scaffold-doctor

Name: scaffold-doctor
Author: michellemayes

29stars1forks

How to install scaffold-doctor?

michellemayes/meta-doctor/scaffold-doctor

$npx -y skills add michellemayes/meta-doctor --skill scaffold-doctor

Installs into the current project.

Summary

Scaffolds a complete scheduled "doctor" agent directly into the user's repo. The doctor audits app health on a recurring schedule, routes code-fixable findings to an automated repair pipeline that opens pull requests, and files deduped tracker tickets for everything else. A Slack

Files · 1

View on GitHub

SKILL.md

1	# scaffold-doctor
2
3	Generates a production-ready "doctor" agent inside the target repo by following
4	the pattern described in [`references/architecture.md`](references/architecture.md).
5	Read that file first — it defines the module table, data flow diagram,
6	three run modes, and the error-isolation invariant that all generated code must
7	preserve.
8
9	---
10
11	## Interactive Flow
12
13	Ask questions one at a time, in the order listed below. Use multiple choice
14	wherever possible. Stack questions come first because they determine which
15	modules get generated and which swap-table rows apply.
16
17	---
18
19	### Q1 — Stack
20
21	Ask each sub-question on its own turn.
22
23	1a. Language / runtime
24	```
25	1) TypeScript (default)
26	2) Python
27	3) Other — describe
28	```
29
30	1b. LLM provider
31	```
32	1) Amazon Bedrock (default)
33	2) Anthropic API direct
34	```
35
36	1c. Scheduler
37	```
38	1) AWS EventBridge + Lambda (default)
39	2) GitHub Actions
40	3) POSIX cron / systemd timer
41	4) Kubernetes CronJob
42	```
43
44	1d. VCS
45	```
46	1) GitHub (default)
47	2) GitLab
48	```
49
50	1e. Tracker
51	```
52	1) Linear (default)
53	2) Jira
54	3) GitHub Issues
55	```
56
57	1f. Chat
58	```
59	1) Slack (default)
60	2) Microsoft Teams
61	3) Discord
62	```
63
64	For any non-default choice, note which swap-table row applies (see
65	[`references/swap-table.md`](references/swap-table.md)) — generation will
66	follow that row exactly.
67
68	---
69
70	### Q2 — Product name
71
72	> What is your application's name? (Used as the doctor's identity in the system
73	> prompt, e.g. "Acme Doctor".)
74
75	Free text. Store as `PRODUCT_NAME`.
76
77	---
78
79	### Q3 — Target repo
80
81	> What is the owner/name of the GitHub (or GitLab) repo the doctor will read
82	> from and open PRs against? (e.g. `acme-corp/backend`)
83
84	Free text. Store as `REPO_SLUG`.
85
86	---
87
88	### Q4 — Log source
89
90	> Which log groups should the doctor scan? List ARNs or names, one per line.
91	> Press enter twice when done. (Default: CloudWatch Logs; describe a different
92	> source if applicable.)
93
94	Collect as `LOG_GROUPS[]`. Default lookback: 60 minutes.
95
96	---
97
98	### Q5 — Data source for audits
99
100	> Which query tool should runbooks use?
101	```
102	1) AWS Athena (default)
103	2) Generic SQL (Postgres, MySQL, BigQuery…) — specify driver
104	3) Prometheus
105	4) None — runbooks use only log/metric/VCS tools
106	```
107
108	Store as `DATA_SOURCE`. For non-default choices, generation follows the
109	matching swap-table row in [`references/swap-table.md`](references/swap-table.md).
110
111	---
112
113	### Q6 — Model and region
114
115	> Primary model ID and AWS region (Bedrock) or plain model string (Anthropic
116	> API)?
117	>
118	> Defaults: `us.anthropic.claude-sonnet-4-6`, `us-east-1`. Press enter to
119	> accept.
120
121	Store as `MODEL_ID` and `AWS_REGION`. Fallback model defaults to
122	`us.anthropic.claude-haiku-3-5` (Bedrock) or `claude-haiku-3-5` (Anthropic API).
123
124	---
125
126	### Q7 — Budgets
127
128	Present defaults; user may override any or all.
129
130	\| Parameter \| Default \|
131	\|---\|---\|
132	\| `maxPrsPerRun` \| 3 \|
133	\| `maxTicketsPerRun` \| 5 \|
134	\| `maxIterations` (per runbook) \| 15 \|
135	\| `maxTokensPerCall` \| 4096 \|
136	\| `patchCandidates` \| 3 \|
137	\| `maxRepairAttempts` (per CI PR) \| 3 \|
138
139	---
140
141	### Q8 — Runbook schedule assignment
142
143	> The example runbook `scan-logs` will be registered as a daily runbook.
144	> Do you have additional runbooks to schedule now? (You can add more later with
145	> `add-doctor-runbook`.)
146	```
147	1) No, start with scan-logs only (default)
148	2) Yes — list names and daily/weekly for each
149	```
150
151	Collect `DAILY_RUNBOOKS[]` (starts with `["scan-logs"]`) and
152	`WEEKLY_RUNBOOKS[]` (starts empty).
153
154	---
155
156	## Generation Directive
157
158	Once all answers are collected, generate the doctor engine by following the
159	reference docs below. Do not restate their content — implement exactly
160	what they specify.
161
162	### Module list and file tree
163
164	Follow [`references/architecture.md`](references/architecture.md) — Module
165	Responsibilities table and Default-Stack File Tree. Generate every listed
166	module. For non-default stack choices, apply the relevant rows from
167	[`references/swap-table.md`](references/swap-table.md) before generating.
168
169	### `run-skill` and `prompts/system`
170
171	Follow [`references/agentic-loop.md`](references/agentic-loop.md) completely.
172
173	Port these two reference snippets verbatim (translated to the target
174	language/stack):
175
176	- Model fallback (§5): the `createMessage` wrapper that retries once on
177	the fallback model before propagating.
178	- Anti-loop guard (§4): the `lastCalls` tail-check that injects a nudge
179	text block when th

People who liked this also viewed

6 picks

Type

scaffold-doctor

29stars1forks

View on GitHub

How to install scaffold-doctor?

michellemayes/meta-doctor/scaffold-doctor

$npx -y skills add michellemayes/meta-doctor --skill scaffold-doctor

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# scaffold-doctor
2
3	Generates a production-ready "doctor" agent inside the target repo by following
4	the pattern described in [`references/architecture.md`](references/architecture.md).
5	Read that file first — it defines the module table, data flow diagram,
6	three run modes, and the error-isolation invariant that all generated code must
7	preserve.
8
9	---
10
11	## Interactive Flow
12
13	Ask questions one at a time, in the order listed below. Use multiple choice
14	wherever possible. Stack questions come first because they determine which
15	modules get generated and which swap-table rows apply.
16
17	---
18
19	### Q1 — Stack
20
21	Ask each sub-question on its own turn.
22
23	1a. Language / runtime
24	```
25	1) TypeScript (default)
26	2) Python
27	3) Other — describe
28	```
29
30	1b. LLM provider
31	```
32	1) Amazon Bedrock (default)
33	2) Anthropic API direct
34	```
35
36	1c. Scheduler
37	```
38	1) AWS EventBridge + Lambda (default)
39	2) GitHub Actions
40	3) POSIX cron / systemd timer
41	4) Kubernetes CronJob
42	```
43
44	1d. VCS
45	```
46	1) GitHub (default)
47	2) GitLab
48	```
49
50	1e. Tracker
51	```
52	1) Linear (default)
53	2) Jira
54	3) GitHub Issues
55	```
56
57	1f. Chat
58	```
59	1) Slack (default)
60	2) Microsoft Teams
61	3) Discord
62	```
63
64	For any non-default choice, note which swap-table row applies (see
65	[`references/swap-table.md`](references/swap-table.md)) — generation will
66	follow that row exactly.
67
68	---
69
70	### Q2 — Product name
71
72	> What is your application's name? (Used as the doctor's identity in the system
73	> prompt, e.g. "Acme Doctor".)
74
75	Free text. Store as `PRODUCT_NAME`.
76
77	---
78
79	### Q3 — Target repo
80
81	> What is the owner/name of the GitHub (or GitLab) repo the doctor will read
82	> from and open PRs against? (e.g. `acme-corp/backend`)
83
84	Free text. Store as `REPO_SLUG`.
85
86	---
87
88	### Q4 — Log source
89
90	> Which log groups should the doctor scan? List ARNs or names, one per line.
91	> Press enter twice when done. (Default: CloudWatch Logs; describe a different
92	> source if applicable.)
93
94	Collect as `LOG_GROUPS[]`. Default lookback: 60 minutes.
95
96	---
97
98	### Q5 — Data source for audits
99
100	> Which query tool should runbooks use?
101	```
102	1) AWS Athena (default)
103	2) Generic SQL (Postgres, MySQL, BigQuery…) — specify driver
104	3) Prometheus
105	4) None — runbooks use only log/metric/VCS tools
106	```
107
108	Store as `DATA_SOURCE`. For non-default choices, generation follows the
109	matching swap-table row in [`references/swap-table.md`](references/swap-table.md).
110
111	---
112
113	### Q6 — Model and region
114
115	> Primary model ID and AWS region (Bedrock) or plain model string (Anthropic
116	> API)?
117	>
118	> Defaults: `us.anthropic.claude-sonnet-4-6`, `us-east-1`. Press enter to
119	> accept.
120
121	Store as `MODEL_ID` and `AWS_REGION`. Fallback model defaults to
122	`us.anthropic.claude-haiku-3-5` (Bedrock) or `claude-haiku-3-5` (Anthropic API).
123
124	---
125
126	### Q7 — Budgets
127
128	Present defaults; user may override any or all.
129
130	\| Parameter \| Default \|
131	\|---\|---\|
132	\| `maxPrsPerRun` \| 3 \|
133	\| `maxTicketsPerRun` \| 5 \|
134	\| `maxIterations` (per runbook) \| 15 \|
135	\| `maxTokensPerCall` \| 4096 \|
136	\| `patchCandidates` \| 3 \|
137	\| `maxRepairAttempts` (per CI PR) \| 3 \|
138
139	---
140
141	### Q8 — Runbook schedule assignment
142
143	> The example runbook `scan-logs` will be registered as a daily runbook.
144	> Do you have additional runbooks to schedule now? (You can add more later with
145	> `add-doctor-runbook`.)
146	```
147	1) No, start with scan-logs only (default)
148	2) Yes — list names and daily/weekly for each
149	```
150
151	Collect `DAILY_RUNBOOKS[]` (starts with `["scan-logs"]`) and
152	`WEEKLY_RUNBOOKS[]` (starts empty).
153
154	---
155
156	## Generation Directive
157
158	Once all answers are collected, generate the doctor engine by following the
159	reference docs below. Do not restate their content — implement exactly
160	what they specify.
161
162	### Module list and file tree
163
164	Follow [`references/architecture.md`](references/architecture.md) — Module
165	Responsibilities table and Default-Stack File Tree. Generate every listed
166	module. For non-default stack choices, apply the relevant rows from
167	[`references/swap-table.md`](references/swap-table.md) before generating.
168
169	### `run-skill` and `prompts/system`
170
171	Follow [`references/agentic-loop.md`](references/agentic-loop.md) completely.
172
173	Port these two reference snippets verbatim (translated to the target
174	language/stack):
175
176	- Model fallback (§5): the `createMessage` wrapper that retries once on
177	the fallback model before propagating.
178	- Anti-loop guard (§4): the `lastCalls` tail-check that injects a nudge
179	text block when th

People who liked this also viewed

6 picks

Type