CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

bound

Use when what a task needs decided is not yet visible — which decisions it turns on, what to keep or entrust, and with whom — before work proceeds on that undrawn boundary; not where the direction a source gives already settles it, or nothing would await the user. The user can then see each decision, what rests on it, its cost if wrong, what settled it, and what stays open and with whom.

Quality

56%

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 ./horismos/skills/bound/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

35%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 embeds a formal Lean specification and repeatedly restated governance rules in a monolithic 43KB document, burying a workable round-by-round protocol in redundancy. A reader cannot extract the workflow or produce a concrete round without substantial interpretation; the one reference file is well-signaled but underused.

Suggestions

Move the Lean specification to a file under `references/` and state each governance rule exactly once in SKILL.md — the acceptance and owed-turn rules currently appear six or more times across Definition, FLOW, LOOP, CONVERGENCE, doc comments, Protocol, and Rules.

Add one worked example of a first round (a small task with 3-4 map entries showing relevance, cost-if-wrong, proposals, and the acceptance path) so the abstract round-composition rules have a concrete referent.

Replace the FLOW/LOOP/CONVERGENCE prose commentaries with a single numbered step list that consolidates the standing conditions ("no turn owed", "nothing awaits") as explicit checkpoints.

DimensionReasoningScore

Conciseness

The body is a 43KB monolith in which the same governance rules are heavily padded across many restatements — the acceptance/owed-turn rules appear in "The boundary stands in one of two ways…", the FLOW block, the LOOP commentary ("Neither scan exhaustion nor a visit count constitutes the person's acceptance"), the `AcceptanceSupported` and `owed` doc comments, Protocol bullets, and Rules ("scan exhaustion or a visit count supplies no acceptance"). This is pervasive padding, not the 'several unnecessary sections' of anchor 2, though it is original rather than known-concept explanation.

1 / 5

Actionability

The Protocol section gives round-by-round procedural guidance ("At the first round, show the relevant whole draft before asking the user to choose its applicable parts or examination depth"), but no templates, worked examples, or concrete output formats exist, and the ~550-line Lean block is a formal specification whose axioms are explicitly declared judgments ("a black box to the contract, yours to make"), not executable instructions. Above 2 because the Protocol bullets do give specific sequencing; below 4 because nothing shows what a round or map actually looks like.

3 / 5

Workflow Clarity

A real sequence is present across FLOW and Protocol (first round → present the whole map → rounds serving opened axes and corrections → acceptance or `converge`), and standing conditions ("no turn of the user's is owed", "nothing awaits the person's disposition") function as checkpoints. Below 4 because these steps and checkpoints are scattered through dense prose with no consolidated step list — a reader must parse the entire document to extract the workflow.

3 / 5

Progressive Disclosure

The single bundle reference is real, clearly signaled, and one level deep ("read `references/round-composition.md` before presenting it"), and the body has section headers. Below 4 because the ~550-line Lean specification is inlined wholesale in SKILL.md — content that clearly belongs in a separate reference file — which is a major placement issue rather than a minor organization gap.

3 / 5

Total

10

/

20

Passed

Description

57%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 has explicit and even negative trigger guidance, but is written in dense, idiosyncratic prose that names outcomes rather than the concrete actions the skill performs. It is distinguishable from other skills, yet reads more as philosophy than as a description a user's request would naturally match.

Suggestions

State 2-3 concrete actions the skill performs (e.g., "constructs a whole map of the task's open decisions, obligations, and dependencies, and shows each with its evidence, cost if wrong, and who settles it") instead of describing the abstract pre-boundary state.

Include natural trigger phrasings users would actually say — "what decisions need to be made", "unclear scope", "what to delegate" — alongside the current framing so the description matches real requests.

Collapse the double-embedded negative conditions ("not where the direction a source gives already settles it, or nothing would await the user") into one short exclusion clause; the current nesting obscures the trigger.

DimensionReasoningScore

Specificity

The description states outcomes rather than actions the skill performs — "The user can then see each decision, what rests on it, its cost if wrong, what settled it, and what stays open" — and its framing phrases ("what a task needs decided is not yet visible", "that undrawn boundary") are abstract, naming the domain without concrete actions. It is above 1 because the domain and deliverables are identifiable, but below 3 because no concrete action verbs (map, construct, show) are ever stated.

2 / 5

Completeness

The 'when' is explicit and unusually specific, including negative conditions ("not where the direction a source gives already settles it, or nothing would await the user"), and the 'what' is present through the outcome sentence; the 'Use when' clause is present so no cap applies. Below 5 because the 'what' is outcome-stated rather than action-stated and lacks concrete trigger phrasing of what the skill does.

4 / 5

Trigger Term Quality

Some relevant keywords are present ("decisions", "what to keep or entrust", "with whom", "what stays open"), but they are composed idiosyncratically — a user would say "what decisions need to be made", "unclear scope", or "what should I delegate", not "what a task needs decided is not yet visible". Below 4 because common natural phrasings and synonyms are missing; above 2 because more than one or two relevant keywords appear.

3 / 5

Distinctiveness Conflict Risk

The niche — pre-work decision-boundary definition — plus explicit negative triggers ("not where the direction a source gives already settles it") make false activation unlikely, and phrases like "entrust" and "with whom" are distinctive. Below 5 because the decisions/delegation territory still overlaps with general planning and scoping skills.

4 / 5

Total

13

/

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

skill_md_line_count

SKILL.md is long (605 lines); consider splitting into references/ and linking

Warning

Total

15

/

16

Passed

Repository

jongwony/epistemic-protocols

Path

horismos/skills/bound/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.