self-extension-workflow

Name: self-extension-workflow
Author: sordi-ai

17stars3forks

How to install self-extension-workflow?

sordi-ai/skill-everything/self-extension-workflow

$npx -y skills add sordi-ai/skill-everything --skill self-extension-workflow

Installs into the current project.

Summary

Apply when executing the self-extension workflow. Six steps from a mistake to a merged rule, with CI gates and CODEOWNERS approval.

Files · 1

View on GitHub

SKILL.md

1	# Self-extension workflow
2	Exact prompt and procedure the agent executes after every mistake. The agent never pushes to `main` directly. Every self-extension is a PR.
3
4	> Five agent steps plus a search-prerequisite, from mistake to merged rule. The agent never pushes to main. Every rule that lands is CODEOWNERS-approved. This page is the contract.
5
6	<!-- target: ~1100 tokens -->
7
8	---
9
10	## TRIGGER CONDITIONS
11	Five conditions. Any one of them starts the workflow.
12
13	The agent starts this workflow when any of these is met:
14
15	- A test fails and the agent wrote the faulty code.
16	- A code review comment corrects the agent.
17	- The agent realises during implementation that its first approach was wrong.
18	- A deployment problem occurs that the agent caused.
19	- The user explicitly says: "That was wrong" / "Remember this".
20
21	---
22
23	## STEP-BY-STEP PROCEDURE
24	Five steps after the search-prerequisite. Every step is mandatory before the next one.
25
26	### Step 0 — Search before write
27	Mandatory. Duplicates in the error log destroy the signal.
28
29	Before creating a new error entry, search the existing error log:
30
31	```text
32	1. Read skills/error-log/SKILL.md.
33	2. Search for similar errors (same category + similar context).
34	3. If a similar entry exists:
35	- Increment its `count` field by 1.
36	- Update `last_seen` to today.
37	- Supplement the context if needed.
38	- Do NOT create a new entry.
39	4. Only if no similar entry exists, continue with Step 1.
40	```
41
42	### Step 1 — Analyse the error
43	The agent answers six questions before writing anything.
44
45	```text
46	1. What exactly did I do? (concrete code/action)
47	2. What should I have done instead?
48	3. Why did I do it wrong? (false assumption, missing knowledge, carelessness)
49	4. Which category applies?
50	- development: code error, wrong implementation
51	- git: commit/branch mistake
52	- deployment: deployment order, configuration
53	- security: security vulnerability
54	- performance: N+1, missing indexes, oversized datasets
55	- domain: wrong understanding of business rules
56	5. How severe was the error? (critical/high/medium/low)
57	6. Which target file does the rule belong in?
58	```
59
60	### Step 2 — Determine new error ID
61	Take the next number after the latest entry.
62
63	```bash
64	# Find last entry in error-log.md
65	grep "id: ERR-" skills/error-log/SKILL.md \| tail -1
66	# Take next number: ERR-2026-004 -> ERR-2026-005
67	```
68
69	### Step 3 — Formulate the entry
70	Action directive only. The CI verb allow-list rejects anything else.
71
72	Use the template at `skills/error-log/_entry-template.md`. The schema at [`schemas/error-entry.json`](../../schemas/error-entry.json) is enforced by CI.
73
74	> [!WARNING]
75	> CI GATE · verb allow-list — `new_rule` must start with one of `Always`, `Never`, `Before`, `After`, `Prefer`, `Avoid`, `Use`, `Do`, `Ensure`. Anything else is rejected.
76
77	Bad: `"SQL injection is dangerous"`
78	Good: `"Never concatenate user input directly into SQL queries. Always use prepared statements."`
79
80	### Step 4 — Determine target file
81	One file per category. Domain rules go to a project file, not a global one.
82
83	\| Error category \| Target file \|
84	\|---\|---\|
85	\| `development` \| `skills/code-quality/SKILL.md` \|
86	\| `git` \| `skills/git-conventions/SKILL.md` \|
87	\| `deployment` \| `skills/review-deployment/SKILL.md` \|
88	\| `security` \| `skills/code-quality/SKILL.md` (Security section) \|
89	\| `performance` \| `skills/code-quality/SKILL.md` (Performance section) \|
90	\| `domain` \| `skills/<project>/SKILL.md` \|
91
92	### Step 5 — Insert the rule
93	At the end of the matching section. Reference the ERR-ID.
94
95	1. Open the target file.
96	2. Find the matching category section.
97	3. Add the new rule at the end of the section.
98	4. Add reference: `Reference: ERR-YYYY-NNN`.
99
100	### Step 6 — Open a PR
101	Never push to `main` directly. The label triggers the lint-rules workflow.
102
103	```bash
104	git checkout -b learn/ERR-YYYY-NNN
105	git add skills/
106	git diff --cached # MANDATORY human review step
107	git commit -m "learn(errors): ERR-YYYY-NNN — <short description>
108
109	Co-Authored-By: <your real name> <your-email@example.com>"
110	git push -u origin learn/ERR-YYYY-NNN
111	gh pr create --label needs-rule-review \
112	--title "learn(errors): ERR-YYYY-NNN" \
113	--body "Auto-generated rule. Reviewer: confirm rule wording and target file. CI \`lint-rules\` must pass."
114	```
115
116	![PR flow — branch, commit, push, PR; lint-rules and auto-approve-rule-pr CI gates run in parallel; CODEOWNERS human gate; squash-merge into main behind branch protection](../../docs/pr-flow.svg)
117
118	Branch through CI gates and CODEOWNERS into a squash-merge on `main`.
119
120	Why a PR (not a direct commit). Four layers of gating.
121
122	\| Layer \| What it gates \| Where \|
123	\|---\|---\|---\|
124	\| `lint-rules` \| Schema + ve

People who liked this also viewed

6 picks

Type

paperclip-create-agentCreate new agents in Paperclip with governance-aware hiring. Use when you need to inspect adapter configuration options, compare existing agent configs, draft a new agent prompt/config, and submit a hSkillsJul 2026157k5
subagent-driven-developmentUse when executing implementation plans with independent tasks in the current sessionSkillsJul 2026154k259k
dispatching-parallel-agentsUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesSkillsJul 2026143k259k
browser-act-skill-forgeForges reusable Skill packages (SKILL.md + scripts) from website exploration via browser-act — no re-exploration later. Use when: user wants a reusable Skill for any website, needs to understand a sitSkillsJul 202685k4.6k
wizardGenerate an interactive bash wizard that walks a human through a manual procedure — third-party setup, a one-off migration, an A→B state transition — opening URLs, capturing values, confirming each stSkillsJul 202675k182k
wonda-cliUsing the Wonda CLI to generate images, videos, music, and audio from the terminal — plus LinkedIn, Reddit, and X/Twitter research and automationSkillsJul 202666k137

self-extension-workflow

17stars3forks

View on GitHub

How to install self-extension-workflow?

sordi-ai/skill-everything/self-extension-workflow

$npx -y skills add sordi-ai/skill-everything --skill self-extension-workflow

Installs into the current project.

Summary

Apply when executing the self-extension workflow. Six steps from a mistake to a merged rule, with CI gates and CODEOWNERS approval.

Files · 1

View on GitHub

SKILL.md

1	# Self-extension workflow
2	Exact prompt and procedure the agent executes after every mistake. The agent never pushes to `main` directly. Every self-extension is a PR.
3
4	> Five agent steps plus a search-prerequisite, from mistake to merged rule. The agent never pushes to main. Every rule that lands is CODEOWNERS-approved. This page is the contract.
5
6	<!-- target: ~1100 tokens -->
7
8	---
9
10	## TRIGGER CONDITIONS
11	Five conditions. Any one of them starts the workflow.
12
13	The agent starts this workflow when any of these is met:
14
15	- A test fails and the agent wrote the faulty code.
16	- A code review comment corrects the agent.
17	- The agent realises during implementation that its first approach was wrong.
18	- A deployment problem occurs that the agent caused.
19	- The user explicitly says: "That was wrong" / "Remember this".
20
21	---
22
23	## STEP-BY-STEP PROCEDURE
24	Five steps after the search-prerequisite. Every step is mandatory before the next one.
25
26	### Step 0 — Search before write
27	Mandatory. Duplicates in the error log destroy the signal.
28
29	Before creating a new error entry, search the existing error log:
30
31	```text
32	1. Read skills/error-log/SKILL.md.
33	2. Search for similar errors (same category + similar context).
34	3. If a similar entry exists:
35	- Increment its `count` field by 1.
36	- Update `last_seen` to today.
37	- Supplement the context if needed.
38	- Do NOT create a new entry.
39	4. Only if no similar entry exists, continue with Step 1.
40	```
41
42	### Step 1 — Analyse the error
43	The agent answers six questions before writing anything.
44
45	```text
46	1. What exactly did I do? (concrete code/action)
47	2. What should I have done instead?
48	3. Why did I do it wrong? (false assumption, missing knowledge, carelessness)
49	4. Which category applies?
50	- development: code error, wrong implementation
51	- git: commit/branch mistake
52	- deployment: deployment order, configuration
53	- security: security vulnerability
54	- performance: N+1, missing indexes, oversized datasets
55	- domain: wrong understanding of business rules
56	5. How severe was the error? (critical/high/medium/low)
57	6. Which target file does the rule belong in?
58	```
59
60	### Step 2 — Determine new error ID
61	Take the next number after the latest entry.
62
63	```bash
64	# Find last entry in error-log.md
65	grep "id: ERR-" skills/error-log/SKILL.md \| tail -1
66	# Take next number: ERR-2026-004 -> ERR-2026-005
67	```
68
69	### Step 3 — Formulate the entry
70	Action directive only. The CI verb allow-list rejects anything else.
71
72	Use the template at `skills/error-log/_entry-template.md`. The schema at [`schemas/error-entry.json`](../../schemas/error-entry.json) is enforced by CI.
73
74	> [!WARNING]
75	> CI GATE · verb allow-list — `new_rule` must start with one of `Always`, `Never`, `Before`, `After`, `Prefer`, `Avoid`, `Use`, `Do`, `Ensure`. Anything else is rejected.
76
77	Bad: `"SQL injection is dangerous"`
78	Good: `"Never concatenate user input directly into SQL queries. Always use prepared statements."`
79
80	### Step 4 — Determine target file
81	One file per category. Domain rules go to a project file, not a global one.
82
83	\| Error category \| Target file \|
84	\|---\|---\|
85	\| `development` \| `skills/code-quality/SKILL.md` \|
86	\| `git` \| `skills/git-conventions/SKILL.md` \|
87	\| `deployment` \| `skills/review-deployment/SKILL.md` \|
88	\| `security` \| `skills/code-quality/SKILL.md` (Security section) \|
89	\| `performance` \| `skills/code-quality/SKILL.md` (Performance section) \|
90	\| `domain` \| `skills/<project>/SKILL.md` \|
91
92	### Step 5 — Insert the rule
93	At the end of the matching section. Reference the ERR-ID.
94
95	1. Open the target file.
96	2. Find the matching category section.
97	3. Add the new rule at the end of the section.
98	4. Add reference: `Reference: ERR-YYYY-NNN`.
99
100	### Step 6 — Open a PR
101	Never push to `main` directly. The label triggers the lint-rules workflow.
102
103	```bash
104	git checkout -b learn/ERR-YYYY-NNN
105	git add skills/
106	git diff --cached # MANDATORY human review step
107	git commit -m "learn(errors): ERR-YYYY-NNN — <short description>
108
109	Co-Authored-By: <your real name> <your-email@example.com>"
110	git push -u origin learn/ERR-YYYY-NNN
111	gh pr create --label needs-rule-review \
112	--title "learn(errors): ERR-YYYY-NNN" \
113	--body "Auto-generated rule. Reviewer: confirm rule wording and target file. CI \`lint-rules\` must pass."
114	```
115
116	![PR flow — branch, commit, push, PR; lint-rules and auto-approve-rule-pr CI gates run in parallel; CODEOWNERS human gate; squash-merge into main behind branch protection](../../docs/pr-flow.svg)
117
118	Branch through CI gates and CODEOWNERS into a squash-merge on `main`.
119
120	Why a PR (not a direct commit). Four layers of gating.
121
122	\| Layer \| What it gates \| Where \|
123	\|---\|---\|---\|
124	\| `lint-rules` \| Schema + ve

People who liked this also viewed

6 picks

Type