CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

maintaining-core-memory

Maintains an agent's core memory, the editable part of its system prompt. Use to modify the agent's core behavior, embed critical knowledge (needed on every turn), or instruct the agent on how to use deferred memory.

Quality

93%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

90%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 dense, high-quality instruction skill: concrete templates and commands, tight prose with no padding, and clear section structure with validation checks woven in. The main gaps are the lack of a single end-to-end workflow with a post-commit verification step, and some detail (prompt-format spec, commit conventions) that could be split into reference files.

DimensionReasoningScore

Conciseness

The body is lean and principle-driven — every line states a rule or its reason ("A healthy file becomes more general over time; a growing count of narrow rules is a sign of failure") and nothing explains concepts Claude already knows (no git/markdown/frontmatter tutorials), with no time-sensitive details. It is not a 4 because no genuine over-explanation could be trimmed without losing guidance.

5 / 5

Actionability

Fully concrete for an instruction-only skill: an exact compiled-prompt template, a frontmatter example, specific config names and defaults (".memfs.config.json", "maxCoreMemoryCharacters" default 65,536), and a copy-paste-ready "git add <changed files> && git commit -m" template with the full message format specified. Per the rubric's code-vs-instruction note, the absence of code is not penalized since the guidance is executable as written.

5 / 5

Workflow Clarity

Sequences and checkpoints are present — the 3-step "Shrink before you grow" list, "finish and commit one before editing for the next", contradiction checks ("Check that a change does not contradict an existing line"), a reachability check after edits, and a feedback loop ("A memory edit is a hypothesis…; later conversations test it"). It is not a 5 because there is no single end-to-end edit→verify→commit sequence and no explicit post-commit verification step.

4 / 5

Progressive Disclosure

No bundle files exist, and the ~120-line body is well organized into clearly headed sections with all guidance navigable in one pass. It is not a 5 because the skill exceeds the ~50-line simple-skill threshold and borderline detail (the compiled prompt-format spec and the commit-message conventions) could live in a separate reference file.

4 / 5

Total

18

/

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: third-person, concise, with a clear what and an explicit 'Use to…' clause listing three concrete triggers. Domain trigger terms are well covered, though a few natural user phrasings (e.g., 'remember this', 'update memory') are absent.

DimensionReasoningScore

Specificity

The description lists three concrete actions with objects — "modify the agent's core behavior", "embed critical knowledge (needed on every turn)", "instruct the agent on how to use deferred memory" — comprehensively covering the skill's capabilities. It is not a 4 because there are no meaningful coverage gaps; it uses third-person voice ("Maintains"), so no voice penalty applies.

5 / 5

Completeness

It explicitly answers both questions: the what ("Maintains an agent's core memory, the editable part of its system prompt") and a concrete when ("Use to modify the agent's core behavior, embed critical knowledge (needed on every turn), or instruct the agent on how to use deferred memory"). The explicit 'Use to…' clause with three specific triggers matches the top anchor, so the missing-trigger cap does not apply.

5 / 5

Trigger Term Quality

Good keyword coverage of the domain's own vocabulary — "core memory", "system prompt", "core behavior", "deferred memory" — which a user of this system would naturally say. It is not a 5 because common user phrasings like "remember this", "update memory", or "MEMORY.md" are missing, but it clearly exceeds anchor 3's 'some relevant keywords'.

4 / 5

Distinctiveness Conflict Risk

A clear niche — editing an agent's compiled system prompt / core memory and its deferred-memory pointers — with distinct trigger terms unlikely to fire for unrelated skills. Conflict risk with generic note-taking or memory skills is minimal.

5 / 5

Total

19

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

letta-ai/letta-code

Path

src/skills/builtin/maintaining-core-memory/SKILL.md

Reviewed

Table of Contents

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.