build-project

Hands-on project tutor for the AI Engineering from Scratch Projects section. Guides a learner through one stage of a real project per session: read the stage lesson, predict, write the code, run the stage grader, reflect, and record progress in PROJECTS-LEARNING.md. Gives hints, never full solutions. Trigger phrases: "build a project", "next project stage", "continue my project", "start the research report agent".

Quality

92%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

88%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 well-built workflow skill: fully actionable commands, a grader-driven validation loop with hint escalation, and clean delegation of teaching content to external lesson files. Weaker points are minor redundancy in the Rules section and non-workflow content (host contract table, note template) that could live elsewhere or be trimmed.

Suggestions

Trim the Rules section to only non-redundant rules (e.g. keep 'Never claim a pass you did not see in grader output' but drop restatements of the one-stage-per-invocation and learner-types-the-code rules already covered in the intro and Build step).

Move the host invocation contract table into a small reference file (or the description's trigger guidance) so SKILL.md opens with the tutoring workflow itself.

Consider moving the PROJECTS-LEARNING.md note template to a references/ file (e.g. references/progress-template.md) to keep the body focused on the teach-run-reflect loop.

Dimension	Reasoning	Score
Conciseness	The body is lean, procedural, and free of concept explanations Claude already knows — commands, repo layout, note format, and a 3-level hint policy are all non-obvious. However, the Rules section restates earlier content: 'One stage per invocation' repeats the intro, and 'The learner types the code. You never paste a full stage solution' repeats the Build and Debug steps. This minor trimmable redundancy fits the 4 anchor rather than the 5 anchor where every token earns its place.	4 / 5
Actionability	Gives copy-paste-ready commands ('python3 scripts/project_test.py <project-id> --init <workspace>', '--stage <N> --path <workspace>', '--strict --report completion.json'), exact per-stage file paths, a complete PROJECTS-LEARNING.md template, and a concrete conceptual-mode fallback. Placeholders are inherently project-specific and are resolved by reading project.json, which the skill instructs — fully executable guidance for an instruction-only skill.	5 / 5
Workflow Clarity	A clear Step 0→1→2→3 sequence with an explicit validation loop: run the grader, on failure read the test name and escalate hints (question → concept → specific line), re-run; regressions are surfaced because 'the grader runs stages 1 to N'. Checkpoints are explicit ('Never claim a pass you did not see in grader output', conceptual mode marks results 'Pending', never 'Pass'). Matches the 5 anchor with feedback loops and error recovery.	5 / 5
Progressive Disclosure	No bundle files exist, and teaching content is correctly delegated one level deep to lesson docs, project.json, and projects/SUBMITTING.md rather than inlined, with well-organized headers throughout. However, the host-invocation contract table and the inline note-format template are non-workflow content living in the main file — minor organization gaps that fit the 4 anchor rather than the 5 anchor's fully appropriate placement.	4 / 5
	Total	18 / 20 Passed

Description

96%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: specific, third-person, comprehensive actions with an explicit trigger-phrase clause that answers both what and when. The only weakness is that two of the four trigger phrases ('build a project', 'continue my project') are generic enough to risk firing outside the intended course context.

Suggestions

Qualify the generic trigger phrases, e.g. change "build a project" to "build a project" within a phrase like "build an AI Engineering from Scratch project", so the skill cannot fire on unrelated project-building requests.

Drop or narrow "continue my project" to "continue my course project" to further reduce conflict with general project workflows.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions ('read the stage lesson, predict, write the code, run the stage grader, reflect, and record progress in PROJECTS-LEARNING.md. Gives hints, never full solutions') with a named artifact, in consistent third-person voice ('Guides a learner', 'Gives hints'). Every capability is specific and concrete, matching the comprehensive-coverage anchor rather than the 4 anchor's 'minor gaps'.	5 / 5
Completeness	Clearly answers both 'what' (tutor guiding one stage per session through the listed actions) and 'when' via an explicit 'Trigger phrases:' clause with concrete phrases. This matches the 5 anchor where both what and when are explicit with concrete triggers, not the 4 anchor where the 'when' is merely present.	5 / 5
Trigger Term Quality	Explicit natural trigger phrases a learner would actually type: "build a project", "next project stage", "continue my project", "start the research report agent". These cover start, resume, and named-project variations, matching the comprehensive-synonyms anchor; only the 4 anchor has 'a few natural terms missing', which is not the case here.	5 / 5
Distinctiveness Conflict Risk	'AI Engineering from Scratch Projects section' and 'start the research report agent' carve a distinct niche, but the generic triggers 'build a project' and 'continue my project' could fire on ordinary project-building requests unrelated to the course. Minor overlap risk with closely related coding-tutor skills matches the 4 anchor; not 5 because those generic phrases lack a course-scoping qualifier.	4 / 5
	Total	19 / 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
referenced_paths_exist	Referenced path issues: 2 missing	Warning

	Total	14 / 16 Passed

Repository: rohitg00/ai-engineering-from-scratch
Path: .claude/skills/build-project/SKILL.md
Commit: c02ca08

Reviewed: 4 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.