CtrlK
BlogDocsLog inGet started
Tessl Logo

harness-init

One-time, guided setup of the local agent-first coding harness — the polling loop, the deterministic dispatcher, its config, the AGENTS.md contract, and the worktree provisioning that makes per-feature worktrees runnable. Use when a developer wants to set up, bootstrap, install, or initialize the coding harness for the first time in a project.

71

Quality

89%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

The canonical home for this skill is harness-init in tdg-ninja/context-specs-factory-ai

SKILL.md
Quality
Evals
Security

Quality

Content

92%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.

This is a well-engineered instruction skill: an 11-step guided flow with concrete commands, explicit confirm/validate/fix loops at every risky point, and clean one-level-deep progressive disclosure into a verified bundle of references and assets. The only slack is minor deliberate repetition of invariant rules, which costs a little token efficiency without hurting clarity.

DimensionReasoningScore

Conciseness

The body is dense and assumes Claude's competence throughout — no padding or explanation of known concepts — but a few passages repeat load-bearing rules already stated elsewhere (e.g. Invariant 6 in Preconditions, the Step 2 wrapper rationale, and Hard rules), and some rationale sentences could be trimmed.

4 / 5

Actionability

Every step carries executable specifics: `git worktree add --detach ../<repo>-harness origin/main`, the exact `.gitignore` block, per-file copy mappings with the chmod/sourced distinction, `gh release view --repo anthropics/claude-code-action`, `/loop 5m /poll-and-dispatch`, and `REVIEW_CLEAN_MARKER=` — covering the happy path, re-run path, and STUCK path with no real gaps.

5 / 5

Workflow Clarity

Steps 0-10 are clearly sequenced with explicit validation checkpoints (preflight [MISS] gating, the host-worktree provisioning doubling as bootstrap validation with fix-and-re-run, the Step 9 dry-run with debug-before-proceeding), feedback loops (two-strike retry, re-run diffing instead of clobbering), and a hard-rules checklist; the sensitive secret-copying operation has explicit consent and direction checks, so no validation cap applies.

5 / 5

Progressive Disclosure

The body is a guided-flow overview where each step signals exactly which reference to read first, references are one level deep, detail is appropriately split (canonical assets dropped in verbatim, Software 3.0 files generated via reference guidance), and every referenced path (9 references, all assets including workflows/claude-review.yml, scripts/preflight.sh) verified to exist in the bundle.

5 / 5

Total

19

/

20

Passed

Description

83%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 the niche precisely, enumerates the concrete artifacts being set up, and pairs an explicit what with an explicit, verb-rich when clause. Its only weakness is that specificity rests on enumerated components rather than distinct concrete actions, and trigger coverage could include more colloquial variants.

DimensionReasoningScore

Specificity

"guided setup of the polling loop, the deterministic dispatcher, its config, the AGENTS.md contract, and the worktree provisioning" enumerates several concrete deliverables, but they are objects of a single verb ("setup") rather than multiple distinct concrete actions, so it falls between the 1-2-action anchor and comprehensive-action anchor.

4 / 5

Completeness

It explicitly answers both questions: what ("One-time, guided setup of the local agent-first coding harness" with enumerated components) and when ("Use when a developer wants to set up, bootstrap, install, or initialize the coding harness for the first time in a project"), matching the top anchor's structure with concrete trigger phrases.

5 / 5

Trigger Term Quality

"Use when a developer wants to set up, bootstrap, install, or initialize the coding harness" covers the natural user verbs well (set up/bootstrap/install/initialize) plus "for the first time in a project", but misses common colloquial variants like "get started with" or "onboard the harness".

4 / 5

Distinctiveness Conflict Risk

"local agent-first coding harness" plus "for the first time" carves a clear niche with minimal overlap risk, but the generic trigger verbs ("install", "set up") leave minor conflict risk with closely related harness or loop-setup skills.

4 / 5

Total

17

/

20

Passed

Validation

93%

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

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

referenced_paths_exist

Referenced path issues: 6 missing, 1 deeper-than-1-level

Warning

Total

15

/

16

Passed

Repository
tdg-ninja/context-specs-claude-code
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.