bound

The user cannot yet see what a task needs them to decide, or which decisions to keep or entrust: map the whole task first, then open each decision to the depth needed.

Quality

50%

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

Fix and improve this skill with Tessl

tessl review fix ./horismos/skills/bound/SKILL.md

Quality

Content

23%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 dominated by an elaborate Lean 4 pseudo-formalization that buries a modest set of actionable protocol rules under hundreds of lines of abstract, self-referential contract prose. A reader can eventually reconstruct the intended interaction pattern, but at enormous token cost and with little concrete, executable guidance.

Suggestions

Move the entire Lean 4 formalization block into a reference file (e.g., references/formal-model.md), keeping SKILL.md to a concise statement of the protocol steps and linking to it for the formal contract.

Replace abstract directives with concrete, imperative instructions and at least one worked example — e.g., a sample first-round map with two or three decision entries, their proposals, and the acceptance prompt.

Express the run state machine (awaiting disposition / turn owed / boundary stands / withdrawn) as a short plain-language decision list or small table instead of nested qualifying clauses, making checkpoints explicit.

Dimension	Reasoning	Score
Conciseness	Roughly 600 of the ~655 lines are a Lean 4 formalization that restates ordinary interaction rules in dense notation and axiom-comment prose, extensively padding what could be expressed in a small fraction of the tokens — a severe token-budget violation even though nothing in it teaches the model something it doesn't already know.	1 / 5
Actionability	The Protocol and Rules sections give high-level directives ("show the concrete content, assumptions, alternatives, and dependent consequences"), but there are no concrete commands, templates, or worked examples; the Lean block uses axioms as judgment placeholders and is not copy-paste-executable guidance for real work.	2 / 5
Workflow Clarity	A sequence is discernible and consistently mirrored across FLOW, Phase Transitions, Protocol, and Rules (observe → present map → rounds → converge/withdraw), but checkpoints are implicit and the layered, self-referential conditions (e.g., the boundary standing 'only where no turn of the user's is still owed') obscure the actual execution order.	3 / 5
Progressive Disclosure	The single reference (references/round-composition.md) is real, one level deep, and clearly signaled, but the massive Lean formalization — content that clearly belongs in a separate reference file — is inlined monolithically in SKILL.md.	2 / 5
	Total	8 / 20 Passed

Description

53%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 communicates a coherent, fairly distinct purpose around surfacing and resolving open decisions, but it reads as a condition statement rather than a capability-plus-trigger description. It lacks explicit 'Use when...' trigger guidance and natural user phrasing, capping both completeness and trigger quality.

Suggestions

Add an explicit trigger clause, e.g. 'Use when the user starts a task whose scope, goal, or decision ownership is unclear, or asks what they need to decide or approve.'

Include natural user phrasings and synonyms — 'what decisions do I need to make', 'scope', 'delegate', 'approval' — so the description matches how users actually ask for this.

Rewrite as a third-person capability statement ('Maps the decision structure of a task...') instead of leading with a description of the user's deficient state.

Dimension	Reasoning	Score
Specificity	The description names its domain (decision structure of a task) and two actions — "map the whole task first" and "open each decision to the depth needed" — but these are abstract directives rather than a comprehensive list of concrete capabilities, matching the anchor for 1-2 concrete actions without full coverage.	3 / 5
Completeness	The 'what' is clear (map the whole task, open each decision to the needed depth), but the 'when' is only weakly implied by the opening condition clause; there is no 'Use when...' or equivalent explicit trigger guidance, which caps this dimension at 3.	3 / 5
Trigger Term Quality	Terms like "decide", "which decisions to keep or entrust", and "map the whole task" are relevant, but the framing "The user cannot yet see what a task needs them to decide" is a condition statement rather than natural user phrasing, and common variations (e.g., "what decisions do I need to make", "scope", "delegate") are missing — some keywords, missing synonyms.	3 / 5
Distinctiveness Conflict Risk	The niche — mapping which decisions a user should make or entrust before proceeding — is fairly distinct and unlikely to trigger for file-manipulation or extraction skills, with only minor overlap risk against general planning/scoping skills.	4 / 5
	Total	13 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (659 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: jongwony/epistemic-protocols
Path: horismos/skills/bound/SKILL.md
Commit: 10abcb3

Reviewed: 5 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.