verification-before-completion

Name: verification-before-completion
Author: totvs

108stars50forks

How to install verification-before-completion?

totvs/engpro-advpl-tlpp-skills/verification-before-completion

$npx -y skills add totvs/engpro-advpl-tlpp-skills --skill verification-before-completion

Installs into the current project.

Summary

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always

Files · 1

View on GitHub

SKILL.md

1	# Verification Before Completion
2
3	## Overview
4
5	Claiming work is complete without verification is dishonesty, not efficiency.
6
7	Core principle: Evidence before claims, always.
8
9	Violating the letter of this rule is violating the spirit of this rule.
10
11	## The Iron Law
12
13	```
14	NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
15	```
16
17	If you haven't run the verification command in this message, you cannot claim it passes.
18
19	## The Gate Function
20
21	```
22	BEFORE claiming any status or expressing satisfaction:
23
24	1. IDENTIFY: What command proves this claim?
25	2. RUN: Execute the FULL command (fresh, complete)
26	3. READ: Full output, check exit code, count failures
27	4. VERIFY: Does output confirm the claim?
28	- If NO: State actual status with evidence
29	- If YES: State claim WITH evidence
30	5. ONLY THEN: Make the claim
31
32	Skip any step = lying, not verifying
33	```
34
35	## Common Failures
36
37	\| Claim \| Requires \| Not Sufficient \|
38	\|-------\|----------\|----------------\|
39	\| Tests pass \| Test command output: 0 failures \| Previous run, "should pass" \|
40	\| Linter clean \| Linter output: 0 errors \| Partial check, extrapolation \|
41	\| Build succeeds \| Build command: exit 0 \| Linter passing, logs look good \|
42	\| Bug fixed \| Test original symptom: passes \| Code changed, assumed fixed \|
43	\| Regression test works \| Red-green cycle verified \| Test passes once \|
44	\| Agent completed \| VCS diff shows changes \| Agent reports "success" \|
45	\| Requirements met \| Line-by-line checklist \| Tests passing \|
46
47	## Red Flags - STOP
48
49	- Using "should", "probably", "seems to"
50	- Expressing satisfaction before verification ("Great!", "Perfect!", "Done!", etc.)
51	- About to commit/push/PR without verification
52	- Trusting agent success reports
53	- Relying on partial verification
54	- Thinking "just this once"
55	- Tired and wanting work over
56	- ANY wording implying success without having run verification
57
58	## Rationalization Prevention
59
60	\| Excuse \| Reality \|
61	\|--------\|---------\|
62	\| "Should work now" \| RUN the verification \|
63	\| "I'm confident" \| Confidence ≠ evidence \|
64	\| "Just this once" \| No exceptions \|
65	\| "Linter passed" \| Linter ≠ compiler \|
66	\| "Agent said success" \| Verify independently \|
67	\| "I'm tired" \| Exhaustion ≠ excuse \|
68	\| "Partial check is enough" \| Partial proves nothing \|
69	\| "Different words so rule doesn't apply" \| Spirit over letter \|
70
71	## Key Patterns
72
73	Tests:
74	```
75	✅ [Run test command] [See: 34/34 pass] "All tests pass"
76	❌ "Should pass now" / "Looks correct"
77	```
78
79	Regression tests (TDD Red-Green):
80	```
81	✅ Write → Run (pass) → Revert fix → Run (MUST FAIL) → Restore → Run (pass)
82	❌ "I've written a regression test" (without red-green verification)
83	```
84
85	Build:
86	```
87	✅ [Run build] [See: exit 0] "Build passes"
88	❌ "Linter passed" (linter doesn't check compilation)
89	```
90
91	Requirements:
92	```
93	✅ Re-read plan → Create checklist → Verify each → Report gaps or completion
94	❌ "Tests pass, phase complete"
95	```
96
97	Agent delegation:
98	```
99	✅ Agent reports success → Check VCS diff → Verify changes → Report actual state
100	❌ Trust agent report
101	```
102
103	## Why This Matters
104
105	From 24 failure memories:
106	- your human partner said "I don't believe you" - trust broken
107	- Undefined functions shipped - would crash
108	- Missing requirements shipped - incomplete features
109	- Time wasted on false completion → redirect → rework
110	- Violates: "Honesty is a core value. If you lie, you'll be replaced."
111
112	## When To Apply
113
114	ALWAYS before:
115	- ANY variation of success/completion claims
116	- ANY expression of satisfaction
117	- ANY positive statement about work state
118	- Committing, PR creation, task completion
119	- Moving to next task
120	- Delegating to agents
121
122	Rule applies to:
123	- Exact phrases
124	- Paraphrases and synonyms
125	- Implications of success
126	- ANY communication suggesting completion/correctness
127
128	## The Bottom Line
129
130	No shortcuts for verification.
131
132	Run the command. Read the output. THEN claim the result.
133
134	This is non-negotiable.

People who liked this also viewed

6 picks

Type

verification-before-completion

108stars50forks

View on GitHub

How to install verification-before-completion?

totvs/engpro-advpl-tlpp-skills/verification-before-completion

$npx -y skills add totvs/engpro-advpl-tlpp-skills --skill verification-before-completion

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Verification Before Completion
2
3	## Overview
4
5	Claiming work is complete without verification is dishonesty, not efficiency.
6
7	Core principle: Evidence before claims, always.
8
9	Violating the letter of this rule is violating the spirit of this rule.
10
11	## The Iron Law
12
13	```
14	NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
15	```
16
17	If you haven't run the verification command in this message, you cannot claim it passes.
18
19	## The Gate Function
20
21	```
22	BEFORE claiming any status or expressing satisfaction:
23
24	1. IDENTIFY: What command proves this claim?
25	2. RUN: Execute the FULL command (fresh, complete)
26	3. READ: Full output, check exit code, count failures
27	4. VERIFY: Does output confirm the claim?
28	- If NO: State actual status with evidence
29	- If YES: State claim WITH evidence
30	5. ONLY THEN: Make the claim
31
32	Skip any step = lying, not verifying
33	```
34
35	## Common Failures
36
37	\| Claim \| Requires \| Not Sufficient \|
38	\|-------\|----------\|----------------\|
39	\| Tests pass \| Test command output: 0 failures \| Previous run, "should pass" \|
40	\| Linter clean \| Linter output: 0 errors \| Partial check, extrapolation \|
41	\| Build succeeds \| Build command: exit 0 \| Linter passing, logs look good \|
42	\| Bug fixed \| Test original symptom: passes \| Code changed, assumed fixed \|
43	\| Regression test works \| Red-green cycle verified \| Test passes once \|
44	\| Agent completed \| VCS diff shows changes \| Agent reports "success" \|
45	\| Requirements met \| Line-by-line checklist \| Tests passing \|
46
47	## Red Flags - STOP
48
49	- Using "should", "probably", "seems to"
50	- Expressing satisfaction before verification ("Great!", "Perfect!", "Done!", etc.)
51	- About to commit/push/PR without verification
52	- Trusting agent success reports
53	- Relying on partial verification
54	- Thinking "just this once"
55	- Tired and wanting work over
56	- ANY wording implying success without having run verification
57
58	## Rationalization Prevention
59
60	\| Excuse \| Reality \|
61	\|--------\|---------\|
62	\| "Should work now" \| RUN the verification \|
63	\| "I'm confident" \| Confidence ≠ evidence \|
64	\| "Just this once" \| No exceptions \|
65	\| "Linter passed" \| Linter ≠ compiler \|
66	\| "Agent said success" \| Verify independently \|
67	\| "I'm tired" \| Exhaustion ≠ excuse \|
68	\| "Partial check is enough" \| Partial proves nothing \|
69	\| "Different words so rule doesn't apply" \| Spirit over letter \|
70
71	## Key Patterns
72
73	Tests:
74	```
75	✅ [Run test command] [See: 34/34 pass] "All tests pass"
76	❌ "Should pass now" / "Looks correct"
77	```
78
79	Regression tests (TDD Red-Green):
80	```
81	✅ Write → Run (pass) → Revert fix → Run (MUST FAIL) → Restore → Run (pass)
82	❌ "I've written a regression test" (without red-green verification)
83	```
84
85	Build:
86	```
87	✅ [Run build] [See: exit 0] "Build passes"
88	❌ "Linter passed" (linter doesn't check compilation)
89	```
90
91	Requirements:
92	```
93	✅ Re-read plan → Create checklist → Verify each → Report gaps or completion
94	❌ "Tests pass, phase complete"
95	```
96
97	Agent delegation:
98	```
99	✅ Agent reports success → Check VCS diff → Verify changes → Report actual state
100	❌ Trust agent report
101	```
102
103	## Why This Matters
104
105	From 24 failure memories:
106	- your human partner said "I don't believe you" - trust broken
107	- Undefined functions shipped - would crash
108	- Missing requirements shipped - incomplete features
109	- Time wasted on false completion → redirect → rework
110	- Violates: "Honesty is a core value. If you lie, you'll be replaced."
111
112	## When To Apply
113
114	ALWAYS before:
115	- ANY variation of success/completion claims
116	- ANY expression of satisfaction
117	- ANY positive statement about work state
118	- Committing, PR creation, task completion
119	- Moving to next task
120	- Delegating to agents
121
122	Rule applies to:
123	- Exact phrases
124	- Paraphrases and synonyms
125	- Implications of success
126	- ANY communication suggesting completion/correctness
127
128	## The Bottom Line
129
130	No shortcuts for verification.
131
132	Run the command. Read the output. THEN claim the result.
133
134	This is non-negotiable.

People who liked this also viewed

6 picks

Type