rig-lifecycle

Name: rig-lifecycle
Author: mvschwarz

56stars9forks

How to install rig-lifecycle?

mvschwarz/openrig/rig-lifecycle

$npx -y skills add mvschwarz/openrig --skill rig-lifecycle

Installs into the current project.

Summary

Use when reasoning about the rig lifecycle operations family (create / start / stop / resume / restore / snapshot / release / unclaim / destroy), reading or trusting rig ps / lifecycle projections after recovery, or designing proof for a lifecycle scenario. Covers the 4 failure

Files · 1

View on GitHub

SKILL.md

1	# Rig Lifecycle
2
3	The family of operations that **create, start, stop, resume, restore,
4	snapshot, release, unclaim, and destroy** OpenRig-managed topologies.
5	Includes the user story after reboot: **"bring my work back without
6	turning a clean recovery into a cleanup project."**
7
8	If lifecycle is brittle, every higher-level primitive inherits that
9	brittleness. Queue, workflow, seat continuity, cross-host operation,
10	and RSI all assume that rigs and seats can be restored into known
11	states.
12
13	## Use this when
14
15	- Operating `rig up / down / restore / resume / snapshot / release / unclaim / destroy`
16	- Reading `rig ps` after a reboot or recovery and deciding what to trust
17	- Designing proof for a lifecycle scenario (clean start / warm resume / host reboot / provider auth loss / partial boot / operator recovery)
18	- Reasoning about restore-outcome semantics (`resumed` / `rebuilt` / `fresh` / `failed` / `attention_required`)
19	- Auditing whether a lifecycle proof is in-process bedrock vs requires real reboot evidence
20
21	## Don't use this when
22
23	- The operation is single-command and deterministic — use `openrig-user` skill for CLI surface
24	- The work is operator-level configuration of OpenRig itself — use `openrig-operator`
25	- The work is rig spec authoring — use `openrig-architect`
26
27	## Failure modes (4)
28
29	1. Auto-restore creates a partially restored rig that must be cleaned up before real recovery. Partial restore looks like recovery but isn't; cleanup-before-recovery becomes the actual workload.
30	2. `rig ps` or lifecycle projections report a healthier state than the runtime actually has. Projections are summaries; the runtime is truth. Don't trust projections silently.
31	3. Provider auth is unavailable after reboot and the system treats that as implementation work instead of a human/environment decision. Auth issues are environmental; route to human.
32	4. Resume succeeds for one runtime/provider but fails for another scenario that was never tested. Per-runtime parity assumptions break silently; matrix proof catches them.
33
34	## Restore-honesty rule (load-bearing)
35
36	Failed resume is FAILED loudly. No automatic fresh fallback. Fresh
37	launch is explicit follow-up only. This is enforced architecturally
38	at the daemon level (per `docs/as-built/architecture/architecture-rules-and-event-system.md` rule 15).
39
40	The locked restore-outcome vocabulary:
41
42	\| Outcome \| Meaning \|
43	\|---\|---\|
44	\| `resumed` \| Native runtime resumed the same conversation \|
45	\| `rebuilt` \| New process assembled from artifacts (`session_source: mode: rebuild`) \|
46	\| `fresh` \| New process with no prior continuity \|
47	\| `failed` \| Restore attempted and failed; no automatic fallback \|
48	\| `attention_required` \| Recoverable blocker (provider auth refused, etc.); needs operator action \|
49	\| `n-a` \| Not applicable (terminal nodes, etc.) \|
50
51	Codex auth-refusal returns `attention_required` (recoverable); Claude
52	`looksLikeClaudeLoginPrompt` returns `failed/login_required` (terminal).
53	Cross-runtime alignment is an open follow-up question.
54
55	## Proof standard
56
57	Lifecycle proof should include real reboot or VM-reboot evidence,
58	not only daemon-unit evidence. The minimum useful matrix covers:
59
60	\| Scenario \| What it proves \|
61	\|---\|---\|
62	\| Clean start \| Boot from spec into known state \|
63	\| Warm resume \| `rig down` → `rig up <name>` resumes seats \|
64	\| Host reboot / tmux socket absence \| Recovery from lost tmux connection \|
65	\| Provider auth loss \| Codex/Claude auth refusal handled honestly \|
66	\| Partial boot / partial failure \| Some seats up, some failed; honest reporting \|
67	\| Intentional operator recovery \| Operator-initiated restore from snapshot \|
68
69	Tier 1: in-process bedrock (daemon-unit evidence). Tier 2: real reboot
70	or disposable Tart VM. Tier 1 alone is not lifecycle proof; it's
71	bedrock.
72
73	## Default policy is part of the primitive
74
75	Treat default policy as part of the primitive, not an afterthought.
76	Lifecycle defaults that are too optimistic for the reliability level
77	actually proven create silent harm:
78
79	- `auto-

People who liked this also viewed

6 picks

Type

rig-lifecycle

56stars9forks

View on GitHub

How to install rig-lifecycle?

mvschwarz/openrig/rig-lifecycle

$npx -y skills add mvschwarz/openrig --skill rig-lifecycle

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Rig Lifecycle
2
3	The family of operations that **create, start, stop, resume, restore,
4	snapshot, release, unclaim, and destroy** OpenRig-managed topologies.
5	Includes the user story after reboot: **"bring my work back without
6	turning a clean recovery into a cleanup project."**
7
8	If lifecycle is brittle, every higher-level primitive inherits that
9	brittleness. Queue, workflow, seat continuity, cross-host operation,
10	and RSI all assume that rigs and seats can be restored into known
11	states.
12
13	## Use this when
14
15	- Operating `rig up / down / restore / resume / snapshot / release / unclaim / destroy`
16	- Reading `rig ps` after a reboot or recovery and deciding what to trust
17	- Designing proof for a lifecycle scenario (clean start / warm resume / host reboot / provider auth loss / partial boot / operator recovery)
18	- Reasoning about restore-outcome semantics (`resumed` / `rebuilt` / `fresh` / `failed` / `attention_required`)
19	- Auditing whether a lifecycle proof is in-process bedrock vs requires real reboot evidence
20
21	## Don't use this when
22
23	- The operation is single-command and deterministic — use `openrig-user` skill for CLI surface
24	- The work is operator-level configuration of OpenRig itself — use `openrig-operator`
25	- The work is rig spec authoring — use `openrig-architect`
26
27	## Failure modes (4)
28
29	1. Auto-restore creates a partially restored rig that must be cleaned up before real recovery. Partial restore looks like recovery but isn't; cleanup-before-recovery becomes the actual workload.
30	2. `rig ps` or lifecycle projections report a healthier state than the runtime actually has. Projections are summaries; the runtime is truth. Don't trust projections silently.
31	3. Provider auth is unavailable after reboot and the system treats that as implementation work instead of a human/environment decision. Auth issues are environmental; route to human.
32	4. Resume succeeds for one runtime/provider but fails for another scenario that was never tested. Per-runtime parity assumptions break silently; matrix proof catches them.
33
34	## Restore-honesty rule (load-bearing)
35
36	Failed resume is FAILED loudly. No automatic fresh fallback. Fresh
37	launch is explicit follow-up only. This is enforced architecturally
38	at the daemon level (per `docs/as-built/architecture/architecture-rules-and-event-system.md` rule 15).
39
40	The locked restore-outcome vocabulary:
41
42	\| Outcome \| Meaning \|
43	\|---\|---\|
44	\| `resumed` \| Native runtime resumed the same conversation \|
45	\| `rebuilt` \| New process assembled from artifacts (`session_source: mode: rebuild`) \|
46	\| `fresh` \| New process with no prior continuity \|
47	\| `failed` \| Restore attempted and failed; no automatic fallback \|
48	\| `attention_required` \| Recoverable blocker (provider auth refused, etc.); needs operator action \|
49	\| `n-a` \| Not applicable (terminal nodes, etc.) \|
50
51	Codex auth-refusal returns `attention_required` (recoverable); Claude
52	`looksLikeClaudeLoginPrompt` returns `failed/login_required` (terminal).
53	Cross-runtime alignment is an open follow-up question.
54
55	## Proof standard
56
57	Lifecycle proof should include real reboot or VM-reboot evidence,
58	not only daemon-unit evidence. The minimum useful matrix covers:
59
60	\| Scenario \| What it proves \|
61	\|---\|---\|
62	\| Clean start \| Boot from spec into known state \|
63	\| Warm resume \| `rig down` → `rig up <name>` resumes seats \|
64	\| Host reboot / tmux socket absence \| Recovery from lost tmux connection \|
65	\| Provider auth loss \| Codex/Claude auth refusal handled honestly \|
66	\| Partial boot / partial failure \| Some seats up, some failed; honest reporting \|
67	\| Intentional operator recovery \| Operator-initiated restore from snapshot \|
68
69	Tier 1: in-process bedrock (daemon-unit evidence). Tier 2: real reboot
70	or disposable Tart VM. Tier 1 alone is not lifecycle proof; it's
71	bedrock.
72
73	## Default policy is part of the primitive
74
75	Treat default policy as part of the primitive, not an afterthought.
76	Lifecycle defaults that are too optimistic for the reliability level
77	actually proven create silent harm:
78
79	- `auto-

People who liked this also viewed

6 picks

Type