CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

agent-graphs

Create and manage agent graphs — directed graphs of configs connected by edges with handoff logic. Use when building multi-agent workflows where configs route to each other.

Quality

78%

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

Fix and improve this skill with Tessl

tessl review fix ./skills/agentcontrol/agent-graphs/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

75%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 a well-structured, largely actionable guide: named MCP tools with parameters, a complete JSON example, a sequenced workflow with verification steps, and useful edge-case and anti-pattern tables. Weaknesses are minor — a bit of generic advice padding, an underspecified create-agent-config sub-step, and no error-recovery loop after verification.

DimensionReasoningScore

Conciseness

The body is dominated by tight tables, enumerated tool parameters, and a complete JSON example, with only minor trimmable padding such as the "You're using a skill that will guide you through" preamble and the generic "Core Principles" bullets ("One Agent, One Job", "Design Before Building"). It is not a 5 because those few sections state things Claude already knows, and not a 3 because the padding is minor relative to the ~140 lines.

4 / 5

Actionability

Concrete, executable guidance throughout: named MCP tools with parameter lists ("projectKey", "rootConfigKey", "edges"), a copy-paste-ready JSON payload for create-agent-graph, and specific edge-case actions ("Use update-agent-graph with confirm: true — provide the complete new edge list"). It stops short of 5 because the create-agent-config sub-step is underspecified ("Set up variations with appropriate models and prompts" gives no concrete payload).

4 / 5

Workflow Clarity

A clear four-step sequence (Design → Ensure Nodes Exist → Create → Verify) with verification checkpoints in both Step 2 ("Verify each config exists with get-agent-configs") and Step 4 (confirm structure, edges, root config, handoff data). It is not a 5 because there is no error-recovery feedback loop (e.g., what to do if verification fails), though the destructive delete operation is kept out of the main flow and flagged in Edge Cases, so no cap applies.

4 / 5

Progressive Disclosure

Well-organized single-file skill with clear sections (Prerequisites, Core Concepts, Workflow, Edge Cases, What NOT to Do) and a clearly signaled external link, and no bundle files exist to reference. It is not a 5 because at ~140 lines the MCP tool catalog and scenario tables sit at the edge of what belongs inline and could be split into a reference file.

4 / 5

Total

16

/

20

Passed

Description

70%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 solid description that explicitly covers both what the skill does and when to use it, in third person, with domain-specific trigger terms. Its main weakness is action specificity — 'create and manage' summarizes rather than enumerating capabilities — and a trigger clause that could overlap other multi-agent skills.

Suggestions

Enumerate concrete capabilities in the description (e.g., 'create graphs with edges and handoff data, inspect existing graphs, update routing, verify end-to-end paths') instead of the generic 'manage'.

Sharpen the 'Use when' clause with additional trigger phrases users would naturally say, such as 'routing between agents', 'agent handoffs', or 'wiring configs into a multi-agent pipeline', to improve both trigger coverage and distinctiveness.

DimensionReasoningScore

Specificity

"Create and manage agent graphs" names the domain and two actions, with structural detail ("configs connected by edges with handoff logic"), matching the 'names domain and 1-2 concrete actions' anchor. It is not a 4 because 'manage' is generic and the description does not enumerate several specific operations (update, delete, inspect, verify) the way the level-4 example lists 'extracts text, fills forms, converts pages'.

3 / 5

Completeness

Both parts are explicit: the 'what' ("Create and manage agent graphs — directed graphs of configs connected by edges with handoff logic") and the 'when' ("Use when building multi-agent workflows where configs route to each other"). It is not a 5 because the 'when' clause offers a single trigger scenario rather than multiple concrete trigger phrases as in the level-5 example.

4 / 5

Trigger Term Quality

Natural terms include "agent graphs", "multi-agent workflows", "configs route to each other", and "handoff logic", giving good keyword coverage a user would plausibly say. It falls short of 5 because common synonyms like 'agent orchestration', 'routing between agents', or 'agent chaining' are missing, and it is above 3 since several relevant natural phrases are present rather than just a few.

4 / 5

Distinctiveness Conflict Risk

The LaunchDarkly-specific 'agent graphs' niche with configs, edges, and handoff logic is mostly distinct with clear domain signals. It is not a 5 because the trigger phrase "building multi-agent workflows" is broad enough to overlap with generic multi-agent orchestration skills.

4 / 5

Total

15

/

20

Passed

Validation

100%

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

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

launchdarkly/ai-tooling

Path

skills/agentcontrol/agent-graphs/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.