-landmarks.md, and archive entries are reproduced inline, much of which Claude could derive or that could be split out.	1 / 3
Actionability	The body gives concrete, copy-paste-ready guidance: specific triggers ('~20 tool calls', '~50+ tool calls'), a file structure with exact paths, ready-to-use shell aliases and a session-archive() function, and fully fleshed-out templates with realistic examples — matching the 'fully executable, specific examples' anchor.	3 / 3
Workflow Clarity	Multi-step processes are explicitly sequenced with checklists: Tier 1/2/3 each pair a trigger with an action list, plus Session Start/End protocols, a decision heuristic, and periodic self-audit questions, satisfying the 'clear sequence with checklists' anchor. The operations are not destructive/batch so missing validate-fix-retry loops do not cap the score.	3 / 3
Progressive Disclosure	There are no bundle files and the entire skill is one monolithic SKILL.md; the well-organized section headers provide structure, but large reusable templates that belong in separate reference files are inlined, matching the 'some structure but content that should be separate is inline' anchor.	2 / 3
	Total	9 / 12 Passed

Description

35%

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 field is a terse, abstract list of three capability areas with no concrete actions or natural trigger language. It states what the skill covers but relies on a separate when-to-use field rather than embedding triggers, leaving it decent but not exemplary.

Suggestions

Rewrite the description as concrete actions, e.g. 'Checkpoints session state to files, summarizes context in tiers, and resumes work from saved state.'

Append an explicit 'Use when...' trigger clause with natural user phrases such as 'starting a long session', 'hitting context limits', or 'resuming work after a break'.

Replace abstract jargon ('tiered summarization', 'context preservation') with terms a user would actually say to keep trigger_term_quality above 1.

Dimension	Reasoning	Score
Specificity	The description names three capability areas ('Context preservation, tiered summarization, resumability') but uses abstract noun phrases rather than listing concrete actions, so it stops at 'names domain and some actions, but not comprehensive' rather than the multi-action level 3.	2 / 3
Completeness	It states the 'what' (preservation, summarization, resumability) but the description field itself omits any 'Use when...' trigger guidance — the when lives in a separate frontmatter field — so per the rubric a missing explicit trigger clause caps completeness at 2.	2 / 3
Trigger Term Quality	The phrases 'Context preservation, tiered summarization, resumability' are technical/abstract jargon a user would not naturally say; there are no natural trigger keywords like 'session', 'checkpoint', or 'resume work' phrased as a user would, matching the 'no natural keywords' anchor.	1 / 3
Distinctiveness Conflict Risk	The phrasing is somewhat specific to session context but the concepts are generic enough to overlap with other memory/context-management skills, matching the 'somewhat specific but could still overlap' anchor.	2 / 3
	Total	7 / 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
skill_md_line_count	SKILL.md is long (529 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: alinaqi/claude-bootstrap
Commit: 9dacc46

Reviewed: about 8 hours 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.