CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

chaos-dr-test

Use when working with Chaos Engineering steps inside a Harness pipeline. Covers ChaosFault, ChaosProbe, ChaosAction steps (DRTest stages), the Chaos step (DRTest stages), and DRTest stage/pipeline structure. Use when the user asks to add, modify, or create chaos steps, fault injection, probes, or disaster recovery test pipelines. Do not use for standalone Chaos Experiment create or edit; use chaos-experiment for those. Trigger phrases: DR test, DRTest pipeline, disaster recovery test, chaos step, ChaosFault step, ChaosProbe step, ChaosAction step.

Quality

92%

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

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-engineered instruction skill: the body sticks to non-obvious MCP behaviors (auto-derived identifiers, empty-steps create, full-replace PUT semantics, tool-result verification) and delegates step-building mechanics to real, one-level-deep reference files, with explicit validation checkpoints around the destructive update path. Weaknesses are minor: a slightly over-explained YAML conventions paragraph with some repetition between Performance Notes and Troubleshooting, and a scope-establishment reference whose path escapes the skill bundle and is absent from references/.

Suggestions

Tighten the YAML Output Conventions paragraph to a one-line rule (e.g., 'Emit block-style YAML only; the JSON-shaped examples in components.md are authoritative for structure, never for output format') and drop the duplicated full-replace PUT sentence from Performance Notes or Troubleshooting so it lives in one place.

Fix the scope-establishment reference: either copy scope-establishment.md into the skill's own references/ directory and link it relatively (references/scope-establishment.md) or inline the four scope bullets once, since the ../../../ path is absent from the bundle and fragile if the skill is moved.

Reduce the repeated '(pass org_id/project_id using the active scope)' parenthetical by stating once in Scope Rules that all MCP calls in this skill inherit the active scope.

DimensionReasoningScore

Conciseness

The body is dense and procedural throughout — field-by-field scaffold annotations ('auto: <identifier>_pipeline, does NOT change'), the full-replace PUT warning, and the intent keyword table all carry non-obvious Harness-specific knowledge with no 'what is chaos engineering' padding. A few spots could be trimmed: the YAML Output Conventions paragraph over-explains ('appear as minified JSON purely for token efficiency — JSON parses identically to YAML...'), the '(pass org_id/project_id using the active scope...)' parenthetical repeats, and the full-replace PUT point appears in both Performance Notes and Troubleshooting. That matches anchor 4 (efficient, minor over-explanation) rather than anchor 5.

4 / 5

Actionability

The body gives copy-paste-ready MCP call templates with full parameters ('harness_create(resource_type="chaos_dr_test", org_id="<org_id>", project_id="<project_id>", body={"name": "...", "identifier": "..."})'), a complete annotated YAML scaffold, an exact naming-convention derivation, and concrete example utterances routed to files. Specific examples cover the common create/edit/query cases, matching anchor 5.

5 / 5

Workflow Clarity

The flow is clearly sequenced: Step 0 intent detection (with a keyword table and explicit ambiguous fallback question) routes to create.md/edit.md, which hand off to components.md, with a defined save path and loop-back hubs. Despite full-replace PUT being a destructive operation, validation is explicit — 'NEVER claim a tool call succeeded without an explicit success response', 'Wait for the tool's confirmed response before reporting success; a review/elicitation prompt is not a completed save' — with feedback loops (retry on ambiguous response) and error-recovery guidance in Troubleshooting ('always harness_get the current YAML immediately before modifying'). This matches anchor 5 and avoids the destructive-operation cap at 3.

5 / 5

Progressive Disclosure

SKILL.md is a genuine overview that routes to one-level-deep, well-signaled references with purpose statements ('read references/create.md — creates a brand-new DR Test pipeline...'), and all three referenced bundle files (components.md, create.md, edit.md) exist with sensible cross-links. However, the body's first reference — 'Follow references/scope-establishment.md (../../../references/scope-establishment.md)' — points outside the bundle and the file is not present in references/, a dangling path from the bundle's perspective. Good structure with a minor navigation gap matches anchor 4 rather than 5.

4 / 5

Total

18

/

20

Passed

Description

95%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, tightly scoped description: it enumerates all four chaos step types and the DRTest stage/pipeline structure, provides an explicit 'Use when' clause with a comprehensive trigger-phrase list including synonyms, and draws a hard boundary against the chaos-experiment skill. The only weakness is a slightly passive 'Covers...' framing of the core create/edit action, which keeps specificity just below maximum.

DimensionReasoningScore

Specificity

The description names concrete capabilities — 'Covers ChaosFault, ChaosProbe, ChaosAction steps (DRTest stages), the Chaos step (DRTest stages), and DRTest stage/pipeline structure' plus the actions 'add, modify, or create' — with only a minor gap: it never states the top-level action (create/edit DR Test pipelines) as a verb, framing the 'what' as coverage rather than actions. This sits between anchor 4 (several specific actions, minor gaps) and anchor 5; the minor action-framing gap keeps it at 4 rather than 5.

4 / 5

Completeness

It explicitly answers 'what' ('Covers ChaosFault, ChaosProbe, ChaosAction steps... and DRTest stage/pipeline structure') and 'when' with concrete trigger phrases ('Use when the user asks to add, modify, or create chaos steps, fault injection, probes, or disaster recovery test pipelines'), plus a negative boundary. This matches anchor 5 exactly; anchor 4 would require the 'when' to be less explicit, which it is not.

5 / 5

Trigger Term Quality

Trigger phrases are explicitly enumerated: 'DR test, DRTest pipeline, disaster recovery test, chaos step, ChaosFault step, ChaosProbe step, ChaosAction step', plus 'fault injection, probes' in the when-clause. Synonym coverage is comprehensive (DR test / DRTest / disaster recovery test), matching anchor 5; natural terms a Harness user would say are all present.

5 / 5

Distinctiveness Conflict Risk

The niche is narrow (chaos steps inside Harness DRTest pipeline stages) and it actively disambiguates from the closest sibling: 'Do not use for standalone Chaos Experiment create or edit; use chaos-experiment for those.' Triggers like 'ChaosFault step' and 'DRTest pipeline' are distinct from generic chaos-engineering vocabulary, matching anchor 5 with minimal conflict risk.

5 / 5

Total

19

/

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

relative_links

Relative link issues: 1 suspicious

Warning

referenced_paths_exist

Referenced path issues: 1 missing

Warning

Total

14

/

16

Passed

Repository

harness/harness-ai

Path

plugins/claude/skills/chaos-dr-test/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.