forge

Reference-grounded prompt-artifact formation. Reads a target reference doc, surfaces the under-determined contract coordinates, and projects a ready-to-use prompt or standing skill recipe.

Quality

51%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./epistemic-cooperative/skills/forge/SKILL.md

Quality

Content

52%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 presents a well-sequenced, validation-aware multi-phase workflow with strong progressive-disclosure intent and concrete contract/type definitions, but it is verbose and jargon-heavy, delegates its central execution step to another skill, and references adapter files that are not actually present in the bundle.

Suggestions

Eliminate the repeated role-boundary restatements (intro, Rule 10, Boundary Note all restate 'does not run the tool/branch/PR') and trim abstract architectural justifications such as 'deferred-colimit accumulation mechanism made physical'.

Ship the four referenced adapter files (adapters/higgsfield.md, gpt-image.md, codex-goals.md, claude-session.md) or, if they are not yet realized, mark them as candidate/not-yet-shipped so the progressive-disclosure references are fulfilled.

Add concrete, in-skill mechanics for the reverse-induction step (what 'inline lightweight reverse-induction' actually does) instead of leaving the core action as an abstract delegation to /elicit.

Dimension	Reasoning	Score
Conciseness	The body is noticeably verbose: the role boundary ('does not run the downstream tool, open branches, or create PRs') is restated in the intro, Rule 10, and the Boundary Note, and abstract architectural justifications ('deferred-colimit accumulation mechanism made physical', 're-imports the routable/provenance boundary') pad several sections without adding executable value.	2 / 5
Actionability	Concrete structures are provided (the seam-operations table, the Types table with field-level definitions, the numbered Phases, and the operational checklist), but the core execution step ('reverse-induce the under-determined intent into ResolvedIntentIR') is abstract and delegated to /elicit, and the adapter files that hold the real projection logic are referenced but not shipped.	3 / 5
Workflow Clarity	Phases 0–4 are clearly sequenced with explicit validation checkpoints (the staleness guard with metadata fields and curated-seed fallback, the validate step against capabilities, the stale-guide/model-fallback/transport-unsafe flags) and a feedback gate (Accept / Adjust flagged slots → re-project / Regenerate → rebuild), plus a per-cycle checklist; held below 5 because the in-skill mechanics of the central reverse-induction step are delegated rather than specified.	4 / 5
Progressive Disclosure	The disclosure design is sound — an always-loaded Adapter Index (name + reference + PromptArtifact form) with one-level-deep adapter bodies loaded only after selection — but the referenced adapters/<name>.md files (higgsfield, gpt-image, codex-goals, claude-session, said to 'ship now') are absent from the bundle, so the deferral is designed but unfulfilled.	4 / 5
	Total	13 / 20 Passed

Description

51%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 states a clear, distinct purpose with several named actions but is written in heavy architectural jargon and entirely lacks an explicit 'when to use' trigger clause. Trigger-term quality and completeness are the main weaknesses.

Suggestions

Add an explicit 'Use when...' clause naming natural triggers, e.g. 'Use when drafting a prompt or skill recipe that must be grounded in an authoritative reference doc (vendor model guide, Codex Goals spec, best-practices guide).'

Replace jargon like 'under-determined contract coordinates' and 'prompt-artifact formation' with plain user-facing phrasing such as 'fills in the missing details' and 'drafts a prompt'.

Include common synonyms and concrete triggers (e.g. 'draft a prompt', 'ground a prompt in a reference', '/goal string') to improve natural keyword coverage.

Dimension	Reasoning	Score
Specificity	Lists several concrete actions — 'Reads a target reference doc', 'surfaces the under-determined contract coordinates', 'projects a ready-to-use prompt or standing skill recipe' — though the middle action leans on abstract jargon, keeping it just below a 5.	4 / 5
Completeness	The 'what' is stated (reference-grounded prompt-artifact formation with the three actions), but there is no 'Use when...' clause or equivalent trigger guidance, which caps completeness at 3 per the rubric.	3 / 5
Trigger Term Quality	Only 'prompt' and 'skill recipe' are semi-natural user terms; the rest ('reference-grounded prompt-artifact formation', 'under-determined contract coordinates') is technical jargon, and the natural phrases a user would actually say are absent.	2 / 5
Distinctiveness Conflict Risk	It carves a fairly distinct niche (reference-grounded prompt-artifact formation) with low overlap risk against other skills, though the jargon-heavy phrasing weakens how reliably it triggers for the right need.	4 / 5
	Total	13 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: jongwony/epistemic-protocols
Path: epistemic-cooperative/skills/forge/SKILL.md
Commit: 3ffe94c

Reviewed: 1 day 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.