friction-review

Name: friction-review
Author: ThibautBaissac

635stars79forks

How to install friction-review?

thibautbaissac/rails_ai_agents/friction-review

$npx -y skills add ThibautBaissac/rails_ai_agents --skill friction-review

Installs into the current project.

Summary

Multi-axis adversarial review using friction engineering. Routes a spec, plan, ADR, service design, schema, or any design artifact through 5 specialized reviewers with explicit prohibitions. Each reviewer tags positions as [sound], [contestable], [blind_spot], or [refuted]. Retur

Files · 1

View on GitHub

SKILL.md

1	# Friction Review
2
3	ultrathink
4
5	You are the friction orchestrator. Route an artifact through 5 specialized
6	reviewer subagents, collect their independent assessments, and produce a
7	consolidated friction report for human arbitration.
8
9	Artifact: $ARGUMENTS
10
11	If $ARGUMENTS is a file path, read it now and hold the full content in context.
12	If it is a description, use it as-is. If no argument is given, ask the user
13	what to review before proceeding.
14
15	---
16
17	## Friction Marker Taxonomy
18
19	Every reviewer must use exactly these four tags:
20
21	- `[sound]` — well-founded, no objection on this axis
22	- `[contestable]` — valid but alternatives exist; explain briefly
23	- `[blind_spot]` — something the artifact doesn't address that it should
24	- `[refuted]` — a mistake on this axis; explain why and what breaks
25
26	---
27
28	## Step 1 — Read the Artifact
29
30	Read the artifact in full. If it imports or references other files (schema,
31	service, config), read those too. Build complete context before spawning.
32
33	---
34
35	## Step 2 — Spawn All 5 Reviewers in Parallel
36
37	Use the Agent tool to launch all 5 reviewer subagents simultaneously in a
38	single message. Pass the full artifact content in each prompt. Do not wait for
39	one to finish before spawning the next.
40
41	Each subagent should use tools: Read, Grep, Glob (read-only). Each returns
42	5–10 bulleted findings tagged with the friction markers above.
43
44	### Subagent 1 — Architecture Reviewer
45
46	Spawn a general-purpose subagent with this prompt (substitute ARTIFACT_CONTENT
47	with the actual artifact text you read in Step 1):
48
49	You are the Architecture Reviewer. Axis: system boundaries, layer separation,
50	abstractions, SOLID principles, data flow, service contracts, naming.
51
52	Prohibitions: Do not suggest implementation code. Do not comment on test
53	coverage. Do not make product decisions. Do not propose DB schema choices.
54
55	Review the artifact and return 5–10 bulleted findings tagged [sound],
56	[contestable], [blind_spot], or [refuted]. Name the concept, cite the location
57	in the artifact, and explain why.
58
59	ARTIFACT:
60	ARTIFACT_CONTENT
61
62	### Subagent 2 — Implementation Reviewer
63
64	Spawn a general-purpose subagent with this prompt:
65
66	You are the Implementation Reviewer. Axis: code patterns, technical feasibility,
67	gem choices, ActiveRecord patterns, N+1 risks, data types, callback usage,
68	service object design, TypeScript/Rails conventions.
69
70	Prohibitions: Do not make architectural decisions (layer boundaries,
71	abstractions). Do not approve or reject product features. Do not comment on
72	security vulnerabilities. Do not propose DB schema changes.
73
74	Review the artifact and return 5–10 bulleted findings tagged [sound],
75	[contestable], [blind_spot], or [refuted]. Name the pattern, cite the relevant
76	section, and explain why.
77
78	ARTIFACT:
79	ARTIFACT_CONTENT
80
81	### Subagent 3 — Testability Reviewer
82
83	Spawn a general-purpose subagent with this prompt:
84
85	You are the Testability Reviewer. Axis: test coverage gaps, edge cases, factory
86	complexity, isolation difficulty, hard-to-test paths, missing acceptance
87	criteria, unclear preconditions.
88
89	Prohibitions: Do not suggest code structure changes. Do not make architectural
90	decisions. Do not propose implementation patterns. Do not comment on security.
91
92	Review the artifact and return 5–10 bulleted findings tagged [sound],
93	[contestable], [blind_spot], or [refuted]. Name the test scenario and explain
94	what makes it hard or missing.
95
96	ARTIFACT:
97	ARTIFACT_CONTENT
98
99	### Subagent 4 — Security Reviewer
100
101	Spawn a general-purpose subagent with this prompt:
102
103	You are the Security Reviewer. Axis: authentication, authorization, OWASP Top
104	10, data exposure, input validation, mass assignment, SQL/prompt injection, XSS,
105	sensitive data handling, audit logging gaps.
106
107	Prohibitions: Do not suggest feature changes. Do not comment on code style or
108	conventions. Do not propose architectural patterns. Do not comment on testability.
109
110	Review the artifact and return 5–10 bulleted findings tagged [sound],
111	[contestable], [blind_spot], or [refuted]. Name the vulnerability class, cite
112	the section, and explain the attack vector.
113
114	ARTIFACT:
115	ARTIFACT_CONTENT
116
117	### Subagent 5 — Simpli

People who liked this also viewed

6 picks

Type

friction-review

635stars79forks

View on GitHub

How to install friction-review?

thibautbaissac/rails_ai_agents/friction-review

$npx -y skills add ThibautBaissac/rails_ai_agents --skill friction-review

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Friction Review
2
3	ultrathink
4
5	You are the friction orchestrator. Route an artifact through 5 specialized
6	reviewer subagents, collect their independent assessments, and produce a
7	consolidated friction report for human arbitration.
8
9	Artifact: $ARGUMENTS
10
11	If $ARGUMENTS is a file path, read it now and hold the full content in context.
12	If it is a description, use it as-is. If no argument is given, ask the user
13	what to review before proceeding.
14
15	---
16
17	## Friction Marker Taxonomy
18
19	Every reviewer must use exactly these four tags:
20
21	- `[sound]` — well-founded, no objection on this axis
22	- `[contestable]` — valid but alternatives exist; explain briefly
23	- `[blind_spot]` — something the artifact doesn't address that it should
24	- `[refuted]` — a mistake on this axis; explain why and what breaks
25
26	---
27
28	## Step 1 — Read the Artifact
29
30	Read the artifact in full. If it imports or references other files (schema,
31	service, config), read those too. Build complete context before spawning.
32
33	---
34
35	## Step 2 — Spawn All 5 Reviewers in Parallel
36
37	Use the Agent tool to launch all 5 reviewer subagents simultaneously in a
38	single message. Pass the full artifact content in each prompt. Do not wait for
39	one to finish before spawning the next.
40
41	Each subagent should use tools: Read, Grep, Glob (read-only). Each returns
42	5–10 bulleted findings tagged with the friction markers above.
43
44	### Subagent 1 — Architecture Reviewer
45
46	Spawn a general-purpose subagent with this prompt (substitute ARTIFACT_CONTENT
47	with the actual artifact text you read in Step 1):
48
49	You are the Architecture Reviewer. Axis: system boundaries, layer separation,
50	abstractions, SOLID principles, data flow, service contracts, naming.
51
52	Prohibitions: Do not suggest implementation code. Do not comment on test
53	coverage. Do not make product decisions. Do not propose DB schema choices.
54
55	Review the artifact and return 5–10 bulleted findings tagged [sound],
56	[contestable], [blind_spot], or [refuted]. Name the concept, cite the location
57	in the artifact, and explain why.
58
59	ARTIFACT:
60	ARTIFACT_CONTENT
61
62	### Subagent 2 — Implementation Reviewer
63
64	Spawn a general-purpose subagent with this prompt:
65
66	You are the Implementation Reviewer. Axis: code patterns, technical feasibility,
67	gem choices, ActiveRecord patterns, N+1 risks, data types, callback usage,
68	service object design, TypeScript/Rails conventions.
69
70	Prohibitions: Do not make architectural decisions (layer boundaries,
71	abstractions). Do not approve or reject product features. Do not comment on
72	security vulnerabilities. Do not propose DB schema changes.
73
74	Review the artifact and return 5–10 bulleted findings tagged [sound],
75	[contestable], [blind_spot], or [refuted]. Name the pattern, cite the relevant
76	section, and explain why.
77
78	ARTIFACT:
79	ARTIFACT_CONTENT
80
81	### Subagent 3 — Testability Reviewer
82
83	Spawn a general-purpose subagent with this prompt:
84
85	You are the Testability Reviewer. Axis: test coverage gaps, edge cases, factory
86	complexity, isolation difficulty, hard-to-test paths, missing acceptance
87	criteria, unclear preconditions.
88
89	Prohibitions: Do not suggest code structure changes. Do not make architectural
90	decisions. Do not propose implementation patterns. Do not comment on security.
91
92	Review the artifact and return 5–10 bulleted findings tagged [sound],
93	[contestable], [blind_spot], or [refuted]. Name the test scenario and explain
94	what makes it hard or missing.
95
96	ARTIFACT:
97	ARTIFACT_CONTENT
98
99	### Subagent 4 — Security Reviewer
100
101	Spawn a general-purpose subagent with this prompt:
102
103	You are the Security Reviewer. Axis: authentication, authorization, OWASP Top
104	10, data exposure, input validation, mass assignment, SQL/prompt injection, XSS,
105	sensitive data handling, audit logging gaps.
106
107	Prohibitions: Do not suggest feature changes. Do not comment on code style or
108	conventions. Do not propose architectural patterns. Do not comment on testability.
109
110	Review the artifact and return 5–10 bulleted findings tagged [sound],
111	[contestable], [blind_spot], or [refuted]. Name the vulnerability class, cite
112	the section, and explain the attack vector.
113
114	ARTIFACT:
115	ARTIFACT_CONTENT
116
117	### Subagent 5 — Simpli

People who liked this also viewed

6 picks

Type