evanflow-iterate

Name: evanflow-iterate
Author: evanklem

419stars18forks

How to install evanflow-iterate?

evanklem/evanflow/evanflow-iterate

$npx -y skills add evanklem/evanflow --skill evanflow-iterate

Installs into the current project.

Summary

Iterative self-review loop after implementing a plan. Re-read changed code with fresh eyes, fix issues found, re-run quality checks, repeat until clean. For UI work, includes visual verification (view the rendered page). Use after evanflow-executing-plans completes; on success, r

Files · 1

View on GitHub

SKILL.md

1	# EvanFlow: Iterate
2
3
4	## Vocabulary
5
6	See `evanflow` meta-skill. Key terms: deep modules, deletion test, vertical slice.
7
8	## When to Use
9
10	- After `evanflow-executing-plans` finishes all tasks
11	- After any non-trivial implementation
12	- When asked to "polish this" / "review this" / "make sure it's clean"
13
14	SKIP when: the change is one line or trivially correct.
15
16	## The Loop
17
18	Repeat until stopping condition met:
19
20	### 1. Run All Quality Checks
21
22	Run the project's quality checks — exact commands are project-specific (see CLAUDE.md or the project's README). Typical examples across stacks:
23
24	```bash
25	# typecheck — one of:
26	tsc --noEmit # TypeScript
27	pnpm typecheck # if scripted
28	cargo check # Rust
29	go vet ./... # Go
30
31	# lint — one of:
32	pnpm lint
33	eslint .
34	cargo clippy
35	ruff check .
36
37	# test — one of:
38	pnpm test
39	pytest
40	cargo test
41	go test ./...
42	```
43
44	If any check fails: fix and restart the loop. Don't proceed to step 2 with broken checks.
45
46	### 2. Re-Read the Diff With Fresh Eyes
47
48	```bash
49	git diff # working-tree changes
50	git diff HEAD~N..HEAD # if reviewing a series of past commits
51	```
52
53	For each changed file, look critically for:
54
55	- Dead code — leftover console.logs, commented-out blocks, unused imports/vars
56	- Naming — does the name match what the code does? (Ubiquitous language matters; see `evanflow-glossary`.)
57	- Deletion test — does each new module earn its existence? Could removing it improve the code?
58	- Magic strings/numbers — should be enums or constants per CLAUDE.md
59	- Error handling — boundary inputs validated? External calls wrapped? Loading/error/empty states in UI?
60	- Type safety — any `any`, `as`, `@ts-ignore`? Justified?
61	- Security — `authenticatedProcedure` where needed? Resource ownership re-derived from `ctx.user`? Per CLAUDE.md.
62	- Test coverage — does the new behavior have a test? Does the test verify behavior, not internals?
63	- Test assertion correctness — research shows 62% of LLM-generated assertions are wrong. For each assertion, would a one-character bug in the implementation still let it pass? If yes, the assertion is too weak.
64	- Scope creep — anything in the diff that wasn't in the plan?
65	- Comments — only WHY notes that explain non-obvious constraints. Delete WHAT comments.
66
67	Fix what you find. Then restart from step 1.
68
69	### 2.5. Five Failure Modes Check
70
71	Industry research identifies five predictable failure modes in agentic coding. After step 2's diff review, do an explicit pass against each:
72
73	- (a) Hallucinated actions — did the implementation invent file paths, env vars, IDs, function names, library APIs, or other external values that aren't authoritatively confirmed? (Example: a `process.env.STRIPE_SECRET_KEY` reference when the actual var name is `STRIPE_SK`.)
74	- (b) Scope creep — does the diff touch files or behaviors not in the plan? Bundled refactors or stylistic changes that should be separate PRs?
75	- (c) Cascading errors — was a failure suppressed/caught/wrapped in a way that hides root cause from callers? Are there silent fallbacks that mask bugs (try/catch returning empty arrays, default values that paper over missing data)?
76	- (d) Context loss — does the diff contradict earlier decisions in the session, the plan, CLAUDE.md, or `CONTEXT.md`? Names, conventions, invariants?
77	- (e) Tool misuse — used the wrong tool (e.g., Bash for file reads, MCP server when CLI was simpler), or used a tool with wrong parameters (e.g., grep without proper escaping, Edit without reading first)?
78
79	For each mode flagged, fix and restart from step 1.
80
81	### 3. (UI work only) Visual Verification
82
83	If the diff touches frontend page or component files and the change has visible output:
84
85	Default approach (no Playwright needed):
86
87	```bash
88	# Make sure your dev server is running first (e.g., pnpm dev, npm run dev, etc.)
89	chromium --headless --no-sandbox \
90	--screenshot=/tmp/iter-$(date +%s).png \
91	--window-size=1440,900 \
92	http://localhost:<port>/<route>
93	```
94
95	(If your project doesn't have `chromium`, substitute `google-chrome --headless` or `chrome --headless` with the same flags.)
96
97	Then read the screenshot:
98
99	```
100	Read /tmp/iter-*.png
101	```
102
103	Check against:
104	- Any brainstorm mockup or design comp the project maintains
105	- The project's design system (colors, spacing, typography, component patterns documented in CLAUDE.md)
106	- Responsive behavior — also screenshot at `--window-size=390,844` (mobile)
107
108	If you need interaction (click, fill, observe modal): use Playwright MCP. If MCP fails with "chrome not found", configure it to use your installed Ch

People who liked this also viewed

6 picks

Type

evanflow-iterate

419stars18forks

View on GitHub

How to install evanflow-iterate?

evanklem/evanflow/evanflow-iterate

$npx -y skills add evanklem/evanflow --skill evanflow-iterate

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# EvanFlow: Iterate
2
3
4	## Vocabulary
5
6	See `evanflow` meta-skill. Key terms: deep modules, deletion test, vertical slice.
7
8	## When to Use
9
10	- After `evanflow-executing-plans` finishes all tasks
11	- After any non-trivial implementation
12	- When asked to "polish this" / "review this" / "make sure it's clean"
13
14	SKIP when: the change is one line or trivially correct.
15
16	## The Loop
17
18	Repeat until stopping condition met:
19
20	### 1. Run All Quality Checks
21
22	Run the project's quality checks — exact commands are project-specific (see CLAUDE.md or the project's README). Typical examples across stacks:
23
24	```bash
25	# typecheck — one of:
26	tsc --noEmit # TypeScript
27	pnpm typecheck # if scripted
28	cargo check # Rust
29	go vet ./... # Go
30
31	# lint — one of:
32	pnpm lint
33	eslint .
34	cargo clippy
35	ruff check .
36
37	# test — one of:
38	pnpm test
39	pytest
40	cargo test
41	go test ./...
42	```
43
44	If any check fails: fix and restart the loop. Don't proceed to step 2 with broken checks.
45
46	### 2. Re-Read the Diff With Fresh Eyes
47
48	```bash
49	git diff # working-tree changes
50	git diff HEAD~N..HEAD # if reviewing a series of past commits
51	```
52
53	For each changed file, look critically for:
54
55	- Dead code — leftover console.logs, commented-out blocks, unused imports/vars
56	- Naming — does the name match what the code does? (Ubiquitous language matters; see `evanflow-glossary`.)
57	- Deletion test — does each new module earn its existence? Could removing it improve the code?
58	- Magic strings/numbers — should be enums or constants per CLAUDE.md
59	- Error handling — boundary inputs validated? External calls wrapped? Loading/error/empty states in UI?
60	- Type safety — any `any`, `as`, `@ts-ignore`? Justified?
61	- Security — `authenticatedProcedure` where needed? Resource ownership re-derived from `ctx.user`? Per CLAUDE.md.
62	- Test coverage — does the new behavior have a test? Does the test verify behavior, not internals?
63	- Test assertion correctness — research shows 62% of LLM-generated assertions are wrong. For each assertion, would a one-character bug in the implementation still let it pass? If yes, the assertion is too weak.
64	- Scope creep — anything in the diff that wasn't in the plan?
65	- Comments — only WHY notes that explain non-obvious constraints. Delete WHAT comments.
66
67	Fix what you find. Then restart from step 1.
68
69	### 2.5. Five Failure Modes Check
70
71	Industry research identifies five predictable failure modes in agentic coding. After step 2's diff review, do an explicit pass against each:
72
73	- (a) Hallucinated actions — did the implementation invent file paths, env vars, IDs, function names, library APIs, or other external values that aren't authoritatively confirmed? (Example: a `process.env.STRIPE_SECRET_KEY` reference when the actual var name is `STRIPE_SK`.)
74	- (b) Scope creep — does the diff touch files or behaviors not in the plan? Bundled refactors or stylistic changes that should be separate PRs?
75	- (c) Cascading errors — was a failure suppressed/caught/wrapped in a way that hides root cause from callers? Are there silent fallbacks that mask bugs (try/catch returning empty arrays, default values that paper over missing data)?
76	- (d) Context loss — does the diff contradict earlier decisions in the session, the plan, CLAUDE.md, or `CONTEXT.md`? Names, conventions, invariants?
77	- (e) Tool misuse — used the wrong tool (e.g., Bash for file reads, MCP server when CLI was simpler), or used a tool with wrong parameters (e.g., grep without proper escaping, Edit without reading first)?
78
79	For each mode flagged, fix and restart from step 1.
80
81	### 3. (UI work only) Visual Verification
82
83	If the diff touches frontend page or component files and the change has visible output:
84
85	Default approach (no Playwright needed):
86
87	```bash
88	# Make sure your dev server is running first (e.g., pnpm dev, npm run dev, etc.)
89	chromium --headless --no-sandbox \
90	--screenshot=/tmp/iter-$(date +%s).png \
91	--window-size=1440,900 \
92	http://localhost:<port>/<route>
93	```
94
95	(If your project doesn't have `chromium`, substitute `google-chrome --headless` or `chrome --headless` with the same flags.)
96
97	Then read the screenshot:
98
99	```
100	Read /tmp/iter-*.png
101	```
102
103	Check against:
104	- Any brainstorm mockup or design comp the project maintains
105	- The project's design system (colors, spacing, typography, component patterns documented in CLAUDE.md)
106	- Responsive behavior — also screenshot at `--window-size=390,844` (mobile)
107
108	If you need interaction (click, fill, observe modal): use Playwright MCP. If MCP fails with "chrome not found", configure it to use your installed Ch

People who liked this also viewed

6 picks

Type