continuous-improvement-engine

Capture, structure, and recycle lessons from active and closed legal matters. Three modes: in-flight capture (triggered by scope changes, risk events, status updates — highest value), mid-matter review (phase gates or quarterly), and matter close retrospective (full structured findings). Lessons are formatted for immediate reuse, not filed and forgotten. Use when a risk materialises, a scope change lands, a phase completes, or a matter closes and you want to convert what happened into something useful for the next matter. Trigger on: 'capture a lesson', 'what did we learn', 'matter close', 'retrospective', 'lessons learned', 'what went wrong', 'what worked', 'phase gate review', 'debrief', 'extract the learning', 'close the matter', 'what should we do differently', 'pattern from this matter', 'improve the next one'.

Quality

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Passed

No known issues

Quality

Content

77%

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 with concrete schemas and well-sequenced workflows including explicit validation and approval gates. Its weaknesses are conciseness (repeated rationale and a concept-explaining Domain Knowledge section) and progressive disclosure (a long single file with no reference files to offload the schema material).

Suggestions

Move the 'Domain Knowledge — Why Lessons Don't Get Reused' rationale and the repeated failure-mode framing into a short reference file or trim to one sentence; the body restates the capture-lag / format-mismatch / no-feedback-loop argument several times.

Extract the four reusable output schemas (LESSON ENTRY, PROPOSAL, MID-MATTER REVIEW, MATTER CLOSE RETROSPECTIVE) into a references/ file (e.g. OUTPUT_SCHEMAS.md) linked from each mode, leaving the body as an overview — this would lift progressive disclosure and reduce inline length.

Tighten the M365 Connected Mode section, which lists many search-term examples that overlap with the Mode 0 signal table; consolidate to avoid token duplication.

Dimension	Reasoning	Score
Conciseness	Mostly operational and earning its tokens, but the 'Domain Knowledge — Why Lessons Don't Get Reused' section and repeated framing paragraphs (failure-mode rationale restated several times) explain context Claude can infer, so it sits at 'mostly efficient but could be tightened' rather than the lean anchor at 3.	2 / 3
Actionability	Provides fully concrete, copy-ready schemas — LESSON ENTRY, PROPOSAL, MID-MATTER REVIEW, MATTER CLOSE RETROSPECTIVE blocks with exact fields and placeholders — plus explicit output rules, matching the executable/ready anchor at 3 rather than the pseudocode anchor at 2.	3 / 3
Workflow Clarity	Each mode is a clearly sequenced workflow with an explicit identifier hard-gate, pattern-detection prompts, and a non-negotiable approval gate before any SKILL.md self-modification — the validation/checkpoint anchor at 3, not the 'sequence present but checkpoints missing' anchor at 2.	3 / 3
Progressive Disclosure	No bundle files exist, so the ~400-line body is a single monolithic SKILL.md with no one-level-deep references to split out the heavy schema/API material; well-sectioned but inline where separation would help, matching the 2 anchor rather than the clear-overview-with-refs anchor at 3.	2 / 3
	Total	10 / 12 Passed

Description

100%

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 strong: it states concrete capabilities across three modes, supplies abundant natural trigger terms, and explicitly covers both what and when. It carves a distinct niche with clear non-overlap against sibling skills. It is third-person throughout, satisfying the voice guideline.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions across three modes — 'Capture, structure, and recycle lessons', 'in-flight capture', 'mid-matter review', and 'matter close retrospective' — matching the 'lists multiple specific concrete actions' anchor rather than the single-domain anchor at 2.	3 / 3
Completeness	Explicitly answers both what (the three capture/recycle modes) and when ('Use when a risk materialises, a scope change lands, a phase completes, or a matter closes'), satisfying the explicit-trigger anchor at 3 rather than the what-only anchor at 2.	3 / 3
Trigger Term Quality	Dense natural trigger phrasing users would actually say — 'capture a lesson', 'what did we learn', 'matter close', 'retrospective', 'lessons learned', 'what went wrong' — giving good coverage rather than the partial coverage at 2.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear legal-matter lesson-capture niche and is unlikely to conflict — the body even enumerates adjacent skills it does NOT duplicate — rather than the overlapping 'works with document files' anchor at 2.	3 / 3
	Total	12 / 12 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: legalopsconsulting/lpm-skills
Commit: 178ed52

Reviewed: 6 days 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.