coach

Learning telemetry, strategy, and schedule — retention stats, calibration, grader audit, n-of-1 experiments, HTML dashboard. Use for "how am I doing", weekly check-ins, strategy questions, auditing the grader, or adjusting how Engram teaches.

Quality

85%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

70%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.

The body is highly actionable with excellent workflow sequencing and validation, but it is noticeably verbose and monolithic. Inline version-changelog markers and anecdotal rationale padding hurt token efficiency, and the lack of any bundle/reference files means a long wall of inlined detail rather than a lean overview pointing to deeper materials.

Suggestions

Move the inline version annotations (v0.6–v1.14) and bug-discovery anecdotes into a CHANGELOG or 'design notes' reference file so the active instructions stay lean; the rubric penalizes version numbers not placed in a deprecated/old-patterns section.

Split the long per-subcommand detail (grader-audit branch logic, experiment preset designs, check-in branches) into one-level-deep reference files (e.g. GRADER-AUDIT.md, EXPERIMENTS.md) and keep SKILL.md as a concise overview with clearly signaled links, raising progressive disclosure.

Trim narrative rationale and illustrative quote anecdotes to the minimum needed for the decision rule, retaining the executable commands and branch conditions but cutting the backstory prose.

Dimension	Reasoning	Score
Conciseness	The body is noticeably verbose: pervasive inline version markers (v0.6, v0.7, v0.8, v1.3, v1.4, v1.9, v1.14, etc.) appear throughout the active instructions rather than in a deprecated section, and several padded anecdote/bug-history sections ('Found by a dogfood', 'v0.8.0 led with... a learner who had failed five capabilities') could be trimmed. It is not a 1 because it avoids explaining basic concepts Claude already knows and the core guidance is actionable.	2 / 5
Actionability	The body is fully executable: copy-paste-ready bash blocks resolve the engine, then run concrete commands with exact flags (e.g. 'python3 "$ENGRAM" gold > /tmp/engram-gold.json', 'experiment start --json ...', 'assessor-audit --file ... --canary --grader-context ...') covering every subcommand.	5 / 5
Workflow Clarity	Multi-step workflows are clearly sequenced with explicit validation and feedback loops — the audit's three independent spawns with fail-closed 'assessor-audit dies if answers leak', the 'Four steps, and none may be skipped' contribute flow that degrades to silence, and the experiment pre-register/assign/settle pipeline with powered-check gating.	5 / 5
Progressive Disclosure	Section headers are clear and well-organized by subcommand, and external doc references are signaled inline ('docs/08 §2', 'docs/05 P14', CONTRIBUTING-DATA.md), but the file is a ~350-line monolith with detailed branch logic, experiment presets, and rationale inlined that clearly belongs in separate reference files. The simple-skill exception does not apply given the length and external references.	3 / 5
	Total	15 / 20 Passed

Description

100%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.

The description is exemplary: it pairs a concrete capability list with an explicit 'Use for' trigger clause of natural user phrases, in third-person/imperative voice. It matches the rubric's good-example pattern precisely across all four dimensions.

Dimension	Reasoning	Score
Specificity	Lists five concrete capabilities — 'retention stats, calibration, grader audit, n-of-1 experiments, HTML dashboard' — mapping comprehensively to the skill's subcommands, matching the anchor for multiple specific concrete actions with comprehensive coverage.	5 / 5
Completeness	Explicitly answers both 'what' (the capability list) and 'when' via a concrete 'Use for ...' clause with multiple trigger phrases, exactly matching the anchor example pattern.	5 / 5
Trigger Term Quality	The 'Use for' clause gives five natural trigger phrases a user would actually say, including the verbatim quoted 'how am I doing', 'weekly check-ins', 'strategy questions', 'auditing the grader', and 'adjusting how Engram teaches'.	5 / 5
Distinctiveness Conflict Risk	The Engram-specific capabilities (grader audit, n-of-1 experiments, retention stats) and the named 'Engram teaches' system carve a clear niche with distinct triggers and minimal overlap risk.	5 / 5
	Total	20 / 20 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 1 suspicious	Warning

	Total	14 / 16 Passed

Repository: nagisanzenin/engram
Path: skills/coach/SKILL.md
Commit: cbd56cc

Reviewed: 9 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.