agent-memory-systems

Memory is the cornerstone of intelligent agents. Without it, every interaction starts from zero. This skill covers the architecture of agent memory: short-term (context window), long-term (vector stores), and the cognitive architectures that organize them.

Quality

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Passed

No known issues

Quality

Content

50%

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

The body is information-rich with concrete code and useful decision matrices, but it is hurt by verbatim repetition of the description, code wrapped in '"""' instead of fenced code blocks, undefined helper functions, and a monolithic single-file structure with no progressive disclosure. It lands at the middle anchor on all four dimensions.

Suggestions

Wrap all code in proper markdown fenced blocks (```python ... ```) instead of '"""' delimiters so examples render as executable code, and replace or define helper calls (cluster_by_similarity, generate_key, resolve_conflict, embed) so snippets are copy-paste ready.

Remove the verbatim repetition of the frontmatter description from the opening of the body and trim 'Why this breaks' prose to one line each, trusting Claude's existing knowledge.

Split the large Patterns and Sharp Edges sections into one-level-deep reference files (e.g. references/PATTERNS.md, references/SHARP_EDGES.md) with a concise overview in SKILL.md pointing to them, to improve progressive disclosure.

Dimension	Reasoning	Score
Conciseness	The body provides dense concrete code and decision matrices, but the opening paragraphs repeat the frontmatter description verbatim and explain concepts Claude already knows ('Memory isn't just storage - it's retrieval', the CoALA framework intro), and the 'Why this breaks' prose in Sharp Edges could be tightened.	2 / 3
Actionability	There is extensive concrete code using real library APIs (LangMem, Pinecone, Qdrant, ChromaDB, langchain splitters), but code blocks are wrapped in literal '"""' delimiters instead of markdown code fences so they do not render as executable blocks, and several snippets rely on undefined helpers (cluster_by_similarity, generate_key, resolve_conflict) — so it is not fully copy-paste ready.	2 / 3
Workflow Clarity	Patterns are each labeled with 'When to use' and a 'Validation Checks' section lists concrete error/warning checks for memory operations, but the content is a collection of patterns rather than a sequenced multi-step workflow with explicit validate-fix-retry checkpoints tied to a primary task.	2 / 3
Progressive Disclosure	The ~1089-line body is a single monolithic SKILL.md with no bundle files in references/, scripts/, or assets/; it is organized into clear sections (Principles, Tooling, Patterns, Sharp Edges, Validation Checks) but all detailed content that could live in separate reference files is inline.	2 / 3
	Total	8 / 12 Passed

Description

50%

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 uses appropriate third-person voice and names the domain clearly, but it is padded with conceptual fluff, lacks concrete actions, and provides no explicit 'when to use' trigger guidance. It sits at the middle of the rubric on every dimension.

Suggestions

Replace the 'Memory is the cornerstone... starts from zero' opener with concrete capabilities, e.g. 'Design and implement agent memory systems: short-term context management, long-term vector storage, and retrieval-aware chunking strategies.'

Add an explicit 'Use when...' clause listing natural trigger phrases users would say ('agent memory', 'remember across sessions', 'long-term memory', 'vector store', 'RAG', 'memory retrieval').

Drop jargon like 'cognitive architectures' from the description in favor of terms a user would naturally say.

Dimension	Reasoning	Score
Specificity	The description names the domain and its components ('short-term (context window), long-term (vector stores), and the cognitive architectures that organize them') but lists no concrete actions, and the opening 'Memory is the cornerstone of intelligent agents. Without it, every interaction starts from zero.' is conceptual fluff rather than capability description.	2 / 3
Completeness	It answers 'what' (covers agent memory architecture: short-term, long-term, cognitive architectures) but contains no 'Use when...' clause or equivalent explicit trigger guidance, so per the rubric completeness is capped at 2.	2 / 3
Trigger Term Quality	It includes some relevant keywords a user might say ('agent memory', 'context window', 'vector stores'), but 'cognitive architectures' is jargon and common natural variations like 'remember across sessions', 'memory retrieval', 'RAG', or 'embeddings' are absent.	2 / 3
Distinctiveness Conflict Risk	'Agent memory systems' is a recognizable niche, but the skill itself delegates overlapping concerns (rag-pipeline-architecture, vector-database-operations, embedding-model-selection), so it could still overlap with adjacent architect/engineer skills.	2 / 3
	Total	8 / 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 (1089 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: boisenoise/skills-collections
Commit: f36337d

Reviewed: 2 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.