CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

odysseus

Use when the user asks Codex to read or write Odysseus data (todos, email, calendar, memory, documents) or to launch/monitor/stop a Cookbook model-serve task through the scoped Codex Agent API. Requires ODYSSEUS_URL and ODYSSEUS_API_TOKEN.

Quality

82%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

Quality

Content

71%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 highly actionable and lean at the section level, with concrete endpoints, request bodies, scopes, and a genuine debug feedback loop for the cookbook surface. Its weaknesses are structural: everything lives inline in one large SKILL.md with no reference files, the helper-script path is contradictory between sections, and destructive operations lack verification steps.

Suggestions

Split the per-surface endpoint reference (email, calendar, memory, documents, cookbook) into bundled reference files (e.g. references/cookbook.md) and keep SKILL.md as an overview with one-level-deep, clearly signaled pointers.

Pick one canonical path for the helper script — the body uses both 'integrations/codex/scripts/odysseus_api.py' and '~/plugins/odysseus/scripts/odysseus_api.py', and the script is not present in the bundle, so commands are not reliably copy-paste executable.

Add verification steps for destructive and write operations (e.g. GET-confirm a created todo/event before reporting done, confirm a stopped cookbook session before relaunching) to strengthen the workflow feedback loops.

Merge the 'Safety' and 'Forbidden Bypass Pattern' sections, which restate the same restrictions, and drop the opening line that duplicates the frontmatter description.

DimensionReasoningScore

Conciseness

The body is dense and imperative — endpoint, body, and required scope per bullet with no concept explanations — but the 'Safety' and 'Forbidden Bypass Pattern' sections restate the same restrictions, and the opening line duplicates the frontmatter description. This matches 'efficient with minor instances that could be trimmed' rather than the fully lean 5 anchor.

4 / 5

Actionability

Provides concrete endpoints with query parameters, full request bodies ('{"action":"add","title":"Call dentist","due_date":"tomorrow at 5pm"}'), required scopes, and a complete vllm serve command. Not fully copy-paste ready because the helper script is referenced at two contradictory paths ('integrations/codex/scripts/odysseus_api.py' vs '~/plugins/odysseus/scripts/odysseus_api.py') and is not present in the bundle, which keeps it below the 5 anchor.

4 / 5

Workflow Clarity

The cookbook debug loop ('tasks' -> 'output SID 600' -> 'stop SID' -> 'serve' -> re-check output) is a sequenced feedback loop, with checkpoints like the capabilities precheck, 'Treat 403 as an intentional Settings restriction', and 'always cookbook stop your previous attempt before relaunching'. Below 5 because write operations lack verification (no create-then-GET confirm) and DELETE/stop actions have no explicit confirmation step.

4 / 5

Progressive Disclosure

Section headers per surface are well organized, but ~150 lines of full API reference for nine surfaces are inlined in SKILL.md with no bundled reference files at all; the only pointer is an external, inconsistently-pathed script. This matches anchor 3 ('content that should be separate is inline') — not 2, since structure is present, and not 4, since nothing is split out to reference files.

3 / 5

Total

15

/

20

Passed

Description

83%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 strong description with an explicit 'Use when...' trigger clause, concrete enumerated capabilities, and stated prerequisites. Its main weaknesses are minor coverage gaps (chat history, email draft/send) and a few missing natural synonyms (reminders, events, notes) that leave slight trigger and distinctiveness ambiguity.

DimensionReasoningScore

Specificity

Enumerates concrete actions ('read or write Odysseus data (todos, email, calendar, memory, documents)' and 'launch/monitor/stop a Cookbook model-serve task') across two surfaces, but skips chat history and email draft/send, so coverage has minor gaps rather than being comprehensive.

4 / 5

Completeness

Explicitly answers both questions: the 'Use when the user asks Codex to...' clause provides concrete trigger phrases, and the what is concrete (read/write enumerated data types, launch/monitor/stop serve tasks through the scoped API) with prerequisites stated. The 'when' is explicit and specific, exceeding the 4 anchor where it 'could be more explicit'.

5 / 5

Trigger Term Quality

Includes natural user terms like 'todos', 'email', 'calendar', 'memory', 'documents', and 'launch/monitor/stop a Cookbook model-serve task', but omits common synonyms such as 'reminders', 'events', 'notes', or 'chat history'.

4 / 5

Distinctiveness Conflict Risk

Bound to a named system ('Odysseus', 'Codex Agent API', 'Cookbook'), which makes it mostly distinct; bare nouns like 'todos', 'email', and 'calendar' could still overlap with generic productivity skills when the user doesn't mention Odysseus, ruling out 5.

4 / 5

Total

17

/

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

odysseus-dev/odysseus

Path

integrations/codex/skills/odysseus/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.