agent-startup-and-context-ingestion

Name: agent-startup-and-context-ingestion
Author: mvschwarz

56stars9forks

How to install agent-startup-and-context-ingestion?

mvschwarz/openrig/agent-startup-and-context-ingestion

$npx -y skills add mvschwarz/openrig --skill agent-startup-and-context-ingestion

Installs into the current project.

Summary

Use when designing or auditing how an agent becomes useful after launch — AGENTS.md overlays, role files, skills, rig specs, workflow specs, startup checklists, refocus messages, "rig context" surface. Covers the 4 failure modes that make startup context fail (old rig spec misses

Files · 1

View on GitHub

SKILL.md

1	# Agent Startup and Context Ingestion
2
3	How an agent becomes useful after launch: **AGENTS.md overlays, role
4	files, skills, rig specs, workflow specs, startup checklists, refocus
5	messages**, and any future "rig context" command or skill-librarian
6	surface.
7
8	It is the current concrete startup path inside the broader
9	context-engineering-and-retrieval primitive. Startup gets a seat into
10	the right initial shape; context engineering is the larger question of
11	how a seat gets the right context for the work it is doing right now.
12
13	**Most coordination failures are not tool failures; they are context
14	failures.** Agents need to know their role, operating mode, coordination
15	convention, boundaries, and current product intent. If startup context
16	is scattered or stale, agents execute the wrong thing very efficiently.
17
18	## Use this when
19
20	- Authoring a new agent's startup files (role / culture / startup-context)
21	- Refreshing a seat that's been running on stale guidance
22	- Auditing whether agents got the current operating mode (not just what's in old files)
23	- Designing the orchestrator → next-agent context-transmission shape
24	- Building a startup map for OpenRig-building rigs
25
26	## Don't use this when
27
28	- The agent is being created via Agent Starter — the starter's manifest carries startup context
29	- The work is artifact-backed mental-model rebuild from a packet — that's `claude-compaction-restore`
30	- The intent is to ship reusable startup content as a skill — write the skill against the agentskills.io specification, not as inline startup prose
31
32	## Failure modes (4)
33
34	1. A new agent starts from an old rig spec and misses the current operating mode. Specs go stale; current state must be visible at startup, not just historical config.
35	2. Guidance is written to a file that future agents read, but current agents are never told. File edits don't propagate to running sessions. Cultural rollout (broadcast + fleet-changes-feed) is needed alongside file edits.
36	3. A startup file becomes a dumping ground and loses the map-to-canonical-sources role. Startup should point AT canonical sources; it shouldn't TRY to be one.
37	4. The orchestrator transmits implementation instructions without preserving product intent. Instructions decay; intent travels.
38
39	## Proof standard
40
41	Startup proof should:
42
43	1. Launch or refresh a seat
44	2. Inspect what it actually read
45	3. Verify it can state the current role, mode, active constraints, and next handoff convention in its own words
46
47	## Cross-runtime startup paths (5; do not collapse)
48
49	When a seat's startup is artifact-backed mental-model rebuild (active-work
50	reentry case, distinct from reusable Agent Starters / priming packs):
51
52	- See the cross-runtime restore/reentry packet standard v0
53	- Source-trust ranking applies: `rig whoami` > target rigspec > bounded latest transcript > full transcript > touched-files > `restore-summary.json`
54
55	## Memory surfaces consumed at startup
56
57	The primary surfaces a startup ingestion path reads:
58
59	- AGENTS / role / CULTURE / startup overlays (the per-seat identity layer)
60	- startup replay context (what the agent did before the current boot)
61	- restore / reentry packets and Agent Starters (cross-runtime startup path)
62
63	Authority-rank + read-vs-write posture governs which surfaces a
64	startup-ingestion path can write vs only read; treat overlays as
65	write-restricted unless your role explicitly authors them.
66
67	## Startup files vs skills (the distinction)
68
69	Per `agent-startup-guide.md` (product reference doc) and the team handbook:
70
71	\| Startup files \| Skills \|
72	\|---\|---\|
73	\| Rig-specific, role-specific identity \| Reusable SOPs / methodology / knowledge \|
74	\| Tell agent WHO it is, WHAT it's working on, HOW this team operates \| Tell agent HOW to do something (transferable across rigs) \|
75	\| Examples: `role.md`, `CULTURE.md`, `startup/context.md` \| Examples: `openrig-user`, `test-driven-development`, `dogfood` \|
76	\| Authored per-rig \| Authored once, used everywhere \|
77
78	Don't put skill content in startup files. Don't put identity content
79	in skills. The 7-layer additive s

People who liked this also viewed

6 picks

Type

agent-startup-and-context-ingestion

56stars9forks

View on GitHub

How to install agent-startup-and-context-ingestion?

mvschwarz/openrig/agent-startup-and-context-ingestion

$npx -y skills add mvschwarz/openrig --skill agent-startup-and-context-ingestion

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Agent Startup and Context Ingestion
2
3	How an agent becomes useful after launch: **AGENTS.md overlays, role
4	files, skills, rig specs, workflow specs, startup checklists, refocus
5	messages**, and any future "rig context" command or skill-librarian
6	surface.
7
8	It is the current concrete startup path inside the broader
9	context-engineering-and-retrieval primitive. Startup gets a seat into
10	the right initial shape; context engineering is the larger question of
11	how a seat gets the right context for the work it is doing right now.
12
13	**Most coordination failures are not tool failures; they are context
14	failures.** Agents need to know their role, operating mode, coordination
15	convention, boundaries, and current product intent. If startup context
16	is scattered or stale, agents execute the wrong thing very efficiently.
17
18	## Use this when
19
20	- Authoring a new agent's startup files (role / culture / startup-context)
21	- Refreshing a seat that's been running on stale guidance
22	- Auditing whether agents got the current operating mode (not just what's in old files)
23	- Designing the orchestrator → next-agent context-transmission shape
24	- Building a startup map for OpenRig-building rigs
25
26	## Don't use this when
27
28	- The agent is being created via Agent Starter — the starter's manifest carries startup context
29	- The work is artifact-backed mental-model rebuild from a packet — that's `claude-compaction-restore`
30	- The intent is to ship reusable startup content as a skill — write the skill against the agentskills.io specification, not as inline startup prose
31
32	## Failure modes (4)
33
34	1. A new agent starts from an old rig spec and misses the current operating mode. Specs go stale; current state must be visible at startup, not just historical config.
35	2. Guidance is written to a file that future agents read, but current agents are never told. File edits don't propagate to running sessions. Cultural rollout (broadcast + fleet-changes-feed) is needed alongside file edits.
36	3. A startup file becomes a dumping ground and loses the map-to-canonical-sources role. Startup should point AT canonical sources; it shouldn't TRY to be one.
37	4. The orchestrator transmits implementation instructions without preserving product intent. Instructions decay; intent travels.
38
39	## Proof standard
40
41	Startup proof should:
42
43	1. Launch or refresh a seat
44	2. Inspect what it actually read
45	3. Verify it can state the current role, mode, active constraints, and next handoff convention in its own words
46
47	## Cross-runtime startup paths (5; do not collapse)
48
49	When a seat's startup is artifact-backed mental-model rebuild (active-work
50	reentry case, distinct from reusable Agent Starters / priming packs):
51
52	- See the cross-runtime restore/reentry packet standard v0
53	- Source-trust ranking applies: `rig whoami` > target rigspec > bounded latest transcript > full transcript > touched-files > `restore-summary.json`
54
55	## Memory surfaces consumed at startup
56
57	The primary surfaces a startup ingestion path reads:
58
59	- AGENTS / role / CULTURE / startup overlays (the per-seat identity layer)
60	- startup replay context (what the agent did before the current boot)
61	- restore / reentry packets and Agent Starters (cross-runtime startup path)
62
63	Authority-rank + read-vs-write posture governs which surfaces a
64	startup-ingestion path can write vs only read; treat overlays as
65	write-restricted unless your role explicitly authors them.
66
67	## Startup files vs skills (the distinction)
68
69	Per `agent-startup-guide.md` (product reference doc) and the team handbook:
70
71	\| Startup files \| Skills \|
72	\|---\|---\|
73	\| Rig-specific, role-specific identity \| Reusable SOPs / methodology / knowledge \|
74	\| Tell agent WHO it is, WHAT it's working on, HOW this team operates \| Tell agent HOW to do something (transferable across rigs) \|
75	\| Examples: `role.md`, `CULTURE.md`, `startup/context.md` \| Examples: `openrig-user`, `test-driven-development`, `dogfood` \|
76	\| Authored per-rig \| Authored once, used everywhere \|
77
78	Don't put skill content in startup files. Don't put identity content
79	in skills. The 7-layer additive s

People who liked this also viewed

6 picks

Type