CtrlK
BlogDocsLog inGet started
Tessl Logo

hermes-agent

Expert in building self-improving AI agents with tool use, multi-platform messaging, and a closed learning loop. Proficient in LLM orchestration, tool integration, session management, and agent autonomy.

44

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 ./.agents/skills/hermes-agent/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

40%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 rich in accurate, project-specific detail with genuinely actionable commands, tables, and pitfalls, but it is a monolithic A-Z encyclopedia rather than a skill overview. It is padded with a table of contents, a duplicate feature summary, and dated release history, has no sequenced workflow with validation checkpoints, and pushes all detail inline instead of splitting it into referenced files.

Suggestions

Split the body into one-level-deep reference files (e.g. RELEASES.md for the changelog, CONFIG.md for the configuration reference, ARCHITECTURE.md for sections 5-9) and keep SKILL.md as a concise overview with clearly signaled links.

Delete the 55-line Table of Contents and the Key Features Summary table that duplicates the Project Overview; move the dated Release History out of the always-loaded body since time-sensitive version/date detail does not belong there.

Add explicit validation checkpoints to the operational sequences (e.g. after install: verify `hermes --version` and API-key connectivity before proceeding; for plugin/skill authoring: run the test suite with `_isolate_hermes_home` before shipping).

DimensionReasoningScore

Conciseness

The body is 1765 lines with a ~55-line table of contents, a "Key Features Summary" table that largely duplicates the Project Overview bullets, and a six-version Release History changelog full of specific dates and version numbers ("v0.2.0 (March 12, 2026)") that is time-sensitive and not placed in a deprecated/old-patterns section. This matches anchor 2 (noticeably verbose, several padded sections) — not 1 because the material is project-specific knowledge Claude would not already know, and not 3 because the padding (TOC, changelog, duplicate summary) is substantial rather than incidental.

2 / 5

Actionability

There is a large amount of concrete, executable material: copy-paste install commands ("curl -fsSL https://raw.githubusercontent.com/... | bash", "hermes model"), exact file paths, environment-variable and config-file tables, a plugin interface code block, and highly specific pitfalls ("DO NOT hardcode `~/.hermes` paths — Use `get_hermes_home()`"). It falls short of anchor 5 because some code is illustrative or truncated ("# ... plus provider, routing, callback params", a simplified agent loop, and interface stubs like `def store(key, value)` without bodies), leaving minor gaps.

4 / 5

Workflow Clarity

Apart from the Installation section's rough command sequence, the document is 26 topic-ordered encyclopedia sections rather than a sequenced workflow, and there are no validation or verification checkpoints anywhere for operations that are destructive or batch-capable (terminal execution, file manipulation, supply-chain-sensitive installs). This fits anchor 2 (rough sequence present but many gaps, validation absent) — the install steps keep it above 1, but the absence of any validate-and-retry loop and the non-procedural organization keep it well below 3.

2 / 5

Progressive Disclosure

No bundle files exist (references/, scripts/, assets/ are absent) and the body inlines content that clearly belongs in separate files — a full changelog, a configuration reference, per-subsystem deep-dives, and a project-structure listing — all in one monolithic 1765-line SKILL.md. This matches anchor 2 (minimal bundle structure; content that clearly belongs in separate files is inlined); the internal headers and TOC give some navigation, but there are no one-level-deep reference files at all, so it cannot reach 3's "references present" bar.

2 / 5

Total

10

/

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 is in third person and identifies a clear niche, but it reads as a persona/competency statement rather than a task-and-trigger description. It lacks a "Use when..." clause (capping completeness at 3) and omits the concrete platforms, tools, and natural trigger phrases that would make it specific and low-conflict.

Suggestions

Add an explicit trigger clause, e.g. "Use when building or extending AI agents, wiring LLMs to tools, or connecting an agent to messaging platforms like Telegram, Discord, or Slack."

Replace buzzword phrases ("LLM orchestration", "agent autonomy") with concrete actions such as "creates skills from experience", "runs 40+ tools across local, Docker, and SSH backends", "manages session history in SQLite".

Include natural user-facing synonyms and specifics (platform names, "agent framework", "chatbot gateway") so the skill triggers on the phrases users actually say.

DimensionReasoningScore

Specificity

The description names the domain ("self-improving AI agents") and several capability areas ("tool use, multi-platform messaging, and a closed learning loop", "session management"), but phrases like "LLM orchestration", "tool integration", and "agent autonomy" are competency buzzwords rather than concrete actions, and nothing concrete like a platform name or file type is given. It sits between anchor 3 (domain + some actions, not comprehensive) and anchor 4 (several specific actions) — closer to 3 because the action list is generic rather than specific.

3 / 5

Completeness

The "what" is reasonably clear (building self-improving agents with tool use, messaging, and a learning loop), but there is no "Use when..." clause or equivalent explicit trigger guidance, so per the rubric guideline completeness is capped at 3. It is above anchor 2 because the "what" is not vague, but cannot reach 4 without any "when" at all.

3 / 5

Trigger Term Quality

Relevant keywords exist ("AI agents", "tool use", "multi-platform messaging", "LLM"), but common variations users would actually say are missing — no named platforms (Telegram, Discord, Slack), no file extensions, and no synonyms like "chatbot" or "agent framework". This matches anchor 3 (some relevant keywords, missing common variations or synonyms); it is not a 4 because the natural trigger phrases a user would utter are largely absent.

3 / 5

Distinctiveness Conflict Risk

The pairing of "closed learning loop" with "multi-platform messaging" carves a fairly distinct niche that most agent-building skills do not cover, though "LLM orchestration" and "tool integration" overlap with general agent-framework skills. This fits anchor 4 (mostly distinct, minor overlap risk with closely related skills); a 5 would require trigger phrasing that clearly separates it from sibling skills, which is absent.

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 (1766 lines); consider splitting into references/ and linking

Warning

Total

15

/

16

Passed

Repository
fathah/hermes-desktop
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.