CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

pm

Use when the user wants product requirements rather than code: 'write a PRD', 'product requirements', 'spec this feature for the team', 'ooo pm'. Runs a PM-focused interview with automatic question classification and produces a PRD/PM document. Not for implementation requests.

Quality

81%

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

SKILL.md
Quality
Evals
Security

Quality

Content

70%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.

A tightly engineered operational protocol with excellent sequencing and genuinely strong validation/feedback loops, concrete tool calls and sentinels, and no filler. Its weaknesses are repetition of the same lane-handling rules in multiple phrasings, a couple of prose-only tool calls, and a monolithic single-file layout that inlines detail a reference file could carry.

Suggestions

State each lane-handling rule once: consolidate the undispatched/contract-violation rules currently spread across lines 81-86 and 108-110 into a single compact rule set, cutting maybe 10-15 lines of repetition.

Add an explicit arguments block for ouroboros_submit_fanout_results (and one concrete ask_user example), matching the format used for ouroboros_pm_interview, so every tool call is copy-paste ready.

Move the fallback host-setup message and the evidence-block rendering rules into a references/ file (e.g. references/evidence-block.md) with a one-line pointer from Step 3, giving the SKILL.md an overview shape while keeping always-needed loop logic inline.

DimensionReasoningScore

Conciseness

The body is dense, protocol-specific guidance Claude could not know (deferred-schema loading, fan-out submission, refine gate) with essentially no padding about known concepts — but several rules are restated repeatedly: the undispatched-lane rule appears at "Never invent output for a lane you did not run" (line 83) and again at "leave out a lane you submitted as undispatched: a lane that did not run has an empty place, not a clear one" (lines 85-86); the batching rules likewise restate the collect-then-relay requirement twice (line 58). This fits anchor 3 ('mostly efficient but includes some unnecessary explanation or could be tightened'); it does not reach anchor 4's 'only minor instances that could be trimmed', nor sink to anchor 2's several padded sections.

3 / 5

Actionability

Most operations are fully executable: exact tool blocks with arguments (ouroboros_pm_interview with initial_context/cwd/session_id/last_question/answer, the 'generate' retry call), exact sentinel values ('[decide_later]', '[deferred]', '"undispatched": true'), a copy-paste fallback message, and a fixed output template. Minor gaps keep it from anchor 5: the ouroboros_submit_fanout_results call is described in prose without an argument block, the evidence block is explicitly 'placeholders for the shape, not text to copy', and no concrete ask_user call format is shown. Clear score-4 territory ('mostly executable guidance; concrete code or commands with minor gaps'), well above anchor 3's pseudocode level.

4 / 5

Workflow Clarity

The process is a clearly sequenced Step 1–5 with a well-defined Step 3 loop (A/A2/A3/B/C/D), and validation with feedback loops is explicit throughout: re-run tool discovery before each call and retry generation on meta.generation_failed (Step 1, D), resubmit once carrying every lane on contract violations (A2), the refine-gate confirmation with fix options ('Fix the reasoning', 'Let me rewrite it'), and the 'is_complete ONLY by meta.is_complete' rule. This matches anchor 5 ('clear sequence with explicit validation steps; feedback loops for error recovery'); anchor 4 is ruled out because no checkpoints are merely implicit. No destructive/batch cap applies.

5 / 5

Progressive Disclosure

The single 183-line SKILL.md has decent internal structure (numbered steps, lettered sub-sections, fenced blocks) but no bundle files exist and no references are signaled, while substantial detail that could live one level deep is inlined: the ~15-line host-setup fallback message (lines 28-37), the evidence-block rendering rules with a full placeholder template (lines 90-110), and the batched-turn protocol. This fits anchor 3 ('some structure but could be better organized; content that should be separate is inline'); it is above anchor 2 (structure is not minimal) but short of anchor 4, since the file is a full protocol manual rather than an overview pointing to detailed material.

3 / 5

Total

15

/

20

Passed

Description

82%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: explicit 'Use when...' triggers with natural quoted phrases, a clear third-person statement of what the skill does, and an explicit negative boundary against implementation requests. The only weakness is that the capability list is thin — two actions — leaving the full breadth of the skill under-described.

Suggestions

Extend the action list beyond the two named ones (e.g. 'classifies each question, records assumptions and decide-later items as structured answers, and hands off a PM seed artifact for the next pipeline stage') to lift specificity toward anchor 4.

Add one or two common synonyms among the quoted triggers, such as 'requirements doc' or 'product spec', so users who avoid the word 'PRD' still match naturally.

Optionally name the output artifact type in the 'when' clause (e.g. 'when the user wants a written requirements document rather than code') to sharpen the boundary further.

DimensionReasoningScore

Specificity

The description names the domain and two concrete actions — "Runs a PM-focused interview with automatic question classification and produces a PRD/PM document" — but the action list is not comprehensive (no mention of the question loop, assumption/decide-later handling, or the seed handoff artifact). This matches anchor 3 ('names domain and 1-2 concrete actions, but not comprehensive'); it falls short of anchor 4, which expects several specific actions, and is well above anchor 2's minimal/generic actions.

3 / 5

Completeness

Both halves are explicit: what it does ("Runs a PM-focused interview with automatic question classification and produces a PRD/PM document") and when to use it ("Use when the user wants product requirements rather than code", followed by the quoted trigger phrases), plus a negative boundary ("Not for implementation requests"). This squarely matches anchor 5 — both what and when clearly and explicitly answered with concrete trigger phrases — and goes beyond anchor 4, whose 'when' is present but less specific.

5 / 5

Trigger Term Quality

It quotes four natural trigger phrases users would actually say: "'write a PRD'", "'product requirements'", "'spec this feature for the team'", "'ooo pm'" — good, naturally-phrased coverage. A few common synonyms are missing (e.g. 'requirements doc', 'product spec', 'feature spec'), keeping it at anchor 4 rather than anchor 5's comprehensive synonym coverage; it is clearly above anchor 3 because the quoted phrases are varied and idiomatic rather than a single generic keyword.

4 / 5

Distinctiveness Conflict Risk

It carves out a clear niche (PRD generation via a PM-focused interview) and actively reduces conflict risk with the explicit exclusion "Not for implementation requests", distinguishing it from coding/spec-implementation skills. The triggers are distinct to this niche, matching anchor 5 ('clear niche with distinct triggers; minimal conflict risk'); anchor 4's 'minor overlap risk' is not a better fit because the scope boundary is stated outright.

5 / 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

Q00/ouroboros

Path

skills/pm/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.