CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

memory-system

Debug, maintain, and use the joelclaw memory system — observer capture/dreams, Brain .svx stores, neat-memory curation, Telegram delivery, reaction grading, wiki rendering. Load whenever work touches the memory system, the curator, dreams, observations, grading, or a memory surface misbehaves.

Quality

86%

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

82%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, highly actionable operational reference: every debug recipe is symptom-keyed with concrete commands and validation gates, and structure is easy to navigate. Weaknesses are minor — some time-sensitive dates sit outside the retirement sections, the retirement note is stated twice, and recipes rarely close the loop with an explicit recovery path after a failed check.

Suggestions

Consolidate the top 'Memory retirement (2026-07-17)' blockquote into the 'Retired and parked surfaces' section to remove the duplicated retirement story.

Move the inline date annotations ('Since the 2026-07-17 fix', 'taste bench (2026-07-17)') into the retirement/deprecated sections so time-sensitive detail lives in one place.

Add a one-line fix-and-retry step to recipes that currently end at verification (e.g., what to do when the delivery canary exits non-zero or the journal row is missing).

DimensionReasoningScore

Conciseness

Prose is telegraphic and fact-dense (paths, event names, commands) with zero padding or explanation of known concepts, but inline dates ('Since the 2026-07-17 fix', 'taste bench (2026-07-17)') sit outside the retirement/deprecated sections and the top blockquote duplicates the 'Retired and parked surfaces' section, leaving minor trims. Anchor 4 fits better than 5.

4 / 5

Actionability

Commands are copy-paste ready and cover the common cases: 'jq --arg slug "<slug>" '.sent[] | select(.slug == $slug) | {slug, outcome}' ~/.joelclaw/observer-neat-memories.json', 'bun scripts/backfill-session-index.ts' with dry-run then --apply, 'bun scripts/messaging-visible-delivery-canary.ts' with an exit-0 success criterion, plus concrete emoji-to-outcome mappings. Fully executable guidance matching anchor 5.

5 / 5

Workflow Clarity

Recipes are ordered symptom-to-fix with validation gates on risky operations (--dry-run promotion, dry-run backfill, 200-check before sharing, exit-0 canary) and a forked decision path ('Chain alive → ...; Chain dead → ...'). However, most recipes stop at verification without an explicit fix-and-retry loop if the check fails, matching anchor 4 rather than 5.

4 / 5

Progressive Disclosure

No bundle files exist, so this is scored on body structure: clear section headers (location table, debug recipes, usage, retired surfaces, doctor) with one-level-deep external pointers clearly signaled ('read it before deep debugging'). The location table and behavior-card detail could arguably live in separate reference files, so anchor 4's 'minor organization gaps' fits better than 5.

4 / 5

Total

17

/

20

Passed

Description

87%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: it names a concrete domain, enumerates six subsystem capabilities, and includes an explicit 'Load whenever...' trigger clause covering natural phrasings. Minor gaps only — a few natural synonyms (wiki, recall, Telegram) absent from the trigger list and broad top-level verbs.

Suggestions

Add natural trigger synonyms like 'wiki', 'recall', 'Brain', or 'Telegram DMs' to the when-clause so users phrasing around those surfaces also match.

Swap the generic 'use' verb for more concrete actions (e.g., 'write and query memory') to sharpen the what-clause.

DimensionReasoningScore

Specificity

"Debug, maintain, and use" are broad verbs, but the description enumerates six concrete subsystems ("observer capture/dreams, Brain .svx stores, neat-memory curation, Telegram delivery, reaction grading, wiki rendering"). It lists several specific capabilities but doesn't break out the action per subsystem, matching anchor 4 rather than 5.

4 / 5

Completeness

Both halves are explicit: a clear what ("Debug, maintain, and use the joelclaw memory system — ...") and an explicit trigger clause ("Load whenever work touches the memory system, the curator, dreams, observations, grading, or a memory surface misbehaves") with concrete trigger phrases, matching the anchor-5 example structure.

5 / 5

Trigger Term Quality

The when-clause carries natural phrases ("memory system, the curator, dreams, observations, grading, or a memory surface misbehaves") a user would plausibly say. A few natural terms from the what-half (wiki, recall, Telegram/DM, Brain) are missing from the trigger list, so anchor 4 fits better than 5.

4 / 5

Distinctiveness Conflict Risk

The named "joelclaw memory system" plus niche triggers (curator, dreams, neat-memory) form a clear proprietary niche with minimal chance of firing for an unrelated skill, matching anchor 5.

5 / 5

Total

18

/

20

Passed

Validation

87%

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

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

referenced_paths_exist

Referenced path issues: 3 missing

Warning

Total

14

/

16

Passed

Repository

joelhooks/joelclaw

Path

skills/memory-system/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.