repo-visuals-retro

Name: repo-visuals-retro
Author: livlign

18stars1forks

How to install repo-visuals-retro?

livlign/claude-skills/repo-visuals-retro

$npx -y skills add livlign/claude-skills --skill repo-visuals-retro

Installs into the current project.

Summary

Retrospective meta-skill for the repo-visuals skill. Reads accumulated evaluation logs from past runs, spots patterns (recurring low-score criteria, repeated iteration failures, unsatisfied requests), consults other expert skills (skill-creator, frontend-design, etc.) where rel

Files · 1

View on GitHub

SKILL.md

1	# repo-visuals-retro
2
3	The `repo-visuals` skill scores every run on five criteria (see `repo-visuals/SKILL.md` §6). This meta-skill turns that accumulated evidence into skill improvements.
4
5	## When to invoke
6
7	- You have ≥ 5 runs logged in `./evaluations/runs/` in the user's working directory (before that, the sample is noise).
8	- Something feels off across multiple runs and you want a structured look.
9	- A major change to `repo-visuals` is planned and you want the evidence base first.
10
11	Do not run this every session. Retros are more valuable with accumulated samples.
12
13	## Inputs
14
15	- The `repo-visuals` skill's `SKILL.md` — the current skill definition (sibling skill in the same plugin)
16	- `./evaluations/index.md` in the user's working directory — curated aggregate from previous retros
17	- `./evaluations/runs/*.md` in the user's working directory — all per-run raw evaluations since the last retro
18	- Available expert skills in the environment (detected, not assumed)
19
20	## Workflow
21
22	### 1. Read & tabulate
23
24	Read all inputs. Build a small table:
25
26	- Average score per criterion, overall and by repo-type bucket (CLI, library, web-app, etc. — infer from scan metadata in each run)
27	- Variance per criterion (high variance = the skill is inconsistent on that axis)
28	- Free-text feedback clustered by theme
29
30	### 2. Identify patterns
31
32	Name each pattern concretely:
33
34	- "Hero moment delivery averaged 2.6 / 5 across the 4 CLI-tool runs but 4.2 / 5 across library runs"
35	- "Technical polish dropped whenever the stage was < 400 px tall — type legibility likely"
36	- "3 runs had identical user feedback: 'loop seam is jarring'"
37
38	### 3. Consult expert skills
39
40	For each pattern, consult the relevant expert skill(s) if available in the environment. Examples:
41
42	- Visual / design critique → `frontend-design` skill. Feed it the final HTML + screenshots from low-scoring runs; ask for specific design-rule violations.
43	- Skill structure / prompt design → `skill-creator` skill. Ask whether the skill's Phase N language is ambiguous or missing a step.
44	- Evaluation methodology → any `evaluate-plugins` style skill. Ask whether the 5-criterion scorecard is well-calibrated.
45	- Domain-specific knowledge → e.g. an ast-graph / codebase-compare skill if patterns suggest weaknesses in the scan phase.
46
47	Detect which skills are actually available — don't reference non-existent ones.
48
49	### 4. Propose edits
50
51	Write a diff-style proposal: for each pattern, which section of the `repo-visuals` skill's `SKILL.md` changes and why. Example:
52
53	```
54	Pattern: Hero moment delivery weak on CLI tools (2.6/5 avg over 4 runs).
55	Hypothesis: §1.4a probes don't surface CLI-specific hero moments
56	(install speed, clear command output, shell ergonomics).
57	Proposed edit to §1.4a: add a CLI-tool branch to the probing list with
58	3 tool-specific prompts.
59	```
60
61	Show the proposed diff. Wait for user approval before applying.
62
63	### 5. Apply & update aggregate
64
65	On user approval:
66
67	- Edit the `repo-visuals` skill's `SKILL.md` with the approved changes
68	- Append a retro summary to `./evaluations/index.md` (date, patterns identified, edits applied)
69	- Move the raw run files that informed this retro into `./evaluations/runs/processed/` so the next retro only sees new samples
70
71	## Outputs
72
73	- Updated `repo-visuals` skill `SKILL.md`
74	- Updated `./evaluations/index.md` with retro summary
75	- Moved processed run files
76
77	## What this skill does NOT do
78
79	- Does not run `repo-visuals` itself
80	- Does not auto-apply edits without user approval
81	- Does not invent criteria — it only improves the skill against its existing scorecard

People who liked this also viewed

6 picks

Type

repo-visuals-retro

18stars1forks

View on GitHub

How to install repo-visuals-retro?

livlign/claude-skills/repo-visuals-retro

$npx -y skills add livlign/claude-skills --skill repo-visuals-retro

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# repo-visuals-retro
2
3	The `repo-visuals` skill scores every run on five criteria (see `repo-visuals/SKILL.md` §6). This meta-skill turns that accumulated evidence into skill improvements.
4
5	## When to invoke
6
7	- You have ≥ 5 runs logged in `./evaluations/runs/` in the user's working directory (before that, the sample is noise).
8	- Something feels off across multiple runs and you want a structured look.
9	- A major change to `repo-visuals` is planned and you want the evidence base first.
10
11	Do not run this every session. Retros are more valuable with accumulated samples.
12
13	## Inputs
14
15	- The `repo-visuals` skill's `SKILL.md` — the current skill definition (sibling skill in the same plugin)
16	- `./evaluations/index.md` in the user's working directory — curated aggregate from previous retros
17	- `./evaluations/runs/*.md` in the user's working directory — all per-run raw evaluations since the last retro
18	- Available expert skills in the environment (detected, not assumed)
19
20	## Workflow
21
22	### 1. Read & tabulate
23
24	Read all inputs. Build a small table:
25
26	- Average score per criterion, overall and by repo-type bucket (CLI, library, web-app, etc. — infer from scan metadata in each run)
27	- Variance per criterion (high variance = the skill is inconsistent on that axis)
28	- Free-text feedback clustered by theme
29
30	### 2. Identify patterns
31
32	Name each pattern concretely:
33
34	- "Hero moment delivery averaged 2.6 / 5 across the 4 CLI-tool runs but 4.2 / 5 across library runs"
35	- "Technical polish dropped whenever the stage was < 400 px tall — type legibility likely"
36	- "3 runs had identical user feedback: 'loop seam is jarring'"
37
38	### 3. Consult expert skills
39
40	For each pattern, consult the relevant expert skill(s) if available in the environment. Examples:
41
42	- Visual / design critique → `frontend-design` skill. Feed it the final HTML + screenshots from low-scoring runs; ask for specific design-rule violations.
43	- Skill structure / prompt design → `skill-creator` skill. Ask whether the skill's Phase N language is ambiguous or missing a step.
44	- Evaluation methodology → any `evaluate-plugins` style skill. Ask whether the 5-criterion scorecard is well-calibrated.
45	- Domain-specific knowledge → e.g. an ast-graph / codebase-compare skill if patterns suggest weaknesses in the scan phase.
46
47	Detect which skills are actually available — don't reference non-existent ones.
48
49	### 4. Propose edits
50
51	Write a diff-style proposal: for each pattern, which section of the `repo-visuals` skill's `SKILL.md` changes and why. Example:
52
53	```
54	Pattern: Hero moment delivery weak on CLI tools (2.6/5 avg over 4 runs).
55	Hypothesis: §1.4a probes don't surface CLI-specific hero moments
56	(install speed, clear command output, shell ergonomics).
57	Proposed edit to §1.4a: add a CLI-tool branch to the probing list with
58	3 tool-specific prompts.
59	```
60
61	Show the proposed diff. Wait for user approval before applying.
62
63	### 5. Apply & update aggregate
64
65	On user approval:
66
67	- Edit the `repo-visuals` skill's `SKILL.md` with the approved changes
68	- Append a retro summary to `./evaluations/index.md` (date, patterns identified, edits applied)
69	- Move the raw run files that informed this retro into `./evaluations/runs/processed/` so the next retro only sees new samples
70
71	## Outputs
72
73	- Updated `repo-visuals` skill `SKILL.md`
74	- Updated `./evaluations/index.md` with retro summary
75	- Moved processed run files
76
77	## What this skill does NOT do
78
79	- Does not run `repo-visuals` itself
80	- Does not auto-apply edits without user approval
81	- Does not invent criteria — it only improves the skill against its existing scorecard

People who liked this also viewed

6 picks

Type